Camsys Subsystem

System Overview

The S100 camsys subsystem includes the Camera sensor (including SerDes), VIN (including MIPI, CIM), ISP, PYM, GDC, YNR, and STITCH modules.

Abbreviation	Full Name	Description
MIPI	Mobile Industry Processor Interface	Standard defined by the MIPI Alliance for mobile industry processor interfaces
CSI	Camera Serial Interface	Camera serial interface
IPI	Image Pixel Interface	Image transmission interface between MIPI and CIM
FOV	Field of View	Field of view
SER	Serializer	Serializer
SerDes	Serializer and Deserializer	Serializer and deserializer
DES	Deserializer	Deserializer
CIM	Camera Interface Manager	Camera access management module supporting online or offline operation
VIN	Video In (CIM+MIPI+LPWM+VCON)	Video input module
ISP	Image Signal Processor	Image signal processor
PYM	Pyramid	Pyramid processing module: image downscaling and ROI
GDC	Geometric Distortion Correction	Geometric distortion correction module
VPF	Video Process Framework (VIN+ISP+PYM..)	Video processing management module
VIO	Video In/Out (VIN+VPM)	Video input/output module
STITCH	Stitch hardware Module	Image stitching processing module
CAMSYS	Camera System (Camera+VPF)	Camera image system

Camsys Hardware Block Diagram

Submodules

CIM

CIM (Camera Interface Manager) is a dedicated hardware block for receiving MIPI-RX IPI image data. CIM handles simultaneous input of multiple image streams and adjusts the timing of the MIPI IPI interface to match the timing requirements of downstream hardware or DDR, delivering images directly via hardware or through DDR to the ISP and PYM.

• The S100 has three CIM modules: CIM0, CIM1, and CIM4.
• A single CIM supports a maximum input of 4V * 8M * 30fps and supports RAW8, RAW10, RAW12, RAW14, RAW16, RAW20, and YUV422-8Bit image formats.
• CIM0 and CIM1 support direct hardware (OTF) output to ISP and PYM and also support offline output to DDR; CIM4 supports only offline output.
• The maximum input width for IPI0 of CIM0 is 5696; all other IPIs in CIM0 and all IPIs in other CIMs support a maximum input width of 4096.

ISP

ISP (Image Signal Processor) is a dedicated engine for image signal processing.
ISP functions include various algorithmic processing of raw images, image characteristic statistics, color space conversion, and time-division multiplexing control of multiple channels, ultimately producing clearer, more accurate, and higher-quality images.

• The S100 has two ISP modules: ISP0 and ISP1.
• Each ISP hardware IP supports up to 12 sensor inputs.
• Maximum ISP processing resolution: 4096 × 2160.
• ISP processing pipeline is shown below:
  
• MCFE:
  Multi-Context Front End, used for multi-channel scheduling control and buffer management in ISP, processing multi-camera images sequentially (one by one).
• RAW Domain:
  RAW-domain image processing includes input port (with input crop functionality), channel switch, input formatter, sensor offset linear, digital gain, gamma FE (i.e., decompander), gamma_sqrt, raw frontend, static defect correction, sinter, chromatic aberration correction, gamma_sq, gamma BE, static white balance, radial shading correction, mesh shading correction, digital gain iridix, iridix, demosaic, etc.
• RGB Domain:
  RGB-domain image processing includes purple fringe correction, color matrix, gamma RGB forward SQ, crop, CNR, gamma RGB reverse SQ, RGB gamma, etc.
• Output formatter:
  Performs color space (CS) conversion, transforming RGB channel data into formats such as YUV, and handles output control.

YNR

YNR is a Digital Noise Reduction module operating in the YUV domain, supporting both 2DNR and 3DNR modes.

• The S100 has one YNR module, YNR1, which only supports the ISP1-online-YNR1-online-PYM1 scenario.
• In 3DNR mode, the maximum supported resolution is 2048×2048; in 2DNR mode, it supports up to 3840×2160.

PYM

PYM (Pyramid) is a hardware acceleration module that processes input images in pyramid layers and outputs them to DDR.

• The S100 has three PYM modules: PYM0, PYM1, and PYM4.
• SRC layer: Represents the source image layer.
• BL layer: Represents bilinear downsampled layers; BL Base 0~4 correspond to 1/2, 1/4, 1/8, 1/16, and 1/32 of the source image, respectively.
• DS layer: Output layer; each layer can arbitrarily select an input layer (SRC or BL0~4), perform downsampling and ROI processing, and then output to DDR.
• Maximum input width and height: 4096; minimum input width and height: 32.
• Downscaling ratio: (1/2, 1]; upscaling is not supported.
• Performance: PYM0/1 support 4K@120fps; PYM4 supports 4K@90fps but does not support online input.

GDC

GDC is a hardware module capable of performing perspective transformation, distortion correction, and rotation at specific angles (0°, 90°, 180°, 270°) on input images.

Typical supported input resolutions include: 3840×2160, 2688×1944, 1920×1080, 1280×720, 640×480, and 480×320.

Hardware specifications:

• Maximum resolution: 3840×2160
• Minimum resolution: 96×96 (odd-numbered rows or columns are not supported)
• Performance: 3840×2160 @ 60fps
• Operating mode: DDR → GDC → DDR
• Input format: YUV420 semi-planar
• Output format: YUV420 semi-planar

Introduction to GDCTool

GDC Tool is a PC-based utility that enables offline simulation of GDC processing effects. Users can prepare JPEG-format images, load them into GDC Tool for offline correction, and then either directly save a config.bin file for hardware correction or save a layout.json file to generate a config.bin for hardware correction.

Launching GDC Tool

1. Windows Environment

   • Installation prerequisites: Requires Node.js. See: https://nodejs.cn/download/
   • Install dependencies: Open a Windows command prompt, navigate to the GDC tool directory (e.g., gdc-tool-gui-xxxx-windows), and run npm install express.
   • Launch the application: In the command prompt, navigate to the tool directory and run node.exe app.js. Then open Chrome and go to http://localhost:3000/.

2. Unix Environment

   • Installation prerequisites (MacOS): Run brew install node.
   • Install dependencies: In the tool directory, run npm install --production.
   • Launch the application: Run node app.js and open http://localhost:3000/ in your browser.

Transformation Modes in GDC Tool

Six transformation modes are available: Affine, Equisolid, Equisolid (cylinder), Equidistant, Custom, and Keystone+dewarping. These correspond to transformation modes described in the transformation_t section of the GDC Bin API documentation. The following table describes the purpose of each transformation mode:

Transformation Mode	Purpose
Affine	A linear transformation providing simple image rotation without distortion correction
Equisolid	Panoramic transformation with the largest transformation grid
Equisolid (cylinder)	Cylindrical transformation
Equidistant	Equidistant transformation, where distances after transformation remain equidistant
Custom	User-defined custom transformation
Keystone+dewarping	Compared to Equidistant, dewarp_keystone adds two parameters: trapezoid_left_angle and trapezoid_right_angle. By default, both are 90°, yielding the same result as Equidistant.

All transformation types share three common parameters: Pan, Tilt, and Zoom. (Example: Equidistant transformation with input/output resolution of 1280×720.) In the following output images, the blue rectangle indicates the effect when only the specified parameter is set to the given value, while all other parameters remain at their defaults.

• Pan

  Horizontally offsets the transformation grid by a given number of pixels within the range (-1280, +1280), as shown below:
  

• Tile

  Vertically offsets the transformation grid by a given number of pixels within the range (-720, +720), as shown below:
  

• Zoom

  Scales the transformation output by a given factor within the range (0, +∞), where (0, 1) denotes values greater than 0 and less than 1, as shown below:
  

1. Affine

   • Function Description

     Provides a linear transformation.

     

   • Member Description

     Member	Description
     int32_t pan	Default: 0; no modification
     int32_t tilt	Default: 0; no modification
     zoom	Scales the transformation output by the provided factor. When rotation angle is 180° or 270°, this value must be ≥1.03
     double angle (rotation)	Image rotation angle: 0°/90°/180°/270°

     Note!

     Input and output widths must be aligned to 16-byte boundaries.

     When the rotation angle is 180° or 270°, the zoom parameter must be ≥1.03.

2. Equisolid

   • Function Description

     This transformation provides equisolid (panoramic) correction and displays the result as a projection onto a plane.

     

   • Member Description

     Member	Description
     int32_t pan	Default: 0; no modification
     int32_t tilt	Default: 0; no modification
     zoom	Scales the transformation output by the provided factor
     double strengthX	Transformation strength along the X-axis (non-negative parameter)
     double strengthY	Transformation strength along the Y-axis (non-negative parameter)
     double angle (rotation)	Image rotation angle: 0°/90°/180°/270°

     strength x adjustment effect: transformation strength along the X-axis, with values in the range (0, +∞), as shown below:

     strength y debugging effect: the transformation intensity along the Y-axis, with a value range of (0, +∞). As shown below:

     Rotation debugging effect: value range (-180, 180). As shown below:

     Note!

     The width of input and output dimensions must be aligned to 16-byte boundaries.

3. Equisold (cylinder)

   • Function Description

     This transformation provides equirectangular (panoramic) correction and displays the result as a projection onto a plane.

     

   • Member Description

     Member	Description
     int32_t pan	default 0, no modification
     int32_t tilt	default 0, no modification
     zoom	Scales the transformation output by the provided factor
     strength	Intensity of the transformation
     double angle(rotation)	Image rotation angle: 0/90/180/270

     strength debugging effect: transformation intensity (0, +∞). As shown below:

     

     rotation debugging effect: value range (-180, +180). As shown below:

     

     Note!

     The width of input and output dimensions must be aligned to 16-byte boundaries.

4. Equidistant

   • Function Description

     The equidistant transformation includes many parameters that allow it to provide a variety of target planes for projection. This gives users greater freedom to select the desired region of the fisheye frame to be transformed.

     

   • Member Description

     Member	Description
     int32_t pan	default 0, no modification
     int32_t tilt	default 0, no modification
     zoom	Scales the transformation output by the provided factor
     double angle(rotation)	Image rotation angle: 0/90/180/270
     double elevation	Defines the elevation angle of the projection axis, ranging from 0 to 90
     double azimuth	Defines the azimuth angle of the projection axis. If the elevation parameter is 0, azimuth will have no visible effect
     int32_t keep_ratio	When the "keep ratio" parameter is enabled, the FOV height parameter will be ignored, and its value will be automatically calculated to maintain equal stretching intensity in both horizontal and vertical directions
     double FOV_h	Describes the size (in degrees) of the output field of view in the horizontal dimension. Valid values range from 0 to 180
     double FOV_w	Describes the size (in degrees) of the output field of view in the vertical dimension. Valid values range from 0 to 180
     double cylindricity_y	Describes the sphericity of the target projection along the Y-axis. This value ranges from 0 to 1, where 1 represents a spherical shape. If this value is set to 1 while "cylindricity_x" is set to 0, the projection will form a cylinder along the Y-axis
     double cylindricity_x	Describes the sphericity of the target projection along the X-axis. This value ranges from 0 to 1, where 1 represents a spherical shape. If this value is set to 1 while "cylindricity_y" is set to 0, the projection will form a cylinder along the X-axis

     elevation debugging effect:

     

     azimuth debugging effect:

     

     rotation debugging effect:

     

     cylindricity x debugging effect:

     Describes the sphericity of the target projection along the X-axis. This value ranges from 0 to 1, where 1 represents a spherical shape. If this value is set to 1 and cylindricity_y is set to 0, the projection will form a cylinder along the X-axis. As shown below:

     

     cylindricity y debugging effect:

     Describes the sphericity of the target projection along the Y-axis. This value ranges from 0 to 1, where 1 represents a spherical shape. If this value is set to 1 and cylindricity_x is set to 0, the projection will form a cylinder along the Y-axis. As shown below:

     

     Note!

     The width of input and output dimensions must be aligned to 16-byte boundaries. Normal human vision is approximately 90 degrees. For transformations where cylindricity (see below) equals "0", setting both FOV width and height to 180 will cause infinite image stretching. If both cylindricity_x and cylindricity_y are set to 1, the projection will be spherical. If both are set to 0, the transformation will be rectangular.

5. Custom

   • Function Description

     After applying the custom transformation, each polygon in the input image is transformed into a square. In other words, any four adjacent input points of any shape become a square after transformation, as shown in the figure below. However, the shape and position of the polygons will change after transformation.

     

     These are used to create transformations that cannot be described by any of the provided standard types. To correct arbitrary distortion, a special calibration file named config0.txt must be provided to the GDC tool, as shown in the figure below:

     

   • Member Description

     Member	Description
     int32_t pan	default 0, no modification
     int32_t tilt	default 0, no modification
     zoom	Scales the transformation output by the provided factor
     char custom_file[128]	Name of the config.txt file
     custom_tranformation_t custom	Parsed custom transformation structure

     Rules for the Config file should generally observe the following points:

     1. The first line enables full tile in pixel calculation: 1 means enable, 0 means disable.

     2. The second line specifies the number of pixels to skip if full tile is enabled; these values must be greater than 0. Smaller numbers result in slower libgdc performance (slower performance means a larger config.bin file and longer time for libgdc to generate config.bin).

     3. The third line specifies the number of calibration points in vertical and horizontal directions. The first value Y = 1081 indicates 1081 calibration points vertically, and the second value X = 1921 indicates 1921 calibration points horizontally.

     4. The fourth line specifies the center point of the selected region, typically (Y-1)/2, (X-1)/2.

     5. Calibration points must be non-negative integers or floats, and calibration points in adjacent rows must not be duplicated. e.g., The figure below shows a partial excerpt of the data. Rows 5 to 9 contain the coordinate values of calibration points in the source image, formatted as Y: X. In this example, there are a total of 1081x1921 calibration points.

        

     6. Since calibration points must be equally spaced, the output image resolution depends on the number of calibration points.

        

        e.g., Output image Width = 100, Height calculated as 340, calculated as follows: 100/height = (96-1)/(324-1)
        The figure below shows a simpler example of 3x3 coordinate point transformation:

        

6. Keystone + Dewarping

   • Function Description

     

   • Member Description

     Member	Description
     int32_t pan	default 0, no modification
     int32_t tilt	default 0, no modification
     zoom	Scales the transformation output by the provided factor
     double angle(rotation)	Image rotation angle: 0/90/180/270
     double elevation	Defines the elevation angle of the projection axis, ranging from 0 to 90
     double azimuth	Defines the azimuth angle of the projection axis. If the elevation parameter is 0, azimuth will have no visible effect
     int32_t keep_ratio	When the "keep ratio" parameter is enabled, the FOV height parameter will be ignored, and its value will be automatically calculated to maintain equal stretching intensity in both horizontal and vertical directions
     double FOV_h	Describes the size (in degrees) of the output field of view in the horizontal dimension. Valid values range from 0 to 180
     double FOV_w	Describes the size (in degrees) of the output field of view in the vertical dimension. Valid values range from 0 to 180
     double cylindricity_y	Describes the sphericity of the target projection along the Y-axis. This value ranges from 0 to 1, where 1 represents a spherical shape. If this value is set to 1 while "cylindricity_x" is set to 0, the projection will form a cylinder along the Y-axis
     double cylindricity_x	Describes the sphericity of the target projection along the X-axis. This value ranges from 0 to 1, where 1 represents a spherical shape. If this value is set to 1 while "cylindricity_y" is set to 0, the projection will form a cylinder along the X-axis
     double trapezoid_left_angle	Default 90; range 0.1 to 90; in the transformation grid, the angle between the left boundary and the bottom boundary—see actual effect
     double trapezoid_right_angle	Default 90; range 0.1 to 90; in the transformation grid, the angle between the right boundary and the bottom boundary—see actual effect

     Note!

     The width of input and output dimensions must be aligned to 16-byte boundaries.

GDC Tool Transformation Mode Parameter Description

The configuration file can be generated by the GDC tool and saved as layout.json. Different transformation modes have different parameters. Taking custom mode and keystone+dewarping mode as examples, the configuration parameters are explained below.

1. keystone+dewarping mode

   {
       "inputRes": [
           1920, // Width of the input image resolution
           1080  // Height of the input image resolution
       ],
       "param": {
           "fov": 180,        // Field of view of the input image
           "diameter": 1080,  // Diameter of the input image; controls the overall size of the transformation grid
           "offsetX": 0,      // Horizontal offset of the transformation grid
           "offsetY": 0       // Vertical offset of the transformation grid
       },
       "outputRes": [
           1920, // Width of the output image resolution
           1080  // Height of the output image resolution
       ],
       "transformations": [
           {
               "transformation": "Dewarp_keystone", // Transformation mode"position": [ // ROI region settings for the output image
                   0, // Horizontal offset of the output image's ROI
                   0, // Vertical offset of the output image's ROI
                   1920, // Width of the output image's ROI
                   1080 // Height of the output image's ROI
               ],
               "param": {
                   "left_base_angle": 90, // Default: 90; range: 0.1 to 90; in the transformation mesh, the angle of the left boundary relative to the bottom boundary
                   "right_base_angle": 90, // Default: 90; range: 0.1 to 90; in the transformation mesh, the angle of the right boundary relative to the bottom boundary
                   "azimuth": 90, // Defines the azimuth angle of the projection axis. If the elevation parameter is 0, the azimuth will have no visible effect.
                   "elevation": 0, // Defines the elevation angle of the projection axis, ranging from 0 to 90.
                   "rotation": 0, // Rotation angle to be applied to the output image
                   "fovWidth": 90, // Specifies the horizontal field of view of the output image in degrees. Larger values result in a wider horizontal transformation mesh. Valid range: 0 to 180.
                   "fovHeight": 90, // Specifies the vertical field of view of the output image in degrees. Larger values result in a taller vertical transformation mesh. Valid range: 0 to 180.
                   "keepRatio": 0, // When "keepRatio" is set to 1, the fovHeight parameter is ignored and automatically calculated to maintain uniform stretching intensity in both horizontal and vertical directions.
                   "cylindricityX": 1, // Describes the sphericity of the target projection along the X-axis. Values range from 0 to 1, where 1 represents a fully spherical projection. If set to 1 while "cylindricityY" is 0, the projection forms a cylinder along the X-axis.
                   "cylindricityY": 1 // Describes the sphericity of the target projection along the Y-axis. Values range from 0 to 1, where 1 represents a fully spherical projection. If set to 1 while "cylindricityX" is 0, the projection forms a cylinder along the Y-axis.
               },
               "ptz": [
                   0, // pan parameter
                   0, // tilt parameter
                   1 // zoom parameter
               ],
               "roi": { // Input image ROI region settings
                   "x": 0, // Horizontal offset of the input image's ROI
                   "y": 0, // Vertical offset of the input image's ROI
                   "w": 1920, // Width of the input image's ROI
                   "h": 1080 // Height of the input image's ROI
               }
           }
       ],
       "mode": "semiplanar420", // Processing format setting
       "eccMode": "eccDisabled", // ECC mode for processing
       "colourspace": "yuv" // Data format for processing
   }

2. Custom mode

   {
       "inputRes": [
           1280, // Width of the input image resolution
           720 // Height of the input image resolution
       ],
       "param": {
           "fov": 192, // Field of view of the input image
           "diameter": 720, // Diameter of the input image, controlling the overall size of the transformation mesh
           "offsetX": 0, // Horizontal offset of the transformation mesh
           "offsetY": 0 // Vertical offset of the transformation mesh
       },
       "outputRes": [
           560, // Width of the output image resolution
           258 // Height of the output image resolution
       ],
       "transformations": [
           {
               "transformation": "Custom", // Transformation mode
               "position": [ // ROI region settings for the output image
                   0, // Horizontal offset of the output image's ROI
                   0, // Vertical offset of the output image's ROI
                   560, // Width of the output image's ROI (must be ≤ outputRes width)
                   258 // Height of the output image's ROI (must be ≤ outputRes height)
               ],
               "ptz": [
                   0, // pan parameter
                   0, // tilt parameter
                   1 // zoom parameter
               ],
               "roi": { // Invalid in custom mode
                   "x": 0, // Invalid in custom mode
                   "y": 0, // Invalid in custom mode
                   "w": 0, // Invalid in custom mode
                   "h": 0 // Invalid in custom mode
               },
   
   "param": {
                   "customTransformation": "/path_to/camera_0_gdc.txt" // Path to the coordinate mapping file on the device
               }
           }
       ],
       "mode": "semiplanar420", // Processing format setting
       "eccMode": "eccDisabled", // ECC mode for processing
       "colourspace": "yuv" // Data format for processing
   }

   Note!

   1. Always set eccMode to "eccDisabled". Although other ECC modes are selectable, they have no actual effect.
   2. When parameters are fractional, ensure precision of at least 8 decimal places after floating-point computation; otherwise, the generated binary may differ.
   3. When populating the data structure or JSON, users must include all fields shown in the mode examples.
   4. In non-custom modes, the "roi" parameter in the configuration file specifies the ROI of the input image.
   5. The "position" parameter in the configuration file specifies the ROI of the output image.

3. Affine
   Configuration file content as follows:

   {
       "inputRes": [
           1920,
           1080
       ],
       "param": {
           "fov": 160,
           "diameter": 1080,
           "offsetX": 0,
           "offsetY": 0
       },
       "outputRes": [
           1920,
           1080
       ],
       "transformations": [
           {
               "transformation": "Affine",
               "position": [
                   0,
                   0,
                   1920,
                   1080
               ],
               "param": {
                   "rotation": 0
               },
               "ptz": [
                   0,
                   0,
                   1
               ],
               "roi": {
                   "x": 0,
                   "y": 0,
                   "w": 1920,
                   "h": 1080
               }
           }
       ],
       "mode": "semiplanar420",
       "eccMode": "eccDisabled",
       "colourspace": "yuv"
   }

   Input image with transformation mesh shown below:

   

   Output image shown below:

   

4. Equisolid Configuration file content as follows:

   {
       "inputRes": [
           1920,
           1080
       ],
       "param": {
           "fov": 160,
           "diameter": 1080,
           "offsetX": 0,
           "offsetY": 0
       },
       "outputRes": [
           1920,
           1080
       ],
       "transformations": [
           {
               "transformation": "Panoramic",
               "position": [
                   0,
                   0,
                   1920,
                   1080
               ],
               "param": {
                   "strength": 1,
                   "strengthY": 1,
                   "rotation": 0
               },
               "ptz": [
                   0,
                   0,
                   1
               ],
               "roi": {
                   "x": 0,
                   "y": 0,
                   "w": 1920,
                   "h": 1080
               }
           }
       ],
       "mode": "semiplanar420","eccMode": "eccDisabled",
       "colourspace": "yuv"
   }

   Input image with transformation grid as follows:

   

   Output image as follows:

   

5. Equisolid (cylinder) Configuration file content as follows:

   {
       "inputRes": [
           1920,
           1080
       ],
       "param": {
           "fov": 160,
           "diameter": 1080,
           "offsetX": 0,
           "offsetY": 0
       },
       "outputRes": [
           1920,
           1080
       ],
       "transformations": [
           {
               "transformation": "Stereographic",
               "position": [
                   0,
                   0,
                   1920,
                   1080
               ],
               "param": {
                   "strength": 1,
                   "rotation": 0
               },
               "ptz": [
                   0,
                   0,
                   1
               ],
               "roi": {
                   "x": 0,
                   "y": 0,
                   "w": 1920,
                   "h": 1080
               }
           }
       ],
       "mode": "semiplanar420",
       "eccMode": "eccDisabled",
       "colourspace": "yuv"
   }

   Input image with transformation grid as follows:

   

   Output image as follows:

   

6. Equidistant Configuration file content as follows:

   {
       "inputRes": [
           1920,
           1080
       ],
       "param": {
           "fov": 160,
           "diameter": 1080,
           "offsetX": 0,
           "offsetY": 0
       },
       "outputRes": [
           1920,
           1080
       ],
       "transformations": [
           {
               "transformation": "Universal",
               "position": [
                   0,
                   0,
                   1920,
                   1080
               ],
               "param": {
                   "azimuth": 0,
                   "elevation": 0,
                   "rotation": 0,
                   "fovWidth": 90,
                   "fovHeight": 90,
                   "keepRatio": 0,
                   "cylindricityX": 1,
                   "cylindricityY": 1
               },
               "ptz": [
                   0,
                   0,
                   1
               ],
               "roi": {
                   "x": 0,
                   "y": 0,
                   "w": 1920,
                   "h": 1080
               }
           }
       ],
       "mode": "semiplanar420",
       "eccMode": "eccDisabled",
       "colourspace": "yuv"
   }

   Input image with transformation grid as follows:

   

   Output image as follows:

   

7. Custom Input resolution: 1280x720, output resolution: 560x258. Configuration file content as follows:

   {
       "inputRes": [
           1280,
           720
       ],
       "param": {
           "fov": 192,
           "diameter": 720,
           "offsetX": 0,
           "offsetY": 0
       },
       "outputRes": [
           560,
           258
       ],
       "transformations": [
           {
               "transformation": "Custom",
               "position": [
                   0,
                   0,
                   560,
                   258
               ],
               "ptz": [
                   0,
                   0,
                   1
               ],
               "roi": {
                   "x": 0,
                   "y": 0,
                   "w": 0,
                   "h": 0
               },
               "param": {
                   "customTransformation": "/path_to/camera_0_gdc_config_3.1.txt"
               }
           }
       ],
       "mode": "semiplanar420",
       "eccMode": "eccDisabled",
       "colourspace": "yuv"
   }

   Input image with transformation grid as follows:

   

   Output image as follows:

   

8. Keystone + dewarping Configuration file content as follows:

   {"inputRes": [
           1920,
           1080
       ],
       "param": {
           "fov": 180,
           "diameter": 1080,
           "offsetX": 0,
           "offsetY": 0
       },
       "outputRes": [
           1920,
           1080
       ],
       "transformations": [
           {
               "transformation": "Dewarp_keystone",
               "position": [
                   0,
                   0,
                   1920,
                   1080
               ],
               "param": {
                   "left_base_angle": 90,
                   "right_base_angle": 90,
                   "azimuth": 0,
                   "elevation": 0,
                   "rotation": 0,
                   "fovWidth": 90,
                   "fovHeight": 90,
                   "keepRatio": 0,
                   "cylindricityX": 1,
                   "cylindricityY": 1
               },
               "ptz": [
                   0,
                   0,
                   1
               ],
               "roi": {
                   "x": 0,
                   "y": 0,
                   "w": 1920,
                   "h": 1080
               }
           }
       ],
       "mode": "semiplanar420",
       "eccMode": "eccDisabled",
       "colourspace": "yuv"
   }

   Input image with transformation grid is shown below:

   

   Output image is shown below:

   

GDC Bin Related API Reference

The following APIs are used for GDC BIN generation. For GDC module control APIs, refer to the HBN API.

1. hb_vio_gen_gdc_cfg

   【Function Declaration】

   int32_t hb_vio_gen_gdc_cfg(param_t *gdc_parm, window_t *wnds, uint32_t wnd_num, void **cfg_buf, uint64_t *cfg_size)

   【Parameter Description】

   • [IN] param_t *gdc_parm: GDC-related parameters, including resolution, format, etc.
   • [IN] window_t *wnds: Parameters for internal regions within GDC.
   • [IN] uint32_t wnd_num: Number of windows.
   • [OUT] uint32_t **cfg_buf: Generated GDC configuration BIN buffer, allocated internally.
   • [OUT] uint64_t *cfg_size: Size of the GDC configuration BIN file.

   【Return Value】

   • Success: E_OK: Success
   • Failure: E_NOT_OK: Fail, return error code; range: [-10000, -1]

   【Function Description】 Generates the BIN file required for GDC module operation.

2. hb_vio_set_gdc_cfg

   【Function Declaration】

   int32_t hb_vio_set_gdc_cfg(uint32_t pipeline_id, uint32_t *cfg_buf, uint64_t cfg_size)

   【Parameter Description】

   • [IN] uint32_t pipeline_id: Pipeline ID; software channel ID; range: [0, 23], default: 0;
   • [IN] cfg_buf: Configuration buffer of the GDC BIN file.
   • [IN] cfg_size: Size of the GDC BIN file.

   【Return Value】

   • Success: E_OK: Success
   • Failure: E_NOT_OK: Fail, return error code; range: [-10000, -1]

   【Function Description】

   Sets the GDC module's configuration BIN.

3. hb_vio_free_gdc_cfg

   【Function Declaration】

   void hb_vio_free_gdc_cfg(uint32_t *cfg_buf)

   【Parameter Description】

   • [IN] uint32_t* cfg_buf: Buffer of the GDC BIN file.

   【Return Value】

   • NONE

   【Function Description】

   Frees the buffer allocated for the GDC module's configuration BIN.

GDC Bin Related Parameter Descriptions

1. typedef struct param_t

   Name	Type	Min Value	Max Value	Default	Description	Required
   format	frame_format_t				Image format to be processed	Yes
   in	resolution_t				Actual input image resolution	Yes
   out	resolution_t				Actual output image resolution	Yes
   x_offset	int32_t	0		0	Pixel offset of the input region along the x-axis	Yes
   y_offset	int32_t	0		0	Pixel offset of the input region along the y-axis	Yes
   diameter	int32_t				Pixel diameter of the circular input area containing the actual fisheye image within the rectangular input image. For some cameras, this circular image area may be larger or smaller than the rectangular canvas (sometimes cropped). Typically, diameter should match input.height.	Yes
   fov	double	0			Field of view defining the visible angle of the input image, affecting the curvature of the source mesh. Larger FOV results in greater perspective distortion.	Yes

2. typedef enum frame_format frame_format_t

   Name	Type	Min Value	Max Value	Default	Description	Required
   FMT_UNKNOWN	enum				Unknown format
   FMT_LUMINANCE	enum				Not supported
   FMT_PLANAR_444	enum				Not supported
   FMT_PLANAR_420	enum				Not supported
   FMT_SEMIPLANAR_420	enum				NV12
   FMT_GDC_MAX	enum

3. typedef struct resolution_s resolution_t

   Name	Type	Min Value	Max Value	Default	Description	Required
   w	uint32_t				Width (in pixels)
   h	uint32_t				Height (in pixels)

4. typedef struct window_t

   Name	Type	Min Value	Max Value	Default	Description	Required
   out_r	rect_t				Output data size information
   transform	transformation_t	0	6	0	Transformation mode used
   input_roi_r	rect_t				ROI region
   pan	int32_t				Horizontal target displacement (in pixels) centered on the output image
   tilt	int32_t				Vertical target displacement (in pixels) centered on the output image
   zoom	double				Target zoom factor
   strengthX	double				Non-negative transformation strength parameter in the X direction
   strengthY	double				Non-negative transformation strength parameter in the Y direction
   angle	double				Rotation angle of the principal projection axis around itself
   elevation	double				Angle specifying the principal projection axis
   azimuth	double				Angle specifying the principal projection axis, measured clockwise from north
   keep_ratio	int32_t				Maintain the same stretch intensity in both horizontal and vertical directions
   FOV_h	double				Vertical dimension of the output field of view, expressed in degrees
   FOV_w	double				Horizontal dimension of the output field of view, expressed in degrees
   cylindricity_y	double				Cylindricity level of the target projection shape in the vertical direction
   cylindricity_x	double				Cylindricity level of the target projection shape in the horizontal direction
   custom_file[128]	char				Custom transformation description file used in custom mode
   custom	custom_tranformation_t				Transformation information in custom mode
   trapezoid_left_angle	double				Left acute angle between the trapezoid base and its slanted side
   trapezoid_right_angle	double				Right acute angle between the trapezoid base and its slanted side
   check_compute	uint8_t				Currently unused

5. typedef struct rect_s rect_t

   Name	Type	Min Value	Max Value	Default	Description	Required
   x	int32_t				Starting X coordinate
   y	int32_t				Starting Y coordinate
   w	int32_t				Width
   h	int32_t				Height

6. typedef enum gdc_transformation transformation_t

   Name	Type	Min Value	Max Value	Default	Description	Required
   PANORAMIC	enum				Panoramic transformation
   CYLINDRICAL	enum				NA
   STEREOGRAPHIC	enum				Same as distortion correction and panoramic transformation, but the output image is a cylindrical panorama instead of a planar image
   UNIVERSAL	enum				Equidistant transformation
   CUSTOM	enum				User-defined transformation; allows customization of the transformation mesh
   AFFINE	enum				Linear transformation
   DEWARP_KEYSTONE	enum				Non-equidistant transformation selectable relative to equidistant transformation; equidistant transformation is a special case of this

7. typedef struct point_s point_t

   Name	Type	Min	Max	Default	Description	Required
   x	double				x coordinate
   y	double				y coordinate

8. typedef struct custom_tranformation_s custom_tranformation_t

   Name	Type	Min	Max	Default	Description	Required
   full_tile_calc	uint8_t				Whether to enable tile-based calculation; if enabled, libgdcbin performs additional min/max calculations per tile. More tiles yield higher precision and better results but increase bin generation time.
   tile_incr_x	uint16_t				Tile increment in x direction
   tile_incr_y	uint16_t				Tile increment in y direction
   w	int32_t				Number of points in the horizontal direction of the custom transformation grid
   h	int32_t				Number of points in the vertical direction of the custom transformation grid
   centerx	double				Center along the x-axis, typically half the number of horizontal coordinate points
   centery	double				Center along the y-axis, typically half the number of vertical coordinate points
   *points	point_t				Sequence of transformation points defined in config.txt; total count = w * h

STITCH

Introduction

Stitch is a configurable image stitching computation unit capable of blending and stitching multiple images together. It is primarily used for 360-degree surround-view image stitching in automated parking scenarios. Stitch operates based on Regions of Interest (ROIs). Each ROI can perform alpha-beta blending between two source images and write the result into a designated ROI of the target image. This blending approach ensures smoother transitions at stitching boundaries. Additionally, Stitch supports gain adjustment for Y, U, and V channels separately, enabling brightness and chrominance balancing between source images to further enhance stitching quality. Moreover, Stitch allows users to input custom per-pixel alpha-beta weight values, enabling various blending effects such as background blur or image watermarking.
The Stitch hardware supports maximum input and output resolutions of 4096x4096.

Hardware Operating Modes

• Online Blending: No LUT table input required; hardware automatically performs blending and stitching. Requires ROI width = height (w = h). In this mode, the hardware automatically calculates alpha and beta weights for each pixel based on configured parameters such as transition zone width and direction.
• Alpha Blending: Requires an alpha LUT table. The hardware reads alpha weight values from DDR memory for weighted blending. The alpha LUT stores the alpha weight for each pixel within the ROI. For each pixel, the hardware reads Y, UV, and alpha values separately for weighted blending.
• Alpha-Beta Blending: Requires both alpha and beta LUT tables. The hardware reads alpha and beta weight values from DDR for weighted blending.
• Src Copy: No LUT table required; hardware directly copies src0.
• Src Alpha Copy: Requires an alpha LUT table; hardware reads alpha weights from DDR and blends src0 accordingly.

Here, the "LUT table" refers to the buffer storing blending weight parameters.

Hardware Stitching Diagram

By using the two source ROIs shown in the image with different blend modes, the corresponding ROI output results are generated.

Stitching Scheme Overview

The hardware stitching function can merge and blend multiple images into a single output image. Designed flexibly, it uses ROIs as the basic processing unit and employs the alpha blending algorithm. Different ROI partitions and configurations can be defined via configuration parameters to generate various stitching schemes. Additionally, LUT tables are used to optimize transition zones during stitching. In autonomous driving and ADAS Automated Parking Assist (APA) scenarios, this hardware can stitch four IPM (Inverse Perspective Mapping) images—already distortion-corrected from four cameras—into a single 360-degree surround-view image for parking space detection, allowing users to easily view parking lines and surroundings.

Typical Scenario

In an APA scenario with four surround-view cameras, GDC fetches four back-projected images and reference points (CFG BIN) from DDR, outputs four IPM images after distortion correction, and then uses the STITCH hardware module with a pre-defined stitching configuration (CPG PARAM) to generate a bird's-eye-view output.

Placement Layout

1. The four IPM images are placed at specified locations in the output buffer using copy mode.
2. Non-overlapping regions can use direct copy mode.
3. Overlapping ROI regions use Alpha Blend mode for seamless fusion.

LUT Table

The LUT table stores alpha/beta blending coefficients (similar to weight values). Each ROI must generate corresponding per-pixel blending coefficients ranging from 0 to 255, which are sequentially stored in the LUT table memory. When an ROI uses alpha or beta blending mode, these parameters are used for fusion.

For example, in the LUT generation described in the "Coordinate Parameter Example" section:
ROI-0/1: 256×512, ROI-2/3: 560×256, ROI-4/5: 256×218, ROI-6/7: 256×186
LUT: ROI-0 + ROI-1 + ROI-2 + ROI-3 + ROI-4 + ROI-5 + ROI-6 + ROI-7
Currently, the LUT table can be generated using the convert_tool.

Coordinate Parameter Example

ROI partitioning for hardware stitching is directly related to camera mounting positions. Currently, ROI partitions can be generated using the convert-tool. The figure below shows an example of coordinate points for each ROI region.

ROI	Range	SRC0	Start	Size	SRC1	Start	Size	Dest Start	Mode
0	Full left view	Left (frame0)	(0,0)	-256,512	/	/	/	(0,40)	Direct Copy
1	Full right view	Right (frame2)	(0,0)	-256,512	/	/	/	(304,40)	Direct Copy
2	Full rear view	Rear (frame3)	(0,0)	-560,256	/	/	/	(0,366)	Direct Copy
3	Full front view	Front (frame1)	(0,0)	-560,256	/	/	/	(0,0)	Direct Copy
4	Overlap: Left & Front	Left (frame0)	(0,0)	-256,218	Front (frame1)	(0,40)	-256,218	(0,40)	AlphaBlend
5	Overlap: Right & Front	Right (frame2)	(0,0)	-256,218	Front (frame1)	(304,40)	-256,218	(304,40)	AlphaBlend
6	Overlap: Left & Rear	Left (frame0)	(0,366)	-256,186	Rear (frame3)	(0,0)	-256,186	(0,366)	AlphaBlend
7	Overlap: Right & Rear	Right (frame2)	(0,366)	-256,186	Rear (frame3)	(304,0)	-256,218	(-304,366)	AlphaBlend

Data Flow and Performance Metrics

After RDK-S100 connects to cameras, the data flows through subsequent processing modules as shown below:

• MIPI RX: 3 CDPHY lanes, each supporting either DPHY up to 4.5 Gbps/lane × 4 lanes or CPHY up to 3.5 Gbps/trio × 3 trios. Each lane supports 4 virtual channels (VCs), theoretically allowing up to 12 camera inputs.

RDK-S100 software is expected to support up to 6 cameras: RX4 can connect up to 4 cameras via SerDes, while RX0 and RX1 each connect to 1 camera. For non-standard configurations, please consult an FAE for confirmation.

tip

The commercial version offers more comprehensive feature support, deeper hardware capability exposure, and exclusive customization options. To ensure compliance and secure delivery, access to the commercial version will be granted through the following process:

Commercial Version Access Process:

1. Complete a questionnaire: Submit your organization’s information and intended use case.
2. Sign an NDA: We will contact you based on your submission to finalize and sign a Non-Disclosure Agreement.
3. Content release: After NDA execution, we will provide access to commercial documentation via a private channel.

If you wish to access the commercial version, please complete the questionnaire below. We will contact you within 3–5 business days:

Questionnaire link: https://horizonrobotics.feishu.cn/share/base/form/shrcnJQBMIkRm6K79rjXR0hr0Fg

• CIM: Receives input from RX and can output online to ISP0/ISP1 (RAW) or PYM0/PYM1 (YUV), or store offline to DDR for subsequent modules to access via DDR.

• ISP: Two ISP units, each supporting 4 online + 8 offline inputs. Each ISP can handle up to 2×4K@60fps.

• PYM: Three PYM units—PYM0/PYM1 are full-featured and support both online/offline modes, while PYM4 supports offline only, with 4K@60fps processing capability.

• GDC: One GDC unit, supporting offline mode only, with 4K@60fps processing capability.

	CIM	ISP0 / ISP1	PYM0 / PYM1	PYM4	GDC	YNR	STITCH
Per-frame latency at 1080p	3.7151 ms	1.8616 ms	2.2373 ms	2.7616 ms	3.7447 ms	1.7774 ms	1.5739 ms
Per-frame latency at 4K	14.8606 ms	7.4467 ms	7.1356 ms	10.7018 ms	15.0624 ms	7.1096 ms	5.7349 ms

Camsys Input Capability

The S100 Camsys hardware theoretically supports up to 8×4K RAW @30fps + 4×1536p YUV @30fps.
Validated maximum input configurations include:

1. 3×4K RAW (3840×2160) @30fps + 9×1280p RAW (1920×1280) @30fps;
2. 3×4K RAW (3840×2160) @30fps + 5×1280p RAW (1920×1280) @30fps + 4×1536p YUV (1920×1536) @30fps.

Supported Sensors

Type	Sensor Name	Notes
MIPI sensor	IMX219	raw10, 1080p
GMSL sensor	0820c	yuv, 4K & 1080p
	OVX3C	raw12, 1280P
	OVX8B	raw12, 4K

Camera API

Note: This section describes APIs based on the HBN architecture, not V4L2.

Module Description

The RDK-S100 HBN Camera consists of three components: Camera Sensor, Deserializer, and Serializer. The Serializer encapsulates the MIPI TX functionality.
Each component provides attach and detach interfaces and supports binding/unbinding with VIN. For example, binding a camera to VIN determines which MIPI RX and I2C controller on the SoC the camera uses, followed by sensor initialization.

API Reference

1. hbn_camera_create

【Function Declaration】

int32_t hbn_camera_create(camera_config_t *cam_config, camera_handle_t *cam_fd)

【Parameters】

[IN] camera_config_t *cam_config: Pointer to the configuration structure for the camera to be set up.
[OUT] camera_handle_t *cam_fd: Returns a file descriptor (handle) for operating the camera based on the provided configuration.

【Return Value】

Success: RET_OK (0)
Failure: Negative error code

【Function Description】

Creates a camera handle based on the configuration provided in camera_config_t.

【Notes】

The API will check the sensor lib. If the sensor driver code does not comply with the HBN framework specification, an error will be reported during the check.

The API will check cam_config. If the configuration does not match the IP hardware capabilities, an error will be reported during the check.

2. hbn_camera_destroy

【Function Declaration】

int32_t hbn_camera_destroy(camera_handle_t cam_fd)

【Parameter Description】

[IN] camera_handle_t cam_fd: The camera operation handle created by hbn_camera_create;

【Return Value】

Success: RET_OK 0

Failure: Negative error code for exceptions

【Function Description】

Destroy the corresponding software resources based on the camera handle.

【Notes】

hbn_camera_destroy must be used in pair with hbn_camera_create.

hbn_camera_destroy will release the sensor lib. After execution, the sensor will no longer be accessible.

Internally, hbn_camera_destroy calls hbn_camera_detach_from_vin, which triggers the sensor stream stop operation. Therefore, hbn_camera_destroy must be called before hbn_vflow_destroy.

3. hbn_camera_attach_to_vin

【Function Declaration】

int32_t hbn_camera_attach_to_vin(camera_handle_t cam_fd, vpf_handle_t vin_fd)

【Parameter Description】

[IN] camera_handle_t cam_fd: Camera handle created by hbn_camera_create;

[IN] vpf_handle_t vin_fd: VIN node handle created by the hbn_vnode_open interface.

【Return Value】

Success: RET_OK 0

Failure: Negative error code for exceptions

【Function Description】

Bind the camera and VIN node together within the VPF framework using their respective handles, and perform hardware initialization for the camera.

【Notes】

The same camera must not repeatedly call hbn_camera_attach_to_vin; otherwise, an "attach error" will be reported.

4. hbn_camera_detach_from_vin

【Function Declaration】

int32_t hbn_camera_detach_from_vin(camera_handle_t cam_fd)

【Parameter Description】

[IN] camera_handle_t cam_fd: Camera handle created by hbn_camera_create;

【Return Value】

Success: RET_OK 0

Failure: Negative error code for exceptions

【Function Description】

Unbind the camera from the VIN node and perform de-initialization.

【Notes】

hbn_camera_detach_from_vin must be used in pair with hbn_camera_attach_to_vin.

hbn_camera_destroy internally calls hbn_camera_detach_from_vin, so after calling hbn_camera_destroy, there is no need to explicitly call hbn_camera_detach_from_vin.

5. hbn_camera_attach_to_deserial

【Function Declaration】

int32_t hbn_camera_attach_to_deserial(camera_handle_t cam_fd, deserial_handle_t des_fd, camera_des_link_t link)

【Parameter Description】

[IN] camera_handle_t cam_fd: Camera handle created by hbn_camera_create;

[IN] deserial_handle_t des_fd: Deserializer handle created by hbn_deserial_create;

[IN] camera_des_link_t link: The linking method between camera and deserializer, determined by which link the camera is connected to.

【Return Value】

Success: RET_OK 0

Failure: Negative error code for exceptions

【Function Description】

Bind the camera and deserializer using their handles, and perform hardware initialization for both the deserializer and camera.

【Notes】

This interface should only be called when a deserializer exists in the hardware.

After calling hbn_camera_attach_to_deserial, there is no need to call hbn_camera_attach_to_vin; instead, the deserializer will be bound to the VIN node.

6. hbn_camera_detach_from_deserial

【Function Declaration】

int32_t hbn_camera_detach_from_deserial(camera_handle_t cam_fd)

【Parameter Description】

[IN] camera_handle_t cam_fd: Camera handle created by hbn_camera_create;

【Return Value】

Success: RET_OK 0

Failure: Negative error code for exceptions

【Function Description】

Unbind the camera from the deserializer and perform de-initialization.

【Notes】

hbn_camera_detach_from_deserial must be used in pair with hbn_camera_attach_to_deserial.

Before calling this API, hbn_deserial_detach_from_vin must be called first.

7. hbn_camera_start

【Function Declaration】

int32_t hbn_camera_start(camera_handle_t cam_fd)

【Parameter Description】

[IN] camera_handle_t cam_fd: Camera handle created by hbn_camera_create;

【Return Value】

Success: RET_OK 0

Failure: Negative error code for exceptions

[Function Description]

Configure camera registers and start streaming.

【Notes】

If the camera handle has been attached to a vflow, this interface may not need to be called. If it is called, hbn_vflow_start must be called first, followed by hbn_camera_start.

8. hbn_camera_stop

【Function Declaration】

int32_t hbn_camera_stop(camera_handle_t cam_fd)

【Parameter Description】

[IN] camera_handle_t cam_fd: Camera handle created by hbn_camera_create;

【Return Value】

Success: RET_OK 0

Failure: Negative error code for exceptions

【Function Description】

Stop the camera stream.

【Notes】

Must be used in pair with hbn_camera_start.

9. hbn_camera_reset

【Function Declaration】

int32_t hbn_camera_reset(camera_handle_t cam_fd)

【Parameter Description】

[IN] camera_handle_t cam_fd: Camera handle created by hbn_camera_create;

【Return Value】

Success: RET_OK 0

Failure: Negative error code for exceptions

【Function Description】

Reset the sensor by re-initializing it.

【Notes】

If this API is called before attaching the camera to VIN, sensor initialization will be performed via the camera_attach_to_vin interface to achieve the reset effect. If this API is called after attaching the camera to VIN, the sequence sensor_stop → sensor_deinit → sensor_init → sensor_start will be executed to reset the sensor.

10. hbn_camera_change_fps

【Function Declaration】

int32_t hbn_camera_change_fps(camera_handle_t cam_fd, int32_t fps)

【Parameter Description】

[IN] camera_handle_t cam_fd: Camera handle created by hbn_camera_create;

[IN] int32_t fps: Output frame rate of the sensor;

【Return Value】

Success: RET_OK 0

Failure: Negative error code for exceptions

【Function Description】

Dynamically switch the sensor frame rate.

【Notes】

This feature requires implementing the corresponding callback function dynamic_switch_fps in the sensor library.

11. hbn_camera_read_register

【Function Declaration】

int32_t hbn_camera_read_register(camera_handle_t cam_fd, camera_reg_type_t type,
uint32_t reg_addr)

【Parameter Description】

[IN] camera_handle_t cam_fd: Camera handle created by hbn_camera_create;

[IN] camera_reg_type_t type: Type of sensor register to read;

[IN] uint32_t reg_addr: Register address;

【Return Value】

Success: RET_OK 0

Failure: Negative error code for exceptions

【Function Description】

Read the value of a camera register.

【Notes】

None

12. hbn_camera_get_handle

【Function Declaration】

camera_handle_t hbn_camera_get_handle(vpf_handle_t vin_fd, int32_t camera_index)

【Parameter Description】

[IN] vpf_handle_t vin_fd: File descriptor (fd) of the VIN node;

[IN] int32_t camera_index: Port index of the camera;

【Return Value】

Success: RET_OK 0

Failure: Negative error code for exceptions

【Function Description】

Obtain the corresponding camera handle via the VIN node handle or camera port index.

【Notes】

None

13. hbn_camera_init_cfg

【Function Declaration】

int32_t hbn_camera_init_cfg(const char *cfg_file)

【Parameter Description】

[IN] const char *cfg_file: Path to the camera configuration file (JSON format);

【Return Value】

Success: RET_OK 0

Failure: Negative error code for exceptions

【Function Description】

Create a camera handle and a deserializer handle based on the provided configuration and bind them together.

【Notes】

This API creates cameras by parsing a JSON configuration file, which differs from the non-JSON approach used in the sample code. For further details, please consult your FAE.

tip

The commercial version offers more comprehensive feature support, deeper hardware capability exposure, and exclusive customization options. To ensure compliance and secure delivery, access to the commercial version will be granted through the following process:

Commercial Version Access Procedure:

1. Complete a questionnaire: Submit basic information about your organization and intended use case.
2. Sign a Non-Disclosure Agreement (NDA): We will contact you based on your submission, and both parties will sign the NDA upon mutual confirmation.
3. Content release: After the NDA is signed, we will provide access to the commercial version materials through a private channel.

If you wish to obtain the commercial version, please fill out the questionnaire below. We will contact you within 3–5 business days:

Questionnaire link: https://horizonrobotics.feishu.cn/share/base/form/shrcnJQBMIkRm6K79rjXR0hr0Fg

14. hbn_deserial_create

【Function Declaration】

int32_t hbn_deserial_create(deserial_config_t *des_config, deserial_handle_t
*des_fd)

【Parameter Description】

[IN] deserial_config_t *des_config: Pointer to the deserializer configuration parameter structure;

[OUT] deserial_handle_t *des_fd: Deserializer handle created according to the configuration;

【Return Value】

Success: RET_OK 0

Failure: Negative error code for exceptions

【Function Description】

Create a deserializer handle based on the provided configuration.

【Notes】

This API should only be called when a hardware deserializer is present.

This API validates the deserializer configuration; an error will be returned if the configuration exceeds allowed limits.

This API also checks the deserializer library; an error will be returned if it does not comply with HBN architecture specifications.

15. hbn_deserial_destroy

【Function Declaration】

int32_t hbn_deserial_destroy(deserial_handle_t des_fd)

【Parameter Description】

[IN] deserial_handle_t des_fd: Deserializer handle created by hbn_deserial_create;

【Return Value】

Success: RET_OK 0

Failure: Negative error code for exceptions

【Function Description】

Destroy the corresponding software resources based on the deserializer handle.

【Notes】

hbn_deserial_destroy must be used in pair with hbn_deserial_create.

16. hbn_deserial_attach_to_vin

【Function Declaration】

int32_t hbn_deserial_attach_to_vin(deserial_handle_t des_fd, camera_des_link_t
link, vpf_handle_t vin_fd)

【Parameter Description】

[IN] deserial_handle_t des_fd: Deserializer handle, created by hbn_deserial_create;

[IN] camera_des_link_t link: Link index of the deserializer;

[IN] vpf_handle_t vin_fd: VIN node handle to bind to;

【Return Value】

Success: RET_OK 0

Failure: Negative error code for exceptions

【Function Description】

Bind the deserializer to a VIN node.

【Notes】

If the hardware includes a deserializer, the camera should be bound to the deserializer, and the deserializer should be bound to the VIN node.

17. hbn_deserial_detach_from_vin

【Function Declaration】

int32_t hbn_deserial_detach_from_vin(deserial_handle_t des_fd, camera_des_link_t link)

【Parameter Description】

[IN] deserial_handle_t des_fd: Deserializer handle, created by hbn_deserial_create;

[IN] camera_des_link_t link: Link index of the deserializer;

【Return Value】

Success: RET_OK 0

Failure: Negative error code for exceptions

【Function Description】

Unbind the deserializer from the VIN node.

【Notes】

hbn_deserial_detach_from_vin must be used in pair with hbn_deserial_attach_to_vin.

18. hbn_txser_create

【Function Declaration】

int32_t hbn_txser_create(txser_config_t *txs_config, txser_handle_t *txs_fd)

【Parameter Description】

[IN] txser_config_t *txs_config: Pointer to the TX serializer configuration structure;

[OUT] txser_handle_t *txs_fd: TX serializer handle created according to the configuration;

【Return Value】

Success: RET_OK 0

Failure: Negative error code for exceptions

【Function Description】

Create a TX serializer handle based on the provided configuration.

【Notes】

This API should only be called when the hardware includes a serializer.

The API validates the TX serializer configuration; if the configuration exceeds allowed ranges, an error will be returned.

The API also checks the TX serializer library; if it does not comply with HBN architecture specifications, an error will be returned.

19. hbn_txser_destroy

【Function Declaration】

int32_t hbn_txser_destroy(txser_handle_t txs_fd)

【Parameter Description】

[IN] txser_handle_t txs_fd: TX serializer handle, created by hbn_txser_create;

【Return Value】

Success: RET_OK 0

Failure: Negative error code for exceptions

【Function Description】

Destroy software resources associated with the TX serializer handle.

【Notes】

This API should only be called when the hardware includes a serializer.

hbn_txser_destroy must be used in pair with hbn_txser_create.

20. hbn_txser_attach_to_vin

【Function Declaration】

int32_t hbn_txser_attach_to_vin(txser_handle_t txs_fd, camera_txs_csi_t csi, vpf_handle_t vin_fd)

【Parameter Description】

[IN] txser_handle_t txs_fd: TX serializer handle, created by hbn_txser_create;

[IN] camera_txs_csi_t csi: TX CSI index;

[IN] vpf_handle_t vin_fd: VIN node handle to bind to;

【Return Value】

Success: RET_OK 0

Failure: Negative error code for exceptions

【Function Description】

Bind the TX serializer to a VIN node.

【Notes】

This API should only be called when the hardware includes a serializer.

This API initializes the TX serializer hardware.

If the hardware includes a serializer, the camera should be bound to the TX serializer, and the TX serializer should be bound to the VIN node.

21. hbn_txser_detach_from_vin

【Function Declaration】

int32_t hbn_txser_detach_from_vin(txser_handle_t txs_fd, camera_txs_csi_t csi)

【Parameter Description】

[IN] txser_handle_t txs_fd: TX serializer handle, created by hbn_txser_create;

[IN] camera_txs_csi_t csi: TX CSI index;

【Return Value】

Success: RET_OK 0

Failure: Negative error code for exceptions

【Function Description】

Unbind the TX serializer from the VIN node.

【Notes】

hbn_txser_detach_from_vin must be used in pair with hbn_txser_attach_to_vin.

Parameter Description

typedef struct camera_config_s

Name	Type	Min Value	Max Value	Default Value	Description	Required
name[CAMERA_MODULE_NAME_LEN]	char	–	CAMERA_MODULE_NAME_LEN(108)	–	Camera module name, which must correspond to the sensor library name. For example, if the sensor driver is named libimx219.so, then name should be "imx219".	Yes
addr	uint32_t	0x00	0x7f	0x00	Sensor device address, typically a 7-bit I2C address.	Yes
isp_addr	uint32_t	0x00	0x7f	0x00	ISP device address (if any); none by default.	No
eeprom_addr	uint32_t	0x00	0x7f	0x00	EEPROM device address (if any); none by default.	No
serial_addr	uint32_t	0x00	0x7f	0x00	SerDes device address (if any); none by default.	No
sensor_mode	uint32_t	1	5	1	Sensor operating mode, selectable via enum sensor_mode_e as defined below:	Yes
					1: NORMAL_M, linear mode;
					2: DOL2_M, HDR mode combining 2 frames into 1;
					3: DOL3_M, HDR mode combining 3 frames into 1;
					4: DOL4_M, HDR mode combining 4 frames into 1;
					5: PWL_M, HDR mode with internal frame combination in the sensor.
sensor_clk	uint32_t	–	–	0x00	Sensor clock configuration; currently unused and reserved for future use.	No
gpio_enable	uint32_t	0	0xFFFFFFFF	0	Whether to use X5 GPIOs to control camera sensor pins to meet power-up/power-down timing requirements.	Yes
					Example: Use GPIO to control the sensor's XSHUTDN pin.
					Note: Corresponding GPIO numbers must be configured in the device tree (DTS).
					0: Do not use GPIO control;
					Non-zero: Enable GPIO control for the sensor, with each bit enabling a specific GPIO. For example, 0x07 enables three GPIOs: [gpio_a, gpio_b, gpio_c].
gpio_level	uint32_t	0	1	0	If gpio_enable is set, gpio_level bits configure the initial logic levels of the GPIOs controlling sensor pins. The relationship between a GPIO bit and its output sequence is as follows:	Yes
					0: Output low first, sleep 1s, then output high;
					1: Output high first, sleep 1s, then output low.
					Example: 0x05 = 101 (binary). From bit0 to bit2: gpio_a outputs high then low, gpio_b outputs low then high, gpio_c outputs high then low.
					Customize according to the sensor’s power-up timing specification.
bus_select	uint32_t	0	6	0	Selects the I2C bus number for the sensor. Typically fixed by hardware; thus, it is recommended to configure this in the DTS instead of here.	No
					For details on binding the sensor I2C in DTS, refer to the "Camera Bring-up Guide".
bus_timeout	uint32_t	0	–	0	I2C timeout configuration. Only required if bus_select is configured.	No
fps	uint32_t	0	120	0	Sensor frame rate configuration.	Yes
width	uint32_t	0	8192	0	Sensor output image width (in pixels).	Yes
height	uint32_t	0	4096	0	Sensor output image height (in pixels).	Yes
format	uint32_t	–	–	–	Sensor data format. Common values include:	Yes
					RAW8: 0x2A;
					RAW10: 0x2B;
					RAW12: 0x2C;
					YUV422 8-bit: 0x1E
flags	uint32_t	0	-	0	Optional features: diagnostics, recovery, debug, etc.	No
extra_mode	uint32_t	0	-	0	Custom configurations inside each sensor library: mostly used to differentiate modules and toggle features.	Yes
config_index	uint32_t	0	-	0	Custom configurations inside each sensor library: mostly used to differentiate modules and toggle features.	Yes
ts_compensate	uint32_t	0	-	0	Reserved parameter, for future use.	No
mipi_cfg	mipi_config_t	-	-	-	MIPI configuration. Set to NULL to automatically fetch configuration from the sensor driver (via get_csi_attr).	Yes
calib_lname	char	-	-	-	Path to the sensor calibration library. Default path: /usr/hobot/lib/sensor	Yes
sensor_param	char	-	-	-	Sensor custom data.	No
iparam_mode	uint32_t	-	-	-	Reserved parameter, for future use.	No
end_flag	uint32_t	-	-	-	End-of-structure marker. Default value: CAMERA_CONFIG_END_FLAG	Yes

typedef struct deserial_config_s

Name	Type	Min	Max	Default	Description	Required
name	char[CAMERA_MODULE_NAME_LEN]	-	-	-	Deserializer name, e.g., max9296.	Yes
addr	uint32_t	0	-	-	Deserializer device address.	Yes
gpio_enable	uint32_t	0	-	-	GPIO operation enable bit, indexed from VCON.	Yes
gpio_level	uint32_t	0	-	-	GPIO operational state bit, indicating current GPIO status.	Yes
gpio_mfp	uint8_t[CAMERA_DES_GPIO_MAX]	0	CAMERA_DES_GPIO_MAX	0x0	GPIO MFP function selection, used to specify multifunctional GPIO configuration.	Yes
bus_select	uint32_t	0	-	-	I2C bus selection, indexed from VCON.	Yes
bus_timeout	uint32_t	0	-	-	I2C timeout setting, in milliseconds.	Yes
lane_mode	uint32_t	0	-	-	PHY lane mode selection.	Yes
lane_speed	uint32_t	0	-	-	PHY lane speed configuration.	Yes
link_map	uint32_t	0	-	-	Mapping configuration between Link and CSI/VC.	Yes
link_desp	char[CAMERA_DES_LINKMAX][CAMERA_DES_PORTDESP_LEN]	-	-	-	Configuration descriptions for each Link-connected module, used in multi-process scenarios.	Yes
reset_delay	uint32_t	0	-	-	Delay time for reset operation, in milliseconds.	Yes
flags	uint32_t	0	-	-	Optional feature flags, e.g., diagnostics, debugging, etc.	No
poc_cfg	poc_config_t*	-	-	NULL	POC configuration pointer. If NULL, POC functionality is disabled.	No
mipi_cfg	mipi_config_t*	-	-	NULL	MIPI configuration pointer. If NULL, configuration is fetched automatically.	No
deserial_param	char*	-	-	NULL	Pointer to deserializer custom data.	No
end_flag	uint32_t	0	0xFFFFFFFF	-	End-of-structure marker. Default: DESERIAL_CONFIG_END_FLAG	Yes

typedef struct poc_config_s

Name	Type	Min	Max	Default	Description	Required
name	char[CAMERA_MODULE_NAME_LEN]	-	-	-	POC name, e.g., max20087.	Yes
addr	uint32_t	0	-	-	POC device address.	Yes
gpio_enable	uint32_t	0	-	-	GPIO operation enable bit, indexed from VCON.	Yes
gpio_level	uint32_t	0	-	-	GPIO operational state bit, indicating current GPIO status.	Yes
poc_map	uint32_t	0	-	-	Mapping between POC and Link.	Yes
power_delay	uint32_t	0	-	-	Delay time for POC power on/off operations, in milliseconds.	Yes
end_flag	uint32_t	0	0xFFFFFFFF	-	End-of-structure marker for integrity verification. Default: POC_CONFIG_END_FLAG	Yes

Return Value Description

Error Code	Macro Definition	Description	Common Causes and Solutions
0	HBN_STATUS_SUCESS	Success
1	HBN_STATUS_INVALID_NODE	Invalid vnode; corresponding vnode not found
2	HBN_STATUS_INVALID_NODETYPE	Invalid vnode type; corresponding vnode not found	For VIN, vnode type must be HB_VIN
3	HBN_STATUS_INVALID_HWID	Invalid hardware module ID	For VIN, hw_id must be 0
4	HBN_STATUS_INVALID_CTXID	Invalid context ID	Can be set to AUTO_ALLOC_ID for automatic allocation by HBN framework
8	HBN_STATUS_INVALID_NULL_PTR	Null pointer
9	HBN_STATUS_INVALID_PARAMETER	Invalid parameter; version check failed
10	HBN_STATUS_ILLEGAL_ATTR	Invalid parameter
11	HBN_STATUS_INVALID_FLOW	Invalid flow; corresponding flow not found
12	HBN_STATUS_FLOW_EXIST	Flow already exists
13	HBN_STATUS_FLOW_UNEXIST	Flow does not exist
14	HBN_STATUS_NODE_EXIST	Node already exists
15	HBN_STATUS_NODE_UNEXIST	Node does not exist
16	HBN_STATUS_NOT_CONFIG	Reserved
19	HBN_STATUS_ALREADY_BINDED	Node already bound
20	HBN_STATUS_NOT_BINDED	Node not bound
21	HBN_STATUS_TIMEOUT	Timeout
22	HBN_STATUS_NOT_INITIALIZED	Not initialized
23	HBN_STATUS_NOT_SUPPORT	Channel not supported or not activated
24	HBN_STATUS_NOT_PERM	Operation not permitted
25	HBN_STATUS_NOMEM	Insufficient memory
31	HBN_STATUS_JSON_PARSE_FAIL	JSON parsing failed
34	HBN_STATUS_SET_CONTROL_FAIL	Failed to set module control or tuning parameters (e.g., ISP effect parameters)
35	HBN_STATUS_GET_CONTROL_FAIL	Failed to get module control or tuning parameters (e.g., ISP effect parameters)
36	HBN_STATUS_NODE_START_FAIL	Failed to start node
37	HBN_STATUS_NODE_STOP_FAIL	Failed to stop node
42	HBN_STATUS_NODE_ILLEGAL_EVENT	Illegal event during node channel polling
43	HBN_STATUS_NODE_DEQUE_ERROR	Error dequeuing buffer from node channel
51	HBN_STATUS_INVALID_VERSION	Version mismatch between driver and upper-layer library
52	HBN_STATUS_GET_VERSION_ERROR	Failed to retrieve driver module version
128	HBN_STATUS_ERR_UNKNOW	Unknown error

Camera Bring-up

HBN API

Description

In software, the Camera uses a dedicated set of APIs. Modules downstream of the Camera are abstracted as vnodes. These vnodes include VIN, ISP, PYM, and GDC.
Multiple vnodes form a vflow (similar to a pipeline). The Camera and VIN are bound together via the attach interface.
Users only need to call HBN APIs to initialize and bind modules. Once the vflow is created and started, users do not need to manage frame data transfer—the SDK internally passes frames from upstream to downstream.

A vflow consists of one or more vnodes. Each vnode has one input channel and one or more output channels.

Example API usage:

API Reference

1. hbn_vnode_open

【Function Declaration】

hobot_status hbn_vnode_open(hb_vnode_type vnode_type, uint32_t hw_id, int32_t ctx_id, hbn_vnode_handle_t *vnode_fd)

【Parameter Description】

[IN] hb_vnode_type vnode_type: Vnode type. Each hardware module corresponds to a specific vnode type, such as HB_VIN, HB_ISP, HB_PYM, etc.

[IN] uint32_t hw_id: Hardware ID of the module.

[IN] uint32_t ctx_id: Context ID of the module (a software concept). You can either specify a context ID or set it to AUTO_ALLOC_ID to let the SDK allocate one automatically.

[OUT] hbn_vnode_handle_t *vnode_fd: Returns the vnode handle of the module.

【Return Value】

Success: HBN_STATUS_SUCESS (0)
Failure: Negative error code; refer to the Return Value Description table.

【Function Description】

Initializes a specific module, opens its device node, and returns the module's vnode handle.

【Notes】

None.

2. hbn_vnode_close

【Function Declaration】

void hbn_vnode_close(hbn_vnode_handle_t vnode_fd)

【Parameter Description】

[IN] hbn_vnode_handle_t vnode_fd: Vnode handle of the module.

【Return Value】

None

【Function Description】

Closes the module's device node.

【Notes】

If hbn_vflow_destroy has been called, there is no need to call hbn_vnode_close.
hbn_vnode_close should only be used when a module is used independently (e.g., GDC loopback). If the module is part of a vflow, calling hbn_vflow_destroy is sufficient—do not call hbn_vnode_close.

3. hbn_vnode_set_attr

【Function Declaration】

hobot_status hbn_vnode_set_attr(hbn_vnode_handle_t vnode_fd, void *attr)

【Parameter Description】

[IN] hbn_vnode_handle_t vnode_fd: Vnode handle of the module.

[IN] void *attr: Pointer to the module's basic attribute structure. This structure can be vin_attr_t, isp_attr_t, pym_attr_t, etc.—i.e., any structure named <module_name>_attr_t.

【Return Value】

Success: HBN_STATUS_SUCESS (0)
Failure: Negative error code; refer to the Return Value Description table.

【Function Description】

Sets the basic attributes of a module.

【Notes】

None.

4. hbn_vnode_get_attr

【Function Declaration】

hobot_status hbn_vnode_get_attr(hbn_vnode_handle_t vnode_fd, void *attr)

【Parameter Description】

[IN] hbn_vnode_handle_t vnode_fd: Vnode handle of the module.

[OUT] void *attr: Pointer to the module's basic attribute structure (e.g., vin_attr_t, isp_attr_t, pym_attr_t, etc.).*attr: Pointer to the basic attribute structure of the module. The basic attribute structure can be vin_attr_t, isp_attr_t, pym_attr_t, etc.—any attribute structure named with the module name followed by _attr_t.

【Return Value】

Success: HBN_STATUS_SUCCESS 0

Failure: Negative error code for exceptions; refer to the return value description.

【Function Description】

Obtains the basic attributes of a module.

【Notes】

None

5. hbn_vnode_set_attr_ex

【Function Declaration】

hobot_status hbn_vnode_set_attr_ex(hbn_vnode_handle_t vnode_fd, void *attr)

【Parameter Description】

[IN] hbn_vnode_handle_t vnode_fd: vnode handle of the module;

[IN] void *attr: Pointer to the extended attribute structure of the module. The extended attribute structure can be vin_attr_ex_t, etc.—any attribute structure named with the module name followed by _attr_ex_t;

【Return Value】

Success: HBN_STATUS_SUCCESS 0

Failure: Negative error code for exceptions; refer to the return value description.

【Function Description】

Sets the extended attributes of a module, which can be dynamically configured during application runtime.

【Notes】

None

6. hbn_vnode_get_attr_ex

【Function Declaration】

hobot_status hbn_vnode_get_attr_ex(hbn_vnode_handle_t vnode_fd, void *attr)

【Parameter Description】

[IN] hbn_vnode_handle_t vnode_fd: vnode handle of the module;

[OUT] void
*attr: Pointer to the extended attribute structure of the module. The extended attribute structure can be vin_attr_ex_t, etc.—any attribute structure named with the module name followed by _attr_ex_t;

【Return Value】

Success: HBN_STATUS_SUCCESS 0

Failure: Negative error code for exceptions; refer to the return value description.

【Function Description】

Obtains the extended attributes of a module.

【Notes】

None

7. hbn_vnode_set_ochn_attr

【Function Declaration】

hobot_status hbn_vnode_set_ochn_attr(hbn_vnode_handle_t vnode_fd, uint32_t
ochn_id, void *attr)

【Parameter Description】

[IN] hbn_vnode_handle_t vnode_fd: vnode handle of the module;

[IN] uint32_t ochn_id: Output channel ID of the module; refer to the module channel description for valid channel IDs;

[IN] void *attr: Pointer to the output channel attribute structure of the module. The output channel attribute can be vin_ochn_attr_t, isp_ochn_attr_t, etc.—any attribute structure named with the module name followed by _ochn_attr_t;

【Return Value】

Success: HBN_STATUS_SUCCESS 0

Failure: Negative error code for exceptions; refer to the return value description.

【Function Description】

Sets the output channel attributes of a module.

【Notes】

None

8. hbn_vnode_get_ochn_attr

【Function Declaration】

hobot_status hbn_vnode_get_ochn_attr(hbn_vnode_handle_t vnode_fd, uint32_t
ochn_id, void *attr)

【Parameter Description】

[IN] hbn_vnode_handle_t vnode_fd: vnode handle of the module;

[IN] uint32_t ochn_id: Output channel ID of the module; refer to the module channel description for valid channel IDs;

[OUT] void *attr: Pointer to the output channel attribute structure of the module. The output channel attribute can be vin_ochn_attr_t, isp_ochn_attr_t, etc.—any attribute structure named with the module name followed by _ochn_attr_t;

【Return Value】

Success: HBN_STATUS_SUCCESS 0

Failure: Negative error code for exceptions; refer to the return value description.

【Function Description】

Obtains the output channel attributes of a module.

【Notes】

None

9. hbn_vnode_set_ochn_attr_ex

【Function Declaration】

hobot_status hbn_vnode_set_ochn_attr_ex(hbn_vnode_handle_t vnode_fd, uint32_t
ochn_id, void *attr)

【Parameter Description】

[IN] hbn_vnode_handle_t vnode_fd: vnode handle of the module;

[IN] uint32_t ochn_id: Output channel ID of the module; refer to the module channel description for valid channel IDs;

[IN] void *attr: Pointer to the extended output channel attribute structure of the module. The extended output channel attribute can be pym_ochn_attr_ex_t, etc.—any attribute structure named with the module name followed by _ochn_attr_ex_t;

【Return Value】

Success: HBN_STATUS_SUCCESS 0

Failure: Negative error code for exceptions; refer to the return value description.

【Function Description】

Sets the extended output channel attributes of a module, which can be dynamically configured during application runtime.

【Notes】

None

10. hbn_vnode_set_ichn_attr

【Function Declaration】

hobot_status hbn_vnode_set_ichn_attr(hbn_vnode_handle_t vnode_fd, uint32_t
ichn_id, void *attr)

【Parameter Description】

[IN] hbn_vnode_handle_t vnode_fd: vnode handle of the module;

[IN] uint32_t ichn_id: Input channel ID of the module; refer to the module channel description for valid channel IDs;

[IN] void *attr: Pointer to the input channel attribute structure of the module. The input channel attribute can be vin_ichn_attr_t, isp_ichn_attr_t, etc.—any attribute structure named with the module name followed by _ichn_attr_t;

【Return Value】

Success: HBN_STATUS_SUCCESS 0

Failure: Negative error code for exceptions; refer to the return value description.

【Function Description】

Sets the input channel attributes of a module.

【Notes】

None

11. hbn_vnode_get_ichn_attr

【Function Declaration】

hobot_status hbn_vnode_get_ichn_attr(hbn_vnode_handle_t vnode_fd, uint32_t
ichn_id, void *attr)

【Parameter Description】

[IN] hbn_vnode_handle_t vnode_fd: vnode handle of the module;[IN] uint32_t ichn_id: Input channel ID of the module; refer to the module channel description for channel IDs.

[OUT] void *attr: Pointer to the input channel attribute structure of the module. The input channel attributes can be vin_ichn_attr_t, isp_ichn_attr_t, etc.—any attribute ending with the module name followed by _ichn_attr_t.

【Return Value】

Success: HBN_STATUS_SUCCESS 0
Failure: Negative error code for exceptions; refer to the return value description.

【Function Description】

Obtain the input channel attributes of the module.

【Notes】

None

12. hbn_vnode_set_ochn_buf_attr

【Function Declaration】

hobot_status hbn_vnode_set_ochn_buf_attr(hbn_vnode_handle_t vnode_fd, uint32_t
ochn_id, hbn_buf_alloc_attr_t *alloc_attr)

【Parameter Description】

[IN] hbn_vnode_handle_t vnode_fd: vnode handle of the module;

[IN] uint32_t ochn_id: Output channel ID of the module; refer to the module channel description for channel IDs;

[IN] hbn_buf_alloc_attr_t *alloc_attr: Buffer allocation attributes;

【Return Value】

Success: HBN_STATUS_SUCCESS 0
Failure: Negative error code for exceptions; refer to the return value description.

【Function Description】

Set the buffer attributes for the output channel.

【Notes】

None

13. hbn_vnode_start

【Function Declaration】

hobot_status hbn_vnode_start(hbn_vnode_handle_t vnode_fd)

【Parameter Description】

[IN] hbn_vnode_handle_t vnode_fd: vnode handle of the module;

【Return Value】

Success: HBN_STATUS_SUCCESS 0
Failure: Negative error code for exceptions; refer to the return value description.

【Function Description】

Start the module.

【Notes】

The module must be opened before starting.

14. hbn_vnode_stop

【Function Declaration】

hobot_status hbn_vnode_stop(hbn_vnode_handle_t vnode_fd)

【Parameter Description】

[IN] hbn_vnode_handle_t vnode_fd: vnode handle of the module;

【Return Value】

Success: HBN_STATUS_SUCCESS 0
Failure: Negative error code for exceptions; refer to the return value description.

【Function Description】

Stop the module.

【Notes】

None

15. hbn_vnode_getframe

【Function Declaration】

hobot_status hbn_vnode_getframe(hbn_vnode_handle_t vnode_fd, uint32_t ochn_id,
uint32_t millisecondTimeout, hbn_vnode_image_t *out_img)

【Parameter Description】

[IN] hbn_vnode_handle_t vnode_fd: vnode handle of the module;

[IN] uint32_t ochn_id: Output channel ID of the module; refer to the module channel description for channel IDs;

[IN] uint32_t millisecondTimeout: Timeout waiting period (in milliseconds);

[OUT] hbn_vnode_image_t *out_img: Address of the output image buffer structure;

【Return Value】

Success: HBN_STATUS_SUCCESS 0
Failure: Negative error code for exceptions; refer to the return value description.

【Function Description】

Retrieve an image from the module's output channel; this is a blocking interface.

【Notes】

None

16. hbn_vnode_releaseframe

【Function Declaration】

hobot_status hbn_vnode_releaseframe(hbn_vnode_handle_t vnode_fd, uint32_t
ochn_id, hbn_vnode_image_t *img)

【Parameter Description】

[IN] hbn_vnode_handle_t vnode_fd: vnode handle of the module;

[IN] uint32_t ochn_id: Output channel ID of the module; refer to the module channel description for channel IDs;

[IN] hbn_vnode_image_t *img: Address of the image buffer structure;

【Return Value】

Success: HBN_STATUS_SUCCESS 0
Failure: Negative error code for exceptions; refer to the return value description.

【Function Description】

Release the image buffer; the buffer will be returned to the specified output channel.

【Notes】

None

17. hbn_vnode_getframe_group

【Function Declaration】

hobot_status hbn_vnode_getframe_group(hbn_vnode_handle_t vnode_fd, uint32_t
ochn_id, uint32_t millisecondTimeout, hbn_vnode_image_group_t *out_img);

【Parameter Description】

[IN] hbn_vnode_handle_t vnode_fd: vnode handle of the module;

[IN] uint32_t ochn_id: Output channel ID of the module; refer to the module channel description for channel IDs;

[IN] uint32_t millisecondTimeout: Timeout waiting period (in milliseconds);

[OUT] hbn_vnode_image_group_t *out_img: Address of the output image buffer structure;

【Return Value】

Success: HBN_STATUS_SUCCESS 0
Failure: Negative error code for exceptions; refer to the return value description.

【Function Description】

Retrieve a multi-layer aggregated image from the module's output channel; this is a blocking interface.

【Notes】

This interface must be used to obtain output images from ISP and PYM.

18. hbn_vnode_releaseframe_group

【Function Declaration】

hobot_status hbn_vnode_releaseframe_group(hbn_vnode_handle_t vnode_fd, uint32_t
ochn_id, hbn_vnode_image_group_t *img)

【Parameter Description】

[IN] hbn_vnode_handle_t vnode_fd: vnode handle of the module;

[IN] uint32_t ochn_id: Output channel ID of the module; refer to the module channel description for channel IDs;
[IN] hbn_vnode_image_t *img: Address of the image buffer structure;

【Return Value】

Success: HBN_STATUS_SUCCESS 0

Failure: Negative error code for exceptions; refer to the return value description.

【Function Description】

Releases a multi-layer aggregated image buffer. The buffer will be returned to the specified output channel.

【Notes】

None

19. hbn_vnode_sendframe

【Function Declaration】

hobot_status hbn_vnode_sendframe(hbn_vnode_handle_t vnode_fd, uint32_t ichn_id, hbn_vnode_image_t *img)

【Parameter Description】

[IN] hbn_vnode_handle_t vnode_fd: Vnode handle of the module;

[IN] uint32_t ichn_id: Input channel ID of the module; refer to the module channel description for channel IDs;

[IN] hbn_vnode_image_t *img: Address of the input image buffer;

【Return Value】

Success: HBN_STATUS_SUCCESS 0

Failure: Negative error code for exceptions; refer to the return value description.

【Function Description】

Sends an image to the input channel of the module, triggering the module to process it. This is a blocking API that waits until hardware processing completes before returning, with a default timeout of 1 second.

【Notes】

None

20. hbn_vflow_create

【Function Declaration】

hobot_status hbn_vflow_create(hbn_vflow_handle_t *vflow_fd)

【Parameter Description】

[OUT] hbn_vflow_handle_t *vflow_fd: Vflow handle;

【Return Value】

Success: HBN_STATUS_SUCCESS 0

Failure: Negative error code for exceptions; refer to the return value description.

【Function Description】

Creates a vflow and returns the vflow handle.

【Notes】

None

21. hbn_vflow_destroy

【Function Declaration】

void hbn_vflow_destroy(hbn_vflow_handle_t vflow_fd)

【Parameter Description】

[IN] hbn_vflow_handle_t *vflow_fd: Vflow handle;

【Return Value】

None

【Function Description】

Destroys a vflow based on the provided vflow handle.

【Notes】

None

22. hbn_vflow_add_vnode

【Function Declaration】

hobot_status hbn_vflow_add_vnode(hbn_vflow_handle_t vflow_fd, hbn_vnode_handle_t vnode_fd)

【Parameter Description】

[IN] hbn_vflow_handle_t *vflow_fd: Vflow handle;

[IN] hbn_vnode_handle_t vnode_fd: Vnode handle of the module;

【Return Value】

Success: HBN_STATUS_SUCCESS 0

Failure: Negative error code for exceptions; refer to the return value description.

【Function Description】

Adds a module into the vflow for management by the vflow.

【Notes】

None

23. hbn_vflow_bind_vnode

【Function Declaration】

hobot_status hbn_vflow_bind_vnode(hbn_vflow_handle_t vflow_fd, hbn_vnode_handle_t src_vnode_fd, uint32_t out_chn, hbn_vnode_handle_t dst_vnode_fd, uint32_t in_chn)

【Parameter Description】

[IN] hbn_vflow_handle_t *vflow_fd: Vflow handle;

[IN] hbn_vnode_handle_t src_vnode_fd: Vnode handle of the source module;

[IN] uint32_t out_chn: Output channel ID of the source module; refer to the module channel description for channel IDs;

[IN] hbn_vnode_handle_t dst_vnode_fd: Vnode handle of the destination module;

[IN] uint32_t in_chn: Input channel ID of the destination module; refer to the module channel description for channel IDs;

【Return Value】

Success: HBN_STATUS_SUCCESS 0

Failure: Negative error code for exceptions; refer to the return value description.

【Function Description】

Binds two modules together. After binding, data frames from the src_vnode_fd module will automatically flow to the dst_vnode_fd module.

【Notes】

The flow must be created, and the modules must be opened.

24. hbn_vflow_unbind_vnode

【Function Declaration】

hobot_status hbn_vflow_unbind_vnode(hbn_vflow_handle_t vflow_fd, hbn_vnode_handle_t src_vnode_fd, uint32_t out_chn, hbn_vnode_handle_t dst_vnode_fd, uint32_t in_chn)

【Parameter Description】

[IN] hbn_vflow_handle_t *vflow_fd: Vflow handle;

[IN] hbn_vnode_handle_t src_vnode_fd: Vnode handle of the source module;

[IN] uint32_t out_chn: Output channel ID of the source module; refer to the module channel description for channel IDs;

[IN] hbn_vnode_handle_t dst_vnode_fd: Vnode handle of the destination module;

[IN] uint32_t in_chn: Input channel ID of the destination module; refer to the module channel description for channel IDs;

【Return Value】

Success: HBN_STATUS_SUCCESS 0

Failure: Negative error code for exceptions; refer to the return value description.

【Function Description】

Unbinds the src_vnode_fd and dst_vnode_fd modules.

【Notes】

Not supported currently.

25. hbn_vflow_start

【Function Declaration】

hobot_status hbn_vflow_start(hbn_vflow_handle_t vflow_fd)

【Parameter Description】

[IN] hbn_vflow_handle_t vflow_fd: Vflow handle;

【Return Value】

Success: HBN_STATUS_SUCCESS 0
Failure: Exception indicated by a negative error code; refer to the return value description.

【Function Description】

Start a vflow. All vnodes contained within the vflow will be started.

【Notes】

The module vnode must be added to the vflow in advance.

26. hbn_vflow_stop

【Function Declaration】

hobot_status hbn_vflow_stop(hbn_vflow_handle_t vflow_fd)

【Parameter Description】

[IN] hbn_vflow_handle_t vflow_fd: vflow handle;

【Return Value】

Success: HBN_STATUS_SUCCESS 0
Failure: Exception indicated by a negative error code; refer to the return value description.

【Function Description】

Stop a vflow. All vnodes contained within the vflow will be stopped.

【Notes】

This function should be used in pair with hbn_vflow_start.

27. hbn_vflow_get_vnode_handle

【Function Declaration】

hbn_vnode_handle_t hbn_vflow_get_vnode_handle(hbn_vflow_handle_t vflow_fd,
hb_vnode_type vnode_type, uint32_t index)

【Parameter Description】

[IN] hbn_vflow_handle_t vflow_fd: vflow handle;
[IN] hb_vnode_type vnode_type: module ID;
[IN] uint32_t index: context ID, range [0, 7]

【Return Value】

Success: HBN_STATUS_SUCCESS 0
Failure: Exception indicated by a negative error code; refer to the return value description.

【Function Description】

Obtain the vnode handle via module ID and context ID.

【Notes】

The module must be opened in advance.

ISP API

1. hbn_isp_get_ae_statistics

【Function Declaration】

int32_t hbn_isp_get_ae_statistics(hbn_vnode_handle_t vnode_fd, isp_statistics_t *ae_statistics, int32_t time_out)

【Parameter Description】

[IN] hbn_vflow_handle_t vflow_fd: vflow handle;
[IN] int32_t time_out: timeout duration for waiting;
[OUT] isp_statistics_t *ae_statistics: AE statistics;

【Return Value】

Success: HBN_STATUS_SUCCESS 0
Failure: Exception indicated by a negative error code; refer to the return value description.

【Function Description】

Obtain AE statistics (1024-bin histogram) via the vflow_fd of the ISP module. The data is fixed at 2096 bytes, arranged as shown in the figure below:

2. hbn_isp_release_ae_statistics

【Function Declaration】

int32_t hbn_isp_release_ae_statistics(hbn_vnode_handle_t vnode_fd, isp_statistics_t *ae_statistics)

【Parameter Description】

[IN] hbn_vflow_handle_t vflow_fd: vflow handle;
[IN] isp_statistics_t *ae_statistics: AE statistics;

【Return Value】

Success: HBN_STATUS_SUCCESS 0
Failure: Exception indicated by a negative error code; refer to the return value description.

【Function Description】

Release the obtained AE statistics. This function must be used in pair with hbn_isp_get_ae_statistics.

Parameter Description

Common

hbn_vnode_image_t

Name	Type	Description	Max Value	Min Value	Default	Required
info	hbn_frame_info_t	Image info structure	–	–	–	–
buffer	hb_mem_graphic_buf_t	Image memory info	–	–	–	–
metadata	void *	Metadata	–	–	–	–

hbn_frame_info_t

Name	Type	Description	Max Value	Min Value	Default	Required
frame_id	uint32_t	Frame ID	–	–	–	–
timestamps	uint64_t	System timestamp	–	–	–	–
tv	struct timeval	Hardware timestamp	–	–	–	–
trig_tv	struct timeval	External-triggered hardware timestamp	–	–	–	–
bufferindex	int32_t	Buffer index	–	–	–	–

hb_mem_graphic_buf_t

Name	Type	Description	Max Value	Min Value	Default	Required
fd[MAX_GRAPHIC_BUF_COMP]	int32_t	File descriptor	–	–	–	–
plane_cnt	int32_t	Number of planes	–	–	–	–
format	int32_t	Image format	–	–	–	–
width	int32_t	Width	–	–	–	–
height	int32_t	Height	–	–	–	–
stride	int32_t	Width stride	–	–	–	–
vstride	int32_t	Height stride	–	–	–	–
is_contig	int32_t	Whether buffer physical address is contiguous	–	–	–	–
share_id[MAX_GRAPHIC_BUF_COMP]	int32_t	Share ID	–	–	–	–
flags	int64_t	Flags	–	–	–	–
size[MAX_GRAPHIC_BUF_COMP]	uint64_t	Buffer size	–	–	–	–
virt_addr[MAX_GRAPHIC_BUF_COMP]	uint8_t *	Virtual address	–	–	–	–
phys_addr[MAX_GRAPHIC_BUF_COMP]	uint64_t	Physical address	–	–	–	–
offset[MAX_GRAPHIC_BUF_COMP]	uint64_t	Memory offset	–	–	–	–

hbn_vnode_image_group_t

Name	Type	Description	Max Value	Min Value	Default	Required
info	hbn_frame_info_t	Image info structure	–	–	–	–
buf_group	hb_mem_graphic_buf_group_t	Group image memory info	–	–	–	–
metadata	void *	Metadata	–	–	–	–

hb_mem_graphic_buf_group_t

Name	Type	Description	Max Value	Min Value	Default	Required
graph_group[HB_MEM_MAXIMUM_GRAPH_BUF];	hb_mem_graphic_buf_t	Image memory info	–	–	–	–
group_id	int32_t	Group ID	–	–	–	–
bit_map	uint32_t	Bitmask indicating available layers in graph_group	–	–	–	–

VIN

vin_attr_t

Name	Type	Description	Max Value	Min Value	Default	Required
vin_node_attr	vin_node_attr_t	VIN node attribute structure	–	–	–	Yes
magicNumber	uint32_t	Magic number for attribute structure validation; must be set to MAGIC_NUM	–	–	–	Yes

vin_node_attr_t

Name	Type	Description	Max Value	Min Value	Default	Required
cim_attr	cim_attr_t	CIM parameters	–	–	–	Yes
lpwm_attr	lpwm_attr_t	LPWM parameters	–	–	–	No
vcon_attr	vcon_attr_t	VCON parameters	–	–	–	No
magicNumber	uint32_t	Magic number for attribute structure validation; must be set to MAGIC_NUM	–	–	–	Yes

cim_attr_t

Name	Type	Description	Max Value	Min Value	Default	Required
mipi_en	uint32_t	Enable MIPI input	1	0	–	Yes
mipi_rx	uint32_t	MIPI RX index; options: 0, 1, 4	4	0	–	Yes
vc_index	uint32_t	CIM IPI index	3	0	–	Yes
cim_pym_flyby	uint32_t	Enable CIM PYM hardware bypass	1	0	–	Yes
cim_isp_flyby	uint32_t	Enable CIM ISP hardware bypass	1	0	–	Yes

Name	Type	Description	Max	Min	Default	Required
format	uint32_t	MIPI input image format, e.g., raw12 corresponds to 0x2c	0x27	0x1E	-	Yes
width	uint32_t	MIPI input image width	4096	32	-	Yes
height	uint32_t	MIPI input image height	2160	32	-	Yes

vin_ochn_attr_t

Name	Type	Description	Max	Min	Default	Required
ddr_en	uint32_t	Enable CIM DDR output	1	0	-	No
roi_en	uint32_t	Enable CIM ROI channel output	1	0	-	No
emb_en	uint32_t	Enable CIM EMB channel output	1	0	-	No
rawds_en	uint32_t	Enable RAW scaler	1	0	-	No
pingpong_ring	uint32_t	Enable ping-pong buffer	1	0	-	No
ochn_attr_type	vin_ochn_attr_type_e	Output channel type: VIN_MAIN_FRAME (main data path), VIN_ONLINE (online output path), VIN_EMB (embedded data path), VIN_ROI (ROI data path)	-	-	-	Yes
vin_basic_attr	vin_basic_attr_t	Basic VIN attributes	-	-	-	Yes
rawds_attr	vin_rawds_attr_t	VIN RAW scaler attributes	-	-	-	No
roi_attr	struct vin_roi_attr_s	VIN ROI attributes	-	-	-	No
emb_attr	vin_emb_attr_t	VIN embedded attributes	-	-	-	No
magicNumber	uint32_t	Attribute structure checksum; must be set to the fixed value MAGIC_NUM	-	-	-	Yes

vin_basic_attr_t

Name	Type	Description	Max	Min	Default	Required
pack_mode	uint32_t	Enable packing; packing enabled by default if not configured	1	0	1	No
wstride	uint32_t	Output width stride; if set to 0, calculated internally	1	0	1	No
vstride	uint32_t	Output height stride; if set to 0, calculated internally	1	0	1	No
format	uint32_t	Output image format, e.g., raw12 corresponds to 0x2c	0x27	0x1E	-	Yes

ISP

isp_attr_t

Name	Type	Description	Max	Min	Default	Required
channel	isp_channel_t	ISP channel attributes	-	-	-	Yes
sched_mode	sched_mode_e	ISP scheduling mode: 1 = SCHED_MODE_MANUAL (manual mode), 2 = SCHED_MODE_PASS_THRU (fully online mode)	2	1	-	Yes
work_mode	isp_work_mode_e	ISP working mode: 0 = ISP_WORK_MODE_NORMAL (normal mode), 1 = ISP_WORK_MODE_TPG (ISP outputs test pattern), 2 = ISP_WORK_MODE_CIM_TPG (CIM outputs test pattern)	2	0	-	No
hdr_mode	hdr_mode_e	Enable ISP HDR mode	1	0	-	No
size	image_size_t	ISP processing resolution	-	-	-	No
frame_rate	uint32_t	ISP frame rate	120	1	-	No
isp_combine	isp_combine_t	ISP master-slave mode	-	-	-	No
algo_state	uint32_t	Enable 2A algorithms	1	0	-	No

isp_channel_t

Name	Type	Description	Max	Min	Default	Required
hw_id	uint32_t	ISP hardware ID	1	0	-	Yes
slot_id	uint32_t	ISP internal hardware channel: configure 03 for online input, 411 for offline input	11	0	0	Yes

image_size_t

Name	Type	Description	Max	Min	Default	Required
width	uint32_t	ISP processing width	4096	32	-	Yes
height	uint32_t	ISP processing height	2160	32	-	Yes

isp_ichn_attr_t

Name	Type	Description	Max	Min	Default	Required
input_crop_cfg	crop_cfg_t	Input crop configuration	-	-	-	No
in_buf_noclean	uint32_t	Whether to perform cache clean on input buffer	1	0	-	No
in_buf_noncached	uint32_t	Whether to allocate input buffer as non-cacheable memory	1	0	-	No

crop_cfg_t

Name	Type	Description	Max	Min	Default	Required
rect	image_rect_t	Input crop rectangle	-	-	-	No
enable	HB_BOOL	Enable crop	1	0	-	No

image_rect_t

Name	Type	Description	Max	Min	Default	Required
x	uint32_t	X coordinate	-	-	-	No
y	uint32_t	Y coordinate	-	-	-	No
width	uint32_t	Rect width	-	-	-	No
height	uint32_t	Rect height	-	-	-	No

isp_ochn_attr_t

Name	Type	Description	Max	Min	Default	Required
stream_output_mode	isp_stream_output_mode_e	Whether to enable OTF output: 1 = enable, 0 = disable	1	0	0	Yes
axi_output_mode	isp_axi_output_mode_e	DDR output type: AXI_OUTPUT_MODE_DISABLE = 0,	14	0	0	Yes
		AXI_OUTPUT_MODE_RGB888 = 1,
		AXI_OUTPUT_MODE_RAW8 = 2,
		AXI_OUTPUT_MODE_RAW10 = 3,
		AXI_OUTPUT_MODE_RAW12 = 4,
		AXI_OUTPUT_MODE_RAW16 = 5,
		AXI_OUTPUT_MODE_RAW24 = 6,
		AXI_OUTPUT_MODE_YUV444 = 7,
		AXI_OUTPUT_MODE_YUV422 = 8, /* yuv422 */
		AXI_OUTPUT_MODE_YUV420 = 9, /* yuv420 */
		AXI_OUTPUT_MODE_IR8 = 10,
		AXI_OUTPUT_MODE_YUV420_RAW12 = 11, /* yuv420 & raw12 */
		AXI_OUTPUT_MODE_YUV422_RAW12 = 12, /* yuv422 & raw12 */
		AXI_OUTPUT_MODE_YUV420_RAW16 = 13, /* yuv420 & raw16 */
		AXI_OUTPUT_MODE_YUV422_RAW16 = 14, /* yuv422 & raw16 */
output_crop_cfg	crop_cfg_t	Output crop configuration	-	-	-	Yes
out_buf_noinvalid	uint32_t	Whether to perform cache invalidate on output buffer	1	0	0	No
out_buf_noncached	uint32_t	Whether to allocate output buffer as non-cacheable memory	1	0	0	No
buf_num	uint32_t	Number of output buffers to allocate	16	3	0	Yes

YNR

ynr_init_attr

Name	Type	Description	Max	Min	Default	Required
work_mode	uint32_t	YNR working mode	2	1	-	Yes
		1: Manual mode, online link, previous module is SW trigger;
		2: Single-channel online mode (previous PYM hardware directly connected);
slot_id	uint32_t	YNR hardware channel ID	7	0	-	Yes
width	uint32_t	YNR processing width	2048	32		Yes
height	uint32_t	YNR processing height	2048	32		Yes
nr_static_switch	uint32_t	nr3den <<1 | nr2d_en
in_stride	uint32_t	Y stride and UV stride				Yes
nr2d_en	uint32_t	Enable 2D NR	1	0		Yes
nr3d_en	uint32_t	Enable 3D NR	1	0		Yes
dma_output_en	uint32_t	Enable DMA output	1	0		Yes
		If 3D NR is enabled, DMA output must be enabled
debug_en	uint32_t	Enable debug mode	1	0		No

hobot_ynr_channel_input_config

Name	Type	Description	Max	Min	Default	Required
ch_img_width	uint32_t	YNR input width	4096	32	-	Yes
ch_img_height	uint32_t	YNR input height	2160	32	-	Yes

hobot_ynr_channel_output_config

Name	Type	Description	Max	Min	Default	Required
ch_nr3d_pix_out_dma_byps	uint32_t	DMA output count; recommended to set to 0	4096	32	-	Yes
ch_nr3d_debug_en	uint32_t	Debug switch; recommended to set to 0	1	0	-	Yes

PYM

roi_box_t

Name	Type	Description	Max	Min	Default	Required
start_top	uint32_t	Y-axis position to crop from original image	DS layer: <= region_height BL layer:<= bl_base_height bl_base_height = region_width >> (ds_roi_layer + 1)	DS layer: >= region_height - out_height BL layer: >= bl_base_height - out_height	-	Yes
start_left	uint32_t	X-axis position to crop from original image	DS layer: <= region_width BL layer: <= bl_base_width bl_base_width = region_width >> (ds_roi_layer + 1)	DS layer: >= region_width - out_width BL layer: >= bl_base_width - out_width	-	Yes
region_width	uint32_t	Width of cropped region	-	-	-	Yes
region_height	uint32_t	Height of cropped region	-	-	-	Yes
wstride_uv	uint32_t	Output UV plane stride			-	Yes
wstride_y	uint32_t	Output Y plane stride	-	-	-	Yes
vstride	uint32_t	Height stride (hidden parameter; not recommended to configure)	-	-	out_height	Yes
step_v	uint32_t		-	-	(1 << 16) * (out_height - region_height) / out_height	No
step_h	uint32_t		-	-	(1 << 16) * (out_width - region_width) / out_width	No
out_width	uint32_t	Output image width	-	-	-	Yes
out_height	uint32_t	Output image height	-	-	-	Yes
phase_y_v	uint32_t				0	No
phase_y_h	uint32_t				0	No

chn_ctrl_t

Name	Type	Description	Max	Min	Default	Required
pixel_num_before_sol	uint32_t		-	-	2	Yes
invalid_head_lines	uint32_t		-	-	-	No
src_in_width	uint32_t	Input width, aligned to 2	< 4096	> 32	-	Yes
src_in_height	uint32_t	Input height, aligned to 2	< 4096	> 32	-	Yes
src_in_stride_y	uint32_t	Input Y plane stride, aligned to 16	< 4096	> src_in_width	-	Yes
src_in_stride_uv	uint32_t	Input UV stride, aligned to 16	< 4096	> src_in_width	-	Yes
suffix_hb_val	uint32_t		<= 152	>= 16	100	Yes
prefix_hb_val	uint32_t		<= 2	>= 0	2	Yes
suffix_vb_val	uint32_t		<= 20	>= 0	10	Yes
prefix_vb_val	uint32_t		<= 2	>= 0	0	Yes
bl_max_layer_en	uint8_t	When selecting BL layer, enable number of BL layers		> ds_roi_layer[chn]	5	Yes
ds_roi_en	uint8_t	Enable DS layer output (total 6 layers, enabled per bit)	< (1 << 6)	-	-	Yes
ds_roi_uv_bypass	uint8_t	Enable DS layer UV plane output bypass (enabled per bit)	< (1 << 6)	-	-	No
ds_roi_sel[MAX_DS_NUM]	uint8_t	Layer selection: 0 = SRC layer; 1 = BL layer	< 3	-	-	Yes
ds_roi_layer[MAX_DS_NUM]	uint8_t		When ds_roi_sel = 0, must be 0	-	-	Yes
ds_roi_info[MAX_DS_NUM]	roi_box_t	DS layer configuration	-	-	-	Yes

pym_cfg_t

Name	Type	Description	Max	Min	Default	Required
hw_id	uint8_t	pym hardware module ID (0, 1, 4)	-	-	-	Yes
pym_mode	uint8_t	pym operation mode: 1: Manual mode, online link, with preceding module as SW trigger; 2: Single-channel Online mode (direct hardware connection between preceding module and PYM); 3: Offline mode (input: YUV420SP, output: YUV420SP)	<= 3	>= 1	-	Yes
slot_id	uint8_t	pym hardware channel ID	7	0	-	No
out_buf_noinvalid	uint8_t	Whether the module's output buffer internally performs an "invalidate cache" operation	1	0	1	Yes
out_buf_noncached	uint8_t	Whether non-cacheable memory allocation is enabled for the module's output buffer	1	0	-	No
in_buf_noclean	uint8_t	Whether cache clean is performed on the input buffer	1	0	1	Yes
in_buf_noncached	uint8_t	Whether non-cacheable memory allocation is enabled for the module's input buffer (typically feedback buffer)	1	0	-	No
buf_consecutive	uint8_t	Whether memory is contiguous	1	0	-	No
pingpong_ring	uint8_t	Whether ping-pong buffer is enabled			-	No
output_buf_num	uint32_t	Number of output buffers; when PYM is in offline mode, the number of feedback buffers is allocated by default according to this value	<= 64	0	-	Yes
timeout	uint32_t	Timeout duration	<= 10000		-	No
threshold_time	uint32_t				-	No
layer_num_trans_next	int32_t	Number of layers transferred to the subsequent module	< 6	-	-1	Yes
layer_num_share_prev	int32_t		< 6	-	-1	Yes
chn_ctrl	chn_ctrl_t	Configure input/output format and size
fb_buf_num	uint32_t	Number of feedback buffers	<= 16	-	2	Yes
reserved[6]	uint32_t	Reserved field	-	-	-	No
magicNumber	uint32_t	Attribute structure verification value; must be set to the fixed value MAGIC_NUM	-	-	-	Yes

GDC

gdc_cfg_t

Name	Type	Description	Max	Min	Default	Required
input_width	uint32_t	Input image width, aligned to 2	<= 3840	>= 96	-	Yes
input_height	uint32_t	Input image height, aligned to 2	<= 2160	>= 96	-	Yes
output_width	uint32_t	Output image width	<= 3840	>= 96	-	Yes
output_height	uint32_t	Output image height	<= 2160	>= 96	-	Yes
buf_num	uint32_t	Number of normal input buffers	<= 32	0	6	Yes
fb_buf_num	uint32_t	Number of feedback buffers	<= 32	0	2	Yes
in_buf_noclean	uint32_t	Whether cache clean is performed on the input buffer	1	0	1	No
in_buf_noncached	uint32_t	Whether non-cacheable memory allocation is enabled for the module's input buffer (typically feedback buffer)	-	-	-	No
out_buf_noinvalid	uint32_t	Whether the module's output buffer internally performs an "invalidate cache" operation	-	-	1	No
out_buf_noncached	uint32_t	Whether non-cacheable memory allocation is enabled for the module's output buffer	-	-	-	No
gdc_pipeline	uint32_t		-	-	-	No

STITCH

stitch_base_attr

Name	Type	Description	Max	Min	Default	Required
mode	uint32_t	Operation mode				No
		0 - External buffer feedback mode
		1 - Internal buffer feedback mode
		2 - Flow binding mode
roi_nums	uint32_t	Number of ROI regions	12	1		Yes
img_nums	uint32_t	Number of input images	-	1		Yes
alpha_lut	struct	Alpha lookup table				No
	lut_attr
beta_lut	struct	Beta lookup table				No
	lut_attr
blending	struct	Blending attributes				Yes
	blending_attr

lut_attr

Name	Type	Description	Max	Min	Default	Required
share_id	int32_t	Share ID of hbmem buffer
		Required for storing LUT buffer
		Must be allocated via hbmem
vaddr	uint64_t	LUT virtual address				No
offset	uint64_t	Offset				No
size	uint64_t	Size				No

blending_attr

Name	Type	Description	Max	Min	Default	Required
roi_index	uint32_t	ROI index				Yes
blending_mode	uint32_t	Blending mode:				Yes
		BLENDING_MODE_ONLINE = 0, // online mode
		BLENDING_MODE_ALPHA = 1, // alpha mode
		BLENDING_MODE_ALPH = 2, // alpha beta mode
		BLENDING_MODE_SRC = 3, // src copy mode
		BLENDING_MODE_ALPHA_SRC = 5 // alpha src0 mode
direct	uint32_t	Blending direction:				Yes
		BLENDING_DIRECT_LT = 0, // left and top direction
		BLENDING_DIRECT_RB = 1, // right and bottom direction
		BLENDING_DIRECT_LB = 2, // left and bottom direction
		BLENDING_DIRECT_RT = 3, // right and top direction
uv_en	uint32_t	Whether input image contains UV				Yes
src0_index	uint32_t	Source index corresponding to src0				Yes
src1_index	uint32_t	Source index corresponding to src1				Yes
margin	uint32_t	Optional parameter; may be omitted				No
margin_inv	uint32_t	Optional parameter; may be omitted				No
gain_src0_yuv	uint32_t	Fixed to 256 // 0: Y, 1: U, 2: V				Yes
gain_src1_yuv	uint32_t	Fixed to 256 // 0: Y, 1: U, 2: V				Yes

roi_info

Name	Type	Description	Max	Min	Default	Required
roi_index	uint32_t	ROI index				Yes
roi_x	uint32_t	Starting X coordinate				Yes
roi_y	uint32_t	Starting Y coordinate				Yes
roi_w	uint32_t	Width				Yes
roi_h	uint32_t	Height				Yes

stitch_ch_attr

Name	Type	Description	Max	Min	Default	Required
width	uint32_t	Input or output width				Yes
height	uint32_t	Input or output height				Yes
strid[MAX_STH_FRAME_PLAN]	uint32_t	Stride				Yes
rois[MAX_STH_ROI_NUMS]	struct roi_info	ROI region description				Yes

Channel Binding Description

Module	Output Channel ID	Channel Function
VIN	0	Offline channel, outputs camera frames to DDR
	1	Online channel, connects to ISP or PYM
ISP	0	Offline channel, outputs ISP-processed frames to DDR
	1	Online channel, connects to PYM or YNR
YNR	1	Online channel, connects to PYM
PYM	0	Offline channel, outputs PYM images to DDR
GDC	0	Offline channel, outputs GDC-processed frames to DDR

"Online" indicates direct hardware connection; "Offline" indicates output to DDR buffer.

SLOT_ID and Debug Mode Description

1. The ISP's slot_id parameter selects the ISP hardware context. In CIM-direct-to-ISP scenarios, slot_id can be 0–3; in CIM-DDR-ISP scenarios, slot_id can be 4–11. Different channels must use distinct slot_id values. In ISP-online-YNR-online-PYM or ISP-online-PYM scenarios, the slot_id of YNR and PYM must match that of the ISP.
2. The PYM's sched_mode parameter selects the ISP scheduling mode. In CIM-ISP direct hardware connection scenarios, select mode 2 (passthrough); in other scenarios, select mode 1 (manual). In ISP-online-YNR-online-PYM or ISP-online-PYM scenarios, the YNR work_mode and PYM pym_mode must be consistent with the ISP sched_mode.

Return Value Description

Error Code	Macro Definition	Description
0	HBN_STATUS_SUCESS	Success
1	HBN_STATUS_INVALID_NODE	Invalid vnode; corresponding vnode not found
2	HBN_STATUS_INVALID_NODETYPE	Invalid vnode type; corresponding vnode not found
3	HBN_STATUS_INVALID_HWID	Invalid hardware module ID
4	HBN_STATUS_INVALID_CTXID	Invalid context ID
5	HBN_STATUS_INVALID_OCHNID	Invalid output channel ID
6	HBN_STATUS_INVALID_ICHNID	Invalid input channel ID
7	HBN_STATUS_INVALID_FORMAT	Invalid format
8	HBN_STATUS_INVALID_NULL_PTR	Null pointer
9	HBN_STATUS_INVALID_PARAMETER	Invalid parameter; version check failed
10	HBN_STATUS_ILLEGAL_ATTR	Invalid parameter
11	HBN_STATUS_INVALID_FLOW	Invalid flow; corresponding flow not found
12	HBN_STATUS_FLOW_EXIST	Flow already exists
13	HBN_STATUS_FLOW_UNEXIST	Flow does not exist
14	HBN_STATUS_NODE_EXIST	Node already exists
15	HBN_STATUS_NODE_UNEXIST	Node does not exist
16	HBN_STATUS_NOT_CONFIG	Reserved
17	HBN_STATUS_CHN_NOT_ENABLED	Channel not enabled
18	HBN_STATUS_CHN_ALREADY_ENABLED	Channel already enabled
19	HBN_STATUS_ALREADY_BINDED	Node already bound
20	HBN_STATUS_NOT_BINDED	Node not bound
21	HBN_STATUS_TIMEOUT	Timeout
22	HBN_STATUS_NOT_INITIALIZED	Not initialized
23	HBN_STATUS_NOT_SUPPORT	Channel not supported or not activated
24	HBN_STATUS_NOT_PERM	Operation not permitted
25	HBN_STATUS_NOMEM	Insufficient memory
26	HBN_STATUS_INVALID_VNODE_FD	Invalid node file descriptor
27	HBN_STATUS_INVALID_ICHNID_FD	Invalid input channel file descriptor
28	HBN_STATUS_INVALID_OCHNID_FD	Invalid output channel file descriptor
29	HBN_STATUS_OPEN_OCHN_FAIL	Failed to open output channel
30	HBN_STATUS_OPEN_ICHN_FAIL	Failed to open input channel
31	HBN_STATUS_JSON_PARSE_FAIL	JSON parsing failed
32	HBN_STATUS_REQ_BUF_FAIL	Buffer request failed
33	HBN_STATUS_QUERY_BUF_FAIL	Buffer query failed
34	HBN_STATUS_SET_CONTROL_FAIL	Failed to set module control or tuning parameters (e.g., ISP effect parameters)
35	HBN_STATUS_GET_CONTROL_FAIL	Failed to get module control or tuning parameters (e.g., ISP effect parameters)
36	HBN_STATUS_NODE_START_FAIL	Node start failed
37	HBN_STATUS_NODE_STOP_FAIL	Node stop failed
38	HBN_STATUS_NODE_POLL_ERROR	Node channel poll error
39	HBN_STATUS_NODE_POLL_TIMEOUT	Node channel poll timeout
40	HBN_STATUS_NODE_POLL_FRAME_DROP	Frame drop occurred during node channel poll
41	HBN_STATUS_NODE_POLL_HUP	Node channel poll descriptor hang-up
42	HBN_STATUS_NODE_ILLEGAL_EVENT	Illegal event during node channel poll
43	HBN_STATUS_NODE_DEQUE_ERROR	Node channel dequeue buffer error
44	HBN_STATUS_ILLEGAL_BUF_INDEX	Invalid buffer index
45	HBN_STATUS_NODE_QUE_ERROR	Node channel queue buffer error
46	HBN_STATUS_FLUSH_FRAME_ERROR	Node channel frame flush error
47	HBN_STATUS_INIT_BIND_ERROR	Error occurred during JSON parsing and binding
48	HBN_STATUS_ADD_NODE_FAIL	Failed to add node to flow
49	HBN_STATUS_WRONG_CONFIG_ID	Unsupported node ID by the system
50	HBN_STATUS_BIND_NODE_FAIL	Error occurred during flow binding to node
51	HBN_STATUS_INVALID_VERSION	Version mismatch between lower-level driver module and upper-layer library
52	HBN_STATUS_GET_VERSION_ERROR	Error retrieving lower-level driver module version
53	HBN_STATUS_MEM_INIT_FAIL	hbmem memory initialization failed
54	HBN_STATUS_MEM_IMPORT_FAIL	hbmem memory import failed
55	HBN_STATUS_MEM_FREE_FAIL	hbmem memory release failed
56	HBN_STATUS_SYSFS_OPEN_FAIL	System file open failed
57	HBN_STATUS_STRUCT_SIZE_NOT_MATCH	HAL-layer structure size mismatch with kernel layer
58	HBN_STATUS_RGN_UNEXIST	Corresponding RGN data not found
59	HBN_STATUS_RGN_INVALID_OPERATION	Invalid RGN operation
60	HBN_STATUS_RGN_OPEN_FILE_FAIL	RGN module file open failed
128	HBN_STATUS_ERR_UNKNOW	Unknown error

V4L2

Some modules of the S100 Camsys have already been integrated with V4L2, allowing acquisition of Camsys data streams through standard V4L2 programming and open-source tools.

Usage

After system boot-up, camsys runs in hbn mode by default. You can switch to V4L2 mode by loading the camsys V4L2 kernel module (ko).

Switching to V4L2 mode:

  # Unload hbn drivers
  rmmod hobot_isp
  rmmod hobot_cim
  rmmod hobot_mipidbg
  rmmod hobot_mipicsi
  rmmod hobot_pym_jplus
  rmmod hobot_gdc
  rmmod hobot_ynr

  # Load V4L2 drivers
  modprobe videobuf2-common
  modprobe videobuf2-v4l2
  modprobe videobuf2-memops
  modprobe videobuf2-common
  modprobe videobuf2-dma-contig
  modprobe videobuf2-v4l2
  modprobe v4l2-mem2mem
  modprobe imx219
  modprobe v4l_mipicsi
  modprobe v4l2_cim
  modprobe hobot_isp_v4l2
  modprobe pym_v4l_drv
  modprobe gdc_v4l_drv
  modprobe hobot_ynr_v4l2
  modprobe vid_v4l2 scene=[scene num] # See table below for scene num
  or modprobe vid_v4l2 scene_table="xxx"
  nohup isp_service &

Scene construction methods:

1. For existing scenes, directly specify using scene num:

modprobe vid_v4l2 scene=[scene num]

2. For custom scenes, construct using a scene table:

modprobe vid_v4l2 scene_table="{<pre_module><hw_id>-<ctx_id>,<pad>,<next_module><hw_id>-<ctx_id>,<pad>,1}{...}..."
# Parameter explanation:
# Each {} represents a connection between two modules.
# pre_module and next_module specify the connected modules; valid values include cim, isp, ynr, pym, gdc, video, video-m2c.
# hw_id specifies the hardware ID.
# ctx_id specifies the hardware context ID.
# pad is the pad number, usually 0. For example, pym supports multi-channel output and can be configured from 0 to 5.
# The final next_module in a pipeline must be specified as either video or video-m2m.
# Note: The scene string passed via scene_table must NOT contain spaces.
# Example: scene_table="{cim0-0,0,isp0-0,0,1}{isp0-0,0,video,0,1}" constructs a cim0-otf-isp0-ddr scene.
# Example for scene 9 below: scene_table="{cim0-0,0,isp1-4,0,1}{cim1-0,0,isp1-5,0,1}{isp1-4,0,ynr1-4,0,1}{isp1-5,0,ynr1-5,0,1}{ynr1-4,0,pym1-0,0,1}{ynr1-5,0,pym1-1,0,1}{pym1-0,0,video,0,1}{pym1-1,0,video,0,1}"

Scene switching method:

rmmod vid_v4l2
modprobe vid_v4l2  xxx=xxxx### Scene Description

Scene Description

scene num	Scene Summary	Scene Description	Corresponding Video Nodes (Relative)
0	CIM-DDR Output	CIM0 outputs 1 stream to DDR (video0)	video0
		CIM1 outputs 1 stream to DDR	video1
		CIM4 outputs 4 streams to DDR (SerDes scenario)	video2~5
1	CIM-OTF-ISP-DDR	CIM0-OTF-ISP0-DDR	video0
		CIM1-OTF-ISP1-DDR	video1
2	CIM-OTF-ISP-OTF-PYM-DDR (2 streams)	CIM0-OTF-ISP0-OTF-PYM0,	video0 corresponds to first PYM ds0
		PYM outputs one channel
		CIM1-OTF-ISP1-OTF-PYM1,	video1 corresponds to second PYM ds0
		PYM outputs one channel
3	CIM-OTF-ISP-OTF-PYM-DDR	CIM0-OTF-ISP0-OTF-PYM0,	video0video5 correspond to ds05
	2 streams, 6 channels output	PYM outputs 6 channels
		CIM1-OTF-ISP1-OTF-PYM1,	video6video11 correspond to ds0ds5
		PYM outputs 6 channels
4	CIM-DDR-ISP-DDR	CIM0-DDR-ISP0-DDR	video0
		CIM1-DDR-ISP1-DDR	video1
5	CIM-DDR-ISP-OTF-PYM	CIM0-DDR-ISP0-OTF-PYM0	video0
		Outputs one channel
6	CIM-OTF-ISP-DDR-GDC	CIM0-OTF-ISP-DDR-GDC	video0
		Outputs one channel
7	DDR-PYM-DDR Loopback Output	Loopback PYM outputs 6 streams to DDR	video0 ~ 5
		Loopback PYM outputs 6 streams to DDR	video6 ~ 11
	DDR-GDC-DDR Loopback Output	Loopback GDC outputs to DDR	video12
		Loopback GDC outputs to DDR	video13
9	CIM-DDR-ISP-OTF-YNR-PYM	CIM0-DDR-ISP1-OTF-YNR1-OTF-PYM1	video0
		CIM1-DDR-ISP1-OTF-YNR1-OTF-PYM1	video1

(Other link scenarios are currently unsupported and will be continuously updated.)

camsys sample

imx219 + MIPI + CIM + ISP + PYM:

         // Sample configuration for imx219
static mipi_config_t imx219_mipi_config = {
    .rx_enable = 1,
    .rx_attr = {
        .phy = 0,
        .lane = 2,
        .datatype = 0x12b,
        .fps = 30,
        .mclk = 24,
        .mipiclk = 1728,
        .width = 0,
        .height = 0,
        .linelenth = 0,
        .framelenth = 0,
        .settle = 0,
        .channel_num = 0,
        .channel_sel = {0},
    },

    .rx_ex_mask = 0x40,
    .rx_attr_ex = {
        .stop_check_instart = 1,
    },

    .end_flag = MIPI_CONFIG_END_FLAG,
};

static camera_config_t imx219_camera_config = {
        /* 0 */
        .name = "imx219",
        .addr = 0x10,
        .eeprom_addr = 0x51,
        .serial_addr = 0x40,
        .sensor_mode = 1,
        .fps = 30,
        .width = 1920,
        .height = 1080,
        .extra_mode = 0,
        .config_index = 0,
        .mipi_cfg = &imx219_mipi_config, // MIPI configuration; NULL means auto-detection
        .end_flag = CAMERA_CONFIG_END_FLAG,
        .calib_lname = "disable",
};

static isp_cfg_t imx219_isp_config = {
    .isp_attr = {
        .channel = {
            .hw_id = 0,
            .slot_id = 4,
            .ctx_id = -1, //#define AUTO_ALLOC_ID -1
        },
        .work_mode = 0,
        .hdr_mode = 1,
        .size = {
            .width = 1920,
            .height = 1080,
        },
        .frame_rate = 30,
        .sched_mode = 1,
        .algo_state = 1,
        .isp_combine = {
            .isp_channel_mode = 0, //ISP_CHANNEL_MODE_NORMAL
            .bind_channel = {
                .bind_hw_id = 0,
                .bind_slot_id = 0,
            },
        },
        .clear_record = 0, // Not obtained from JSON or code; set to 0
        .isp_sw_ctrl = {
            .ae_stat_buf_en = 1,
            .awb_stat_buf_en = 1,
            .ae5bin_stat_buf_en = 1,
            .ctx_buf_en = 0,
            .pixel_consistency_en = 0,
        },
    },
    .ichn_attr = {
        .input_crop_cfg = {
            .enable = 0,
            .rect = {
                .x = 0,
                .y = 0,
                .width = 0,
                .height = 0,
            },
        },
        .in_buf_noclean = 1,
        .in_buf_noncached = 0,
    },
    .ochn_attr = {
        .output_crop_cfg = {
            .enable = 0,
            .rect = {
                .x = 0,
                .y = 0,
                .width = 0,
                .height = 0,
            },
        },
    .out_buf_noinvalid = 1,
        .out_buf_noncached = 0,
        .output_raw_level = 0, //ISP_OUTPUT_RAW_LEVEL_SENSOR_DATA
        .stream_output_mode = 0, //convert_isp_stream_output(1),
        .axi_output_mode = 9, //convert_isp_axi_output(0),
        .buf_num = 3,
    }
};

static vin_attr_t imx219_vin_attr = {
    .vin_node_attr = {
        .vcon_attr = {
            .bus_main = 2,
            .bus_second = 2,
        },

        .cim_attr = {
            .mipi_en = 1,
            .cim_isp_flyby = 0,
            .cim_pym_flyby = 0,
            .mipi_rx = 0,
            .vc_index = 0,
            .ipi_channels = 1,
            .y_uv_swap = 0, //(uint32_t)vpf_get_json_value(p_node_mipi, "y_uv_swap");
            .func = {
                .enable_frame_id = 1,
                .set_init_frame_id = 1,
                .enable_pattern = 0,
            },
            .rdma_input = {
                .rdma_en = 0,
                .stride = 0,
                .pack_mode = 1,
                .buff_num = 6,
            },
        },
    },

    .vin_ichn_attr = {
        .width =  1920,
        .height = 1080,
        .format = 43,
    },

    .vin_attr_ex = {
        .cim_static_attr = {
            .water_level_mark = 0,
        },
    },

    .vin_ochn_attr = {
        [VIN_MAIN_FRAME] = { //vin_ochn0_attr
            .ddr_en = 1,
            .vin_basic_attr = {
                .format = 43,
                .wstride = 0,
                .pack_mode = 1,
            },
            .pingpong_ring = 1,
            .roi_en = 0,
            .roi_attr = {
                .roi_x = 1280,
                .roi_y = 720,
                .roi_width = 64,
                .roi_height = 64,
            },
            .rawds_en = 0,
            .rawds_attr = {
                .rawds_mode = 0,
            },
        },
    },
    .vin_ochn_buff_attr = {
        [VIN_MAIN_FRAME] = { //vin_ochn0_buff_attr
            .buffers_num = 6,
        },
        [VIN_EMB] = { //vin_ochn3_buff_attr
            .buffers_num = 6,
        },
        [VIN_ROI] = { //vin_ochn4_buff_attr
            .buffers_num = 6,
        },
    },
    .magicNumber = MAGIC_NUMBER,
};

pym_cfg_t pym_common_config = {
        .hw_id = 1,
        .pym_mode = 3,
        .slot_id = 0,
        .pingpong_ring = 0,
        .output_buf_num = 6,
        .fb_buf_num = 2,
        .timeout = 0,
        .threshold_time = 0,
        .layer_num_trans_next = 0,
        .layer_num_share_prev = -1,
        .out_buf_noinvalid = 1,
        .out_buf_noncached = 0,
        .in_buf_noclean = 1,
        .in_buf_noncached = 0,
        .chn_ctrl = {
            .pixel_num_before_sol = DEF_PIX_NUM_BF_SOL,
            .invalid_head_lines = 0,
            .src_in_width = 1920,
            .src_in_height = 1080,
            .src_in_stride_y = 1920,
            .src_in_stride_uv = 1920,
            .suffix_hb_val = DEF_SUFFIX_HB,
            .prefix_hb_val = DEF_PREFIX_HB,
            .suffix_vb_val = DEF_SUFFIX_VB,
            .prefix_vb_val = DEF_PREFIX_VB,
            .ds_roi_en = 1,
            .bl_max_layer_en = DEF_BL_MAX_EN,
            .ds_roi_uv_bypass = 0,
            .ds_roi_sel = {
                [0] = 0,
            },
            .ds_roi_layer = {
                [0] = 0,
            },
            .ds_roi_info = {
                [0] = {
                    .start_left = 0,
                    .start_top = 0,
                    .region_width = 1920,
                    .region_height = 1080,
                    .wstride_uv = 1920,
                    .wstride_y = 1920,
                    .out_width = 1920,
                    .out_height = 1080,
                    .vstride = 1080, //.out_height,
                },
            },
        },
    .magicNumber = MAGIC_NUMBER,
};

         // imx219 initialization
hbn_camera_create(camera_config, &cam_fd);

// cim initialization
hbn_vnode_open(HB_VIN, hw_id, AUTO_ALLOC_ID, &vin_node_handle);
hbn_vnode_set_attr(vin_node_handle, vin_attr);
hbn_vnode_set_ichn_attr(vin_node_handle, 0, vin_ichn_attr);
hbn_vnode_set_ochn_attr(vin_node_handle, (uint32_t)VIN_MAIN_FRAME, vin_ochn_attr);
if (vin_ochn_attr->ddr_en) {
    memset(&alloc_attr, 0, sizeof(hbn_buf_alloc_attr_t));
    alloc_attr.buffers_num = vin_attr->vin_ochn_buff_attr[VIN_MAIN_FRAME].buffers_num;
    alloc_attr.is_contig = 1;
    alloc_attr.flags = (int64_t)((uint64_t)HB_MEM_USAGE_CPU_READ_OFTEN |         (uint64_t)HB_MEM_USAGE_CPU_WRITE_OFTEN | (uint64_t)HB_MEM_USAGE_CACHED);
    hbn_vnode_set_ochn_buf_attr(vin_node_handle, (uint32_t)VIN_MAIN_FRAME, &alloc_attr);
}

// isp initialization
hbn_vnode_open(HB_ISP, hw_id, ctx_id, &isp_node_handle);
hbn_vnode_set_attr(isp_node_handle, &isp_config->isp_attr);
hbn_vnode_set_ichn_attr(isp_node_handle, 0, &isp_config->ichn_attr);
hbn_vnode_set_ochn_attr(isp_node_handled, 0, &isp_config->ochn_attr);

  
// pym initialization
hbn_vnode_open(HB_PYM, pym_cfg->hw_id, AUTO_ALLOC_ID, &pym_node_handle);
hbn_vnode_set_attr(pym_node_handle, pym_cfg);
hbn_vnode_set_ichn_attr(pym_node_handle, 0, pym_cfg);
hbn_vnode_set_ochn_attr(pym_node_handle, 0, pym_cfg);
if (pym_cfg->output_buf_num > 0u) {
    memset(&alloc_attr, 0, sizeof(hbn_buf_alloc_attr_t));
    alloc_attr.buffers_num = pym_cfg->output_buf_num;
    alloc_attr.is_contig = 1;
    alloc_attr.flags = (int64_t)((uint64_t)HB_MEM_USAGE_CPU_READ_OFTEN | (uint64_t)HB_MEM_USAGE_CPU_WRITE_OFTEN);
    if (pym_cfg->out_buf_noncached == 0u) {
        alloc_attr.flags |= (uint64_t)HB_MEM_USAGE_CACHED;
    }
        ret = hbn_vnode_set_ochn_buf_attr(pym_node_handle, 0, &alloc_attr);
}

// vflow initialization
hbn_vflow_create(&vflow_fd);
hbn_vflow_add_vnode(vflow_fd, vin_node_handle);
hbn_vflow_add_vnode(vflow_fd, isp_node_handle);
hbn_vflow_add_vnode(vflow_fd, pym_node_handle);
hbn_camera_attach_to_vin(cam_fd, vin_node_handle);
hbn_vflow_bind_vnode(vflow_fd, vin_node_handle, 0, isp_node_handle, 0);
hbn_vflow_bind_vnode(vflow_fd, isp_node_handle, 0, pym_node_handle, 0);
hbn_vflow_start(vflow_fd);

// Get image from pym and return buffer
hbn_vnode_getframe_group(pym_node_handle, 0, VP_GET_FRAME_TIMEOUT, out_image_group);
fill_image_frame_from_vnode_image_group(frame, ochn_id);
memcpy(frame_buffer, frame.data[0], frame.data_size[0]); // frame_buffer is the obtained complete image
if (frame.plane_count > 1)
    memcpy(frame_buffer + frame.data_size[0], frame.data[1], frame.data_size[1]);
hbn_vnode_releaseframe_group(pym_node_handle, 0, out_image_group);                                                                                                                                                                                          |

0820c + 96712 deserializer + MIPI + CIM + PYM:

// Sample configuration for 0820c
static mipi_config_t ar0820std_mipi_config = {.rx_enable = 1,
    .rx_attr = {
        .phy = 0,
        .lane = 1,
        .datatype = 30,
        .fps = 30,
        .mclk = 24,
        .mipiclk = 810,
        .width = 3840,
        .height = 2160,
        .linelenth = 2149,
        .framelenth = 1125 * 2,
        .settle = 22,
        .channel_num = 1,
        .channel_sel = {0},
    },
};

static camera_config_t ar0820std_camera_config = {
        /* 0 */
        .name = "ar0820std",
        .addr = 0x10,
        .eeprom_addr = 0x51,
        .serial_addr = 0x40,
        .sensor_mode = 0x5,
        .fps = 30,
        .width = 3840,
        .height = 2160,
        .extra_mode = 5,
        .config_index = 512,
        .end_flag = CAMERA_CONFIG_END_FLAG,
        .calib_lname = "disable",
};

static poc_config_t g_poc_cfg[] = {
    {
        .addr = 0x28,
        .poc_map = 0x2013,
        .end_flag = POC_CONFIG_END_FLAG,
    },
};

static deserial_config_t ar0820std_deserial_config = {
    .name = "max96712",
    .addr = 0x29,
    .poc_cfg = &g_poc_cfg[0],
    .end_flag = DESERIAL_CONFIG_END_FLAG,
};

static vin_attr_t ar0820std_vin_attr = {
    .vin_node_attr = {
        .cim_attr = {
            .cim_isp_flyby = 0,
            .cim_pym_flyby = 0,
            .mipi_en = 1,
            .mipi_rx = 4,
            .vc_index = 0,
            .ipi_channels = 1,
            .y_uv_swap = 0, //(uint32_t)vpf_get_json_value(p_node_mipi, "y_uv_swap");
            .func = {
                .enable_frame_id = 1,
                .set_init_frame_id = 1,
                .enable_pattern = 0,
                .skip_frame = 0,
                .input_fps = 0,
                .output_fps = 0,
                .skip_nums = 0,
                .hw_extract_m = 0,
                .hw_extract_n = 0,
                .lpwm_trig_sel = (int32_t)LPWM_CHN_INVALID,
            },
            .rdma_input = {
                .rdma_en = 0,
                .stride = 0,
                .pack_mode = 1,
                .buff_num = 6,
            },
        },
    },

    .vin_ichn_attr = {
        .width =  3840,
        .height = 2160,
        .format = 30,
    },

    .vin_attr_ex = {
        .cim_static_attr = {
            .water_level_mark = 0,
        },
    },

    .vin_ochn_attr = {
        [VIN_MAIN_FRAME] = { //vin_ochn0_attr
            .ddr_en = 1,
            .vin_basic_attr = {
                .format = 30,
                .wstride = 0,
                .vstride = 0,
                .pack_mode = 1,
            },
            .pingpong_ring = 1,
            .roi_en = 0,
            .roi_attr = {
                .roi_x = 1280,
                .roi_y = 720,
                .roi_width = 64,
                .roi_height = 64,
            },
            .rawds_en = 0,
            .rawds_attr = {
                .rawds_mode = 0,
            },
        },
    },

    .vin_ochn_buff_attr = {
        [VIN_MAIN_FRAME] = { //vin_ochn0_buff_attr
            .buffers_num = 6,
        },
        [VIN_EMB] = { //vin_ochn3_buff_attr
            .buffers_num = 6,
        },
        [VIN_ROI] = { //vin_ochn4_buff_attr
            .buffers_num = 6,
        },
    },
    .magicNumber = MAGIC_NUMBER,
};

pym_cfg_t pym_common_config = {
        .hw_id = 1,
        .pym_mode = 3,
        .slot_id = 0,
        .pingpong_ring = 0,
        .output_buf_num = 6,
        .fb_buf_num = 2,
        .timeout = 0,
        .threshold_time = 0,
        .layer_num_trans_next = 0,
        .layer_num_share_prev = -1,
        .out_buf_noinvalid = 1,
        .out_buf_noncached = 0,
        .in_buf_noclean = 1,
        .in_buf_noncached = 0,
        .chn_ctrl = {
            .pixel_num_before_sol = DEF_PIX_NUM_BF_SOL,
            .invalid_head_lines = 0,
            .src_in_width = 1920,
            .src_in_height = 1080,
            .src_in_stride_y = 1920,
            .src_in_stride_uv = 1920,
            .suffix_hb_val = DEF_SUFFIX_HB,
            .prefix_hb_val = DEF_PREFIX_HB,
            .suffix_vb_val = DEF_SUFFIX_VB,
            .prefix_vb_val = DEF_PREFIX_VB,
            .ds_roi_en = 1,
            .bl_max_layer_en = DEF_BL_MAX_EN,
            .ds_roi_uv_bypass = 0,
            .ds_roi_sel = {
                [0] = 0,
            },
            .ds_roi_layer = {
                [0] = 0,
            },
            .ds_roi_info = {
                [0] = {
                    .start_left = 0,
                    .start_top = 0,
                    .region_width = 1920,
                    .region_height = 1080,
                    .wstride_uv = 1920,
                    .wstride_y = 1920,
                    .out_width = 1920,
                    .out_height = 1080,
                    .vstride = 1080, //.out_height,
                },
            },
        },
    .magicNumber = MAGIC_NUMBER,
};

 // AR0820C initialization
hbn_camera_create(camera_config, &cam_fd);

// MAX96712 deserializer initialization
hbn_deserial_create(deserial_config, &des_fd);

// CIM initialization
hbn_vnode_open(HB_VIN, hw_id, AUTO_ALLOC_ID, &vin_node_handle);
hbn_vnode_set_attr(vin_node_handle, vin_attr);
hbn_vnode_set_ichn_attr(vin_node_handle, 0, vin_ichn_attr);
hbn_vnode_set_ochn_attr(vin_node_handle, (uint32_t)VIN_MAIN_FRAME, vin_ochn_attr);
if (vin_ochn_attr->ddr_en) {
    memset(&alloc_attr, 0, sizeof(hbn_buf_alloc_attr_t));
    alloc_attr.buffers_num = vin_attr->vin_ochn_buff_attr[VIN_MAIN_FRAME].buffers_num;
    alloc_attr.is_contig = 1;
    alloc_attr.flags = (int64_t)((uint64_t)HB_MEM_USAGE_CPU_READ_OFTEN |         (uint64_t)HB_MEM_USAGE_CPU_WRITE_OFTEN | (uint64_t)HB_MEM_USAGE_CACHED);
    hbn_vnode_set_ochn_buf_attr(vin_node_handle, (uint32_t)VIN_MAIN_FRAME, &alloc_attr);
}// pym initialization
hbn_vnode_open(HB_PYM, pym_cfg->hw_id, AUTO_ALLOC_ID, &pym_node_handle);
hbn_vnode_set_attr(pym_node_handle, pym_cfg);
hbn_vnode_set_ichn_attr(pym_node_handle, 0, pym_cfg);
hbn_vnode_set_ochn_attr(pym_node_handle, 0, pym_cfg);
if (pym_cfg->output_buf_num > 0u) {
    memset(&alloc_attr, 0, sizeof(hbn_buf_alloc_attr_t));
    alloc_attr.buffers_num = pym_cfg->output_buf_num;
    alloc_attr.is_contig = 1;
    alloc_attr.flags = (int64_t)((uint64_t)HB_MEM_USAGE_CPU_READ_OFTEN | (uint64_t)HB_MEM_USAGE_CPU_WRITE_OFTEN);
    if (pym_cfg->out_buf_noncached == 0u) {
        alloc_attr.flags |= (uint64_t)HB_MEM_USAGE_CACHED;
    }
        ret = hbn_vnode_set_ochn_buf_attr(pym_node_handle, 0, &alloc_attr);
}

// vflow initialization
hbn_vflow_create(&vflow_fd);
hbn_vflow_add_vnode(vflow_fd, vin_node_handle);
hbn_vflow_add_vnode(vflow_fd, pym_node_handle);
hbn_camera_attach_to_deserial(cam_fd, des_fd, 0);
hbn_deserial_attach_to_vin(des_fd, 0, vin_node_handle);
hbn_vflow_bind_vnode(vflow_fd, vin_node_handle, 0, pym_node_handle, 0);
hbn_vflow_start(vp_vflow_contex->vflow_fd);

// Get image from pym and return the buffer
hbn_vnode_getframe_group(pym_node_handle, 0, VP_GET_FRAME_TIMEOUT, out_image_group);
fill_image_frame_from_vnode_image_group(frame, ochn_id);
memcpy(frame_buffer, frame.data[0], frame.data_size[0]); // frame_buffer is the obtained complete image
if (frame.plane_count > 1)
    memcpy(frame_buffer + frame.data_size[0], frame.data[1], frame.data_size[1]);
hbn_vnode_releaseframe_group(pym_node_handle, 0, out_image_group);

GDC STITCH Stitching Sample

This sample uses a data-replay workflow: it reads files from system storage as input images for GDC, calls hbn APIs, performs GDC processing based on the GDC configuration binary file, and then stitches the GDC output images using the stitch API along with the corresponding stitching LUT table file to generate a bird's-eye-view image.

Original rear-view image and its GDC-processed output:

Original front-view image and its GDC-processed output:

Original left-view image and its GDC-processed output:

Original right-view image and its GDC-processed output:

Final stitched output image:

Corresponding ROI region division for stitching:

ROI	Range	SRC0	Start Point	Size	SRC1	Start Point	Size	Destination Start	Mode	Direction
0	Left-view frame2	frame 2	(10, 0)	(390, 778)				(0, 16)	3 Direct copy
1	Right-view frame3	frame 3	(10, 0)	(390, 780)				(506, 14)	3 Direct copy
2	Rear-view frame0	frame 0	(0, 0)	(896, 298)				(0, 598)	3 Direct copy
3	Front-view frame1	frame 1	(4, 0)	(892, 298)				(0, 0)	3 Direct copy
4	Overlap between left and front views	frame 2	(10, 0)	(390, 282)	frame 1	(2, 16)	(390, 282)	(0, 16)	1 Alpha blend	0 Top-left
5	Overlap between right and front views	frame 3	(10, 0)	(388, 284)	frame 1	(508, 14)	(388, 284)	(506, 14)	1 Alpha blend	3 Top-right
6	Overlap between left and rear views	frame 2	(10, 582)	(390, 196)	frame 0	(0, 0)	(390, 196)	(0, 598)	1 Alpha blend	2 Bottom-left
7	Overlap between right and rear views	frame 3	(10, 584)	(390, 196)	frame 0	(506, 0)	(390, 196)	(506, 598)	1 Alpha blend	1 Bottom-right

STITCH configuration parameters:

struct stitch_ch_attr inch_attr[4] = {
        {
                .width = 896,
                .height = 298,
                .strid = {896, 896},
                .rois = {
                        { .roi_index = 0, .roi_x = 0, .roi_y = 0, .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 0, .roi_x = 0, .roi_y = 0, .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 2, .roi_x = 0, .roi_y = 0, .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 0, .roi_x = 0, .roi_y = 0, .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 0, .roi_x = 0, .roi_y = 0, .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 0, .roi_x = 0, .roi_y = 0, .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 6, .roi_x = 0, .roi_y = 0, .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 7, .roi_x = 506, .roi_y = 0, .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 0, .roi_x = 0, .roi_y = 0, .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 0, .roi_x = 0, .roi_y = 0, .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 0, .roi_x = 0, .roi_y = 0, .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 0, .roi_x = 0, .roi_y = 0, .roi_w = 0, .roi_h = 0  },

                }
        },
        {
                .width = 896,
                .height = 298,
                .strid = {896, 896},
                .rois = {
                        { .roi_index = 0, .roi_x = 0, .roi_y = 0, .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 0, .roi_x = 0, .roi_y = 0, .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 0, .roi_x = 0, .roi_y = 0, .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 3, .roi_x = 4, .roi_y = 0, .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 4, .roi_x = 2, .roi_y = 16, .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 5, .roi_x = 508, .roi_y = 14, .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 0, .roi_x = 0, .roi_y = 0, .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 0, .roi_x = 0, .roi_y = 0, .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 0, .roi_x = 0, .roi_y = 0, .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 0, .roi_x = 0, .roi_y = 0, .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 0, .roi_x = 0, .roi_y = 0, .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 0, .roi_x = 0, .roi_y = 0, .roi_w = 0, .roi_h = 0  },

                }
        },
        {
                .width = 400,
                .height = 778,
                .strid = {400, 400},
                .rois = {
                        { .roi_index = 0, .roi_x = 10, .roi_y = 0, .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 0, .roi_x = 0, .roi_y = 0, .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 0, .roi_x = 0, .roi_y = 0, .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 0, .roi_x = 0, .roi_y = 0, .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 4, .roi_x = 10, .roi_y = 0, .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 0, .roi_x = 0, .roi_y = 0, .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 6, .roi_x = 10, .roi_y = 582, .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 0, .roi_x = 0, .roi_y = 0, .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 0, .roi_x = 0, .roi_y = 0, .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 0, .roi_x = 0, .roi_y = 0, .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 0, .roi_x = 0, .roi_y = 0, .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 0, .roi_x = 0, .roi_y = 0, .roi_w = 0, .roi_h = 0  },

                }

        },
        {
                .width = 400,
                .height = 780,
                .strid = {400, 400},
                .rois = {
                        { .roi_index = 0, .roi_x = 0,  .roi_y = 0,   .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 1, .roi_x = 10, .roi_y = 0,   .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 0, .roi_x = 0,  .roi_y = 0,   .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 0, .roi_x = 0,  .roi_y = 0,   .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 0, .roi_x = 0,  .roi_y = 0,   .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 5, .roi_x = 10, .roi_y = 0,   .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 0, .roi_x = 0,  .roi_y = 0,   .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 7, .roi_x = 10, .roi_y = 584, .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 0, .roi_x = 0,  .roi_y = 0,   .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 0, .roi_x = 0,  .roi_y = 0,   .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 0, .roi_x = 0,  .roi_y = 0,   .roi_w = 0, .roi_h = 0  },
                        { .roi_index = 0, .roi_x = 0,  .roi_y = 0,   .roi_w = 0, .roi_h = 0  },

                }
        }
};

struct stitch_ch_attr och_attr = {
        .width = 896,
        .height = 896,
        .strid = {896, 896},
        .rois = {
                { .roi_index = 0, .roi_x =   0, .roi_y =  16, .roi_w = 390, .roi_h = 778  },
                { .roi_index = 1, .roi_x = 506, .roi_y =  14, .roi_w = 390, .roi_h = 780  },
                { .roi_index = 2, .roi_x =   0, .roi_y = 598, .roi_w = 896, .roi_h = 298  },
                { .roi_index = 3, .roi_x =   0, .roi_y =   0, .roi_w = 892, .roi_h = 298  },
                { .roi_index = 4, .roi_x =   0, .roi_y =  16, .roi_w = 390, .roi_h = 282  },
                { .roi_index = 5, .roi_x = 506, .roi_y =  14, .roi_w = 388, .roi_h = 284  },
                { .roi_index = 6, .roi_x =   0, .roi_y = 598, .roi_w = 390, .roi_h = 196  },
                { .roi_index = 7, .roi_x = 506, .roi_y = 598, .roi_w = 390, .roi_h = 196  },
                { .roi_index = 0, .roi_x =   0, .roi_y =   0, .roi_w =   0, .roi_h =   0  },
                { .roi_index = 0, .roi_x =   0, .roi_y =   0, .roi_w =   0, .roi_h =   0  },
                { .roi_index = 0, .roi_x =   0, .roi_y =   0, .roi_w =   0, .roi_h =   0  },
                { .roi_index = 0, .roi_x =   0, .roi_y =   0, .roi_w =   0, .roi_h =   0  },

        }

STITCH initialization

int32_t init_stitch(test_ctx_t *test_ctx)
{
        int32_t ret = 0, i;
        hbn_buf_alloc_attr_t alloc_attr = {0};
        char res_file_name[128] = {0};
        struct stat fileStat;

        ret = hbn_vnode_open(HB_STITCH, 0, -1, &test_ctx->sth_handle);```c
        if (ret < 0) {
                printf("STH vnode open fail\n");
                return -1;
        }

        memset(res_file_name, 0, sizeof(res_file_name));
        sprintf(res_file_name, "%s/%s", g_res_path, "alpha_lut_apa.bin");
        if(stat(res_file_name, &fileStat) != 0) {
                printf("Failed to get file stats. cfg file = %s\n", res_file_name);
                return -1;
        }

        ret = hb_mem_alloc_com_buf(fileStat.st_size, HB_MEM_USAGE_MAP_INITIALIZED |
                                                        HB_MEM_USAGE_PRIV_HEAP_2_RESERVERD | HB_MEM_USAGE_CPU_READ_OFTEN |
                                                        HB_MEM_USAGE_CPU_WRITE_OFTEN | HB_MEM_USAGE_CACHED, &alpha_buffer);
        if (ret < 0) {
                printf("hb_mem_alloc_com_buf alpha_lut faild, ret = %d\n", ret);
                return -1;
        }

        load_file_2_buff(res_file_name, (char *)alpha_buffer.virt_addr, fileStat.st_size);
        hb_mem_flush_buf_with_vaddr((uint64_t)alpha_buffer.virt_addr, fileStat.st_size);

        base_attr.alpha_lut.share_id = alpha_buffer.share_id;
        base_attr.alpha_lut.vaddr = (uint64_t)alpha_buffer.virt_addr;
        base_attr.alpha_lut.size = fileStat.st_size;

        ret = hbn_vnode_set_attr(test_ctx->sth_handle, &base_attr);
        if (ret < 0) {
                printf("STH vnode set attr fail\n");
                return -1;
        }

        for (i = 0; i < SENSOR_NUMS; i++) {
                ret = hbn_vnode_set_ichn_attr(test_ctx->sth_handle, i, &inch_attr[i]);
                if (ret < 0) {
                        printf("STH vnode set ichn attr fail\n");
                        return -1;
                }
        }

        ret = hbn_vnode_set_ochn_attr(test_ctx->sth_handle, 0, &och_attr);
        if (ret < 0) {
                printf("STH vnode set ochn attr fail\n");
                return -1;
        }

        memset(&alloc_attr, 0, sizeof(hbn_buf_alloc_attr_t));
        alloc_attr.buffers_num = 3;
        alloc_attr.is_contig = 1;
        alloc_attr.flags = (int64_t)((uint64_t)HB_MEM_USAGE_CPU_READ_OFTEN |
                        (uint64_t)HB_MEM_USAGE_CPU_WRITE_OFTEN | (uint64_t)HB_MEM_USAGE_MAP_INITIALIZED);
        alloc_attr.flags |= (uint64_t)HB_MEM_USAGE_CACHED;

        ret = hbn_vnode_set_ochn_buf_attr(test_ctx->sth_handle, 0, &alloc_attr);
        if (ret < 0) {
                printf("STH vnode set ochn buf attr fail\n");
                return -1;
        }

        ret = hbn_vnode_start(test_ctx->sth_handle);
        if (ret < 0) {
                printf("STH vnode start fail\n");
                return -1;
        }

        return 0;
}

2v imx219 + MIPI + CIM + ISP + PYM + STITCH stitching and encoding sample

This sample acquires two video streams from a pipeline consisting of two IMX219 sensors, CIM, ISP, PYM, and other modules, then stitches the two streams vertically via the STITCH CODEC module into a single H.264 file named cim-isp-pym-stitch.h264.

Test procedure:

After installing two IMX219 sensors, power on the device and execute the following commands:

sunrise@ubuntu:~$ cd /app/multimedia_demo/camsys_demo/sample_2v_219_stitch_codec/
sunrise@ubuntu:/app/multimedia_demo/camsys_demo/sample_2v_219_stitch_codec$ make
sunrise@ubuntu:/app/multimedia_demo/camsys_demo/sample_2v_219_stitch_codec$ ./sample_2v_219_stitch_codec

The generated cim-isp-pym-stitch.h264 file playback is shown below:

STITCH configuration parameters:

struct stitch_base_attr sth_base_attr = {
		  .mode = 2,
		  .roi_nums = 2,
		  .img_nums = 2,
		  .alpha_lut = {
			.share_id = 0,
			.vaddr = 0,
			.offset = 0,
			.size = 0
		  },
		  .beta_lut = {
			.share_id = 0,
			.vaddr = 0,
			.offset = 0,
			.size = 0
		  },
		  .blending = {{
			  .roi_index = 0,
			  .blending_mode = 3,
			  .direct = 0,
			  .uv_en = 1,
			  .src0_index = 0,
			  .src1_index = 1,
			  .margin = 0,
			  .margin_inv = 128,
			  .gain_src0_yuv = {256, 256, 256},
			  .gain_src1_yuv = {256, 256, 256}
			}, {
			  .roi_index = 1,
			  .blending_mode = 3,
			  .direct = 0,
			  .uv_en = 1,
			  .src0_index = 1,
			  .src1_index = 0,
			  .margin = 0,
			  .margin_inv = 128,
			  .gain_src0_yuv = {256, 256, 256},
			  .gain_src1_yuv = {256, 256, 256}
			}, {
			  .roi_index = 0,
			  .blending_mode = 0,
			  .direct = 0,
			  .uv_en = 0,
			  .src0_index = 0,
			  .src1_index = 0,
			  .margin = 0,
			  .margin_inv = 0,
			  .gain_src0_yuv = {0, 0, 0},
			  .gain_src1_yuv = {0, 0, 0}
			}, {
			  .roi_index = 0,
			  .blending_mode = 0,
			  .direct = 0,
			  .uv_en = 0,
			  .src0_index = 0,
			  .src1_index = 0,
			  .margin = 0,
			  .margin_inv = 0,
			  .gain_src0_yuv = {0, 0, 0},
			  .gain_src1_yuv = {0, 0, 0}
			}, {
			  .roi_index = 0,
			  .blending_mode = 0,
			  .direct = 0,
			  .uv_en = 0,
			  .src0_index = 0,
			  .src1_index = 0,
			  .margin = 0,
			  .margin_inv = 0,
			  .gain_src0_yuv = {0, 0, 0},
			  .gain_src1_yuv = {0, 0, 0}
			}, {
			  .roi_index = 0,
			  .blending_mode = 0,
			  .direct = 0,
			  .uv_en = 0,
			  .src0_index = 0,
			  .src1_index = 0,
			  .margin = 0,
			  .margin_inv = 0,
			  .gain_src0_yuv = {0, 0, 0},
			  .gain_src1_yuv = {0, 0, 0}
			}, {
			  .roi_index = 0,
			  .blending_mode = 0,
			  .direct = 0,
			  .uv_en = 0,
			  .src0_index = 0,
			  .src1_index = 0,
			  .margin = 0,
			  .margin_inv = 0,
			  .gain_src0_yuv = {0, 0, 0},
			  .gain_src1_yuv = {0, 0, 0}
			}, {
			  .roi_index = 0,
			  .blending_mode = 0,
			  .direct = 0,
			  .uv_en = 0,
			  .src0_index = 0,
			  .src1_index = 0,
			  .margin = 0,
			  .margin_inv = 0,
			  .gain_src0_yuv = {0, 0, 0},
			  .gain_src1_yuv = {0, 0, 0}
			}, {
			  .roi_index = 0,
			  .blending_mode = 0,
			  .direct = 0,
			  .uv_en = 0,
			  .src0_index = 0,
			  .src1_index = 0,
              .margin = 0,
			  .margin_inv = 0,
			  .gain_src0_yuv = {0, 0, 0},
			  .gain_src1_yuv = {0, 0, 0}
			}, {
			  .roi_index = 0,
			  .blending_mode = 0,
			  .direct = 0,
			  .uv_en = 0,
			  .src0_index = 0,
			  .src1_index = 0,
			  .margin = 0,
			  .margin_inv = 0,
			  .gain_src0_yuv = {0, 0, 0},
			  .gain_src1_yuv = {0, 0, 0}
			}, {
			  .roi_index = 0,
			  .blending_mode = 0,
			  .direct = 0,
			  .uv_en = 0,
			  .src0_index = 0,
			  .src1_index = 0,
			  .margin = 0,
			  .margin_inv = 0,
			  .gain_src0_yuv = {0, 0, 0},
			  .gain_src1_yuv = {0, 0, 0}
			}, {
			  .roi_index = 0,
			  .blending_mode = 0,
			  .direct = 0,
			  .uv_en = 0,
			  .src0_index = 0,
			  .src1_index = 0,
			  .margin = 0,
			  .margin_inv = 0,
			  .gain_src0_yuv = {0, 0, 0},
			  .gain_src1_yuv = {0, 0, 0}
			}}
};

struct stitch_ch_attr sth_inch_attr[] = {
           {
			.width = 1920,
			.height = 1080,
			.strid = {1920, 1920},
			.rois = {{
				.roi_index = 0,
				.roi_x = 0,
				.roi_y = 0,
				.roi_w = 1920,
				.roi_h = 1080
			  }, {
				.roi_index = 1,
				.roi_x = 0,
				.roi_y = 0,
				.roi_w = 1920,
				.roi_h = 1080
			  },
			}
		  }, {
			.width = 1920,
			.height = 1080,
			.strid = {1920, 1920},
			.rois = {{
				.roi_index = 0,
				.roi_x = 0,
				.roi_y = 0,
				.roi_w = 1920,
				.roi_h = 1080
			  }, {
				.roi_index = 1,
				.roi_x = 0,
				.roi_y = 0,
				.roi_w = 1920,
				.roi_h = 1080
			  },
			}
		  }, {
			.width = 0,
			.height = 0,
			.strid = {0, 0},
		  }, {
			.width = 0,
			.height = 0,
			.strid = {0, 0},
		  }
};

struct stitch_ch_attr sth_och_attr = {
		  .width = 1920,
		  .height = 2160,
		  .strid = {1920, 1920},
		  .rois = {{
			  .roi_index = 0,
			  .roi_x = 0,
			  .roi_y = 0,
			  .roi_w = 1920,
			  .roi_h = 1080
			}, {
			  .roi_index = 1,
			  .roi_x = 0,
			  .roi_y = 1080,
			  .roi_w = 1920,
			  .roi_h = 1080
			},
		  }
};

Create a vflow pipeline:

	int i = 0;
	ret = hbn_vflow_create(&vflow_fd[i]);
	if (ret < 0) {
		printf("hbn_vflow_create[%d]:%d error\n", i, __LINE__);
		goto err;
	}

	ret = hbn_camera_create(&cam_cfg[i], &cam_vnode_fd[i]);
	if (ret < 0) {
		printf("hbn_camera_create[%d]:%d error\n", i, __LINE__);
		goto err1;
	}

	vin_vnode_fd[i] = vin_vnode_create(&vin_attr[i]);
	if (vin_vnode_fd[i] < 0) {
		ret = (int32_t)vin_vnode_fd[i];
		printf("vin_vnode_init[%d]:%d error\n", i, __LINE__);
		goto err1;
	}

	ret = hbn_vflow_add_vnode(vflow_fd[i], vin_vnode_fd[i]);
	if (ret < 0) {
		printf("hbn_vflow_add_vnode[%d]:%d error\n", i, __LINE__);
		goto err1;
	}

	isp_vnode_fd[i] = isp_vnode_create(&isp_cfg[i]);
	if (isp_vnode_fd[i] < 0) {
		ret = (int32_t)isp_vnode_fd[i];
		printf("isp_vnode_init[%d]:%d error\n", i, __LINE__);
		goto err1;
	}

	ret = hbn_vflow_add_vnode(vflow_fd[i], isp_vnode_fd[i]);
	if (ret < 0) {
		printf("hbn_vflow_add_vnode[%d]:%d error\n", i, __LINE__);
		goto err1;
	}

	ynr_vnode_fd[i] = ynr_vnode_create(&ynr_info[i]);
	if (ynr_vnode_fd[i] < 0) {
		ret = (int32_t)ynr_vnode_fd[i];
		printf("ynr_vnode_init[%d]:%d error\n", i, __LINE__);
		goto err1;
	}

	ret = hbn_vflow_add_vnode(vflow_fd[i], ynr_vnode_fd[i]);
	if (ret < 0) {
		printf("hbn_vflow_add_vnode[%d]:%d error\n", i, __LINE__);
		goto err1;
	}

	pym_vnode_fd[i] = pym_vnode_create(&pym_cfg[i]);
	if (pym_vnode_fd[i] < 0) {
		ret = (int32_t)pym_vnode_fd[i];
		printf("pym_vnode_init[%d]:%d error\n", i, __LINE__);
		goto err1;
	}

	ret = hbn_vflow_add_vnode(vflow_fd[i], pym_vnode_fd[i]);
	if (ret < 0) {
		printf("hbn_vflow_add_vnode[%d]:%d error\n", i, __LINE__);
		goto err1;
	}

	sth_vnode_fd = sth_vnode_create();
	if (sth_vnode_fd < 0) {
		ret = (int32_t)sth_vnode_fd;
		printf("sth_vnode_init[%d]:%d error\n", i, __LINE__);
		goto err1;
	}

	ret = hbn_vflow_add_vnode(vflow_fd[i], sth_vnode_fd);
	if (ret < 0) {
		printf("hbn_vflow_add_vnode[%d]:%d error\n", i, __LINE__);
		goto err1;
	}

	ret = hbn_camera_attach_to_vin(cam_vnode_fd[i], vin_vnode_fd[i]);
	if (ret < 0) {
		printf("hbn_vflow_bind_vnode[%d]:%d error\n", i, __LINE__);
		goto err1;
	}

	ret = hbn_vflow_bind_vnode(vflow_fd[i], vin_vnode_fd[i], 0, isp_vnode_fd[i], 0);
	if (ret < 0) {
		printf("hbn_vflow_bind_vnode[%d]:%d error\n", i, __LINE__);
		goto err1;
	}
```Create another vflow and bind both vflows to the same stitch:
```c
	i = 1;
	ret = hbn_vflow_create(&vflow_fd[i]);
	if (ret < 0) {
		printf("hbn_vflow_create[%d]:%d error\n", i, __LINE__);
		goto err1;
	}

	ret = hbn_camera_create(&cam_cfg[i], &cam_vnode_fd[i]);
	if (ret < 0) {
		printf("hbn_camera_create[%d]:%d error\n", i, __LINE__);
		goto err1;
	}

	vin_vnode_fd[i] = vin_vnode_create(&vin_attr[i]);
	if (vin_vnode_fd[i] < 0) {
		ret = (int32_t)vin_vnode_fd[i];
		printf("vin_vnode_init[%d]:%d error\n", i, __LINE__);
		goto err1;
	}

	ret = hbn_vflow_add_vnode(vflow_fd[i], vin_vnode_fd[i]);
	if (ret < 0) {
		printf("hbn_vflow_add_vnode[%d]:%d error\n", i, __LINE__);
		goto err1;
	}

	isp_vnode_fd[i] = isp_vnode_create(&isp_cfg[i]);
	if (isp_vnode_fd[i] < 0) {
		ret = (int32_t)isp_vnode_fd[i];
		printf("isp_vnode_init[%d]:%d error\n", i, __LINE__);
		goto err1;
	}

	ret = hbn_vflow_add_vnode(vflow_fd[i], isp_vnode_fd[i]);
	if (ret < 0) {
		printf("hbn_vflow_add_vnode[%d]:%d error\n", i, __LINE__);
		goto err1;
	}

	ynr_vnode_fd[i] = ynr_vnode_create(&ynr_info[i]);
	if (ynr_vnode_fd[i] < 0) {
		ret = (int32_t)ynr_vnode_fd[i];
		printf("ynr_vnode_init[%d]:%d error\n", i, __LINE__);
		goto err1;
	}

	ret = hbn_vflow_add_vnode(vflow_fd[i], ynr_vnode_fd[i]);
	if (ret < 0) {
		printf("hbn_vflow_add_vnode[%d]:%d error\n", i, __LINE__);
		goto err1;
	}
	pym_vnode_fd[i] = pym_vnode_create(&pym_cfg[i]);;
	if (pym_vnode_fd[i] < 0) {
		ret = (int32_t)pym_vnode_fd[i];
		printf("pym_vnode_init[%d]:%d error\n", i, __LINE__);
		goto err1;
	}

	ret = hbn_vflow_add_vnode(vflow_fd[i], pym_vnode_fd[i]);
	if (ret < 0) {
		printf("bn_vflow_add_vnode[%d]:%d error\n", i, __LINE__);
		goto err1;
	}

	ret = hbn_vflow_add_vnode(vflow_fd[i], sth_vnode_fd);
	if (ret < 0) {
		printf("hbn_vflow_add_vnode[%d]:%d error\n", i, __LINE__);
		goto err1;
	}

	ret = hbn_camera_attach_to_vin(cam_vnode_fd[i], vin_vnode_fd[i]);
	if (ret < 0) {
		printf("hbn_camera_attach_to_vin[%d]:%d error\n", i, __LINE__);
		goto err1;
	}

	ret = hbn_vflow_bind_vnode(vflow_fd[i], vin_vnode_fd[i], 0, isp_vnode_fd[i], 0);
	if (ret < 0) {
		printf("hbn_vflow_bind_vnode[%d]:%d error\n", i, __LINE__);
		goto err1;
	}

	ret = hbn_vflow_bind_vnode(vflow_fd[i], isp_vnode_fd[i], 1, ynr_vnode_fd[i], 0);
	if (ret < 0) {
		printf("hbn_vflow_bind_vnode[%d]:%d error\n", i, __LINE__);
		goto err1;
	}

	ret = hbn_vflow_bind_vnode(vflow_fd[i], ynr_vnode_fd[i], 1, pym_vnode_fd[i], 0);
	if (ret < 0) {
		printf("hbn_vflow_bind_vnode[%d]:%d error\n", i, __LINE__);
		goto err1;
	}

	ret = hbn_vflow_bind_vnode(vflow_fd[i], pym_vnode_fd[i], 0, sth_vnode_fd, 1);
	if (ret < 0) {
		printf("hbn_vflow_bind_vnode[%d]:%d error\n", i, __LINE__);
		goto err1;
	}

Configure the CODEC encoding module:

	//config codec
	ret = codec_config_param(&context, MEDIA_CODEC_ID_H264, sth_och_attr.width, sth_och_attr.height);
	if (ret < 0) {
		printf("codec_config_param error!!!\n");
		goto err1;
	}

	ret = codec_init(&context);
	if (ret < 0) {
		printf("codec_init error!!!\n");
		goto err1;
	}

	ret = codec_start(&context);
	if (ret < 0) {
		printf("codec_init error!!!\n");
		goto err2;
	}

	h264fd = fopen(H264_FNAME, "w+");
    if (h264fd == NULL) {
        printf("open(%s) fail", H264_FNAME);
		ret = -1;
        goto err3;
    }

	ret = hbn_vflow_start(vflow_fd[0]);
	ret |= hbn_vflow_start(vflow_fd[1]);
	if (ret < 0) {
		printf("codec_init error!!!\n");
		goto err3;
	}

Continuously fetch frames, send them to the CODEC for encoding, then retrieve encoded frames from the CODEC and save them as an H.264 file:

	while (imgframe.cnt < 30 * TIMEOUT) {
		ret = hbn_vnode_getframe(sth_vnode_fd, 0, 1000, &imgframe.vnode_buffer);
		printf("sth_worker, ret = %d\n", ret);
		if (ret == 0) {
			ret = codec_set_input(&context, &imgframe);
			if (ret < 0) {
				printf("codec_set_input error!!!\n");
				goto err4;
			}

			ret = codec_get_output(&context, &imgframe);
			if (ret < 0) {
				printf("codec_get_output error!!!\n");
				goto err4;
			}

            ret = write_output_h264(&imgframe, h264fd);
			if (ret < 0) {
				printf("write_output_h264 error!!!\n");
				goto err4;
			}

			ret = codec_release_output(&context, &imgframe);
			if (ret < 0) {
				printf("codec_release_output error!!!\n");
				goto err4;
			}

			hbn_vnode_releaseframe(sth_vnode_fd, 0, &imgframe.vnode_buffer);
		} else {
			printf("hbn_vnode_getframe fail, ret = %d\n", ret);
			goto err4;
		}

		imgframe.cnt++;
	}

V4L2 Sample

imx219 + MIPI + CIM + ISP + PYM:

v4l2-ctl -d 0 --set-fmt-video=width=1920,height=1080,pixelformat=NV12 --stream-mmap --stream-count=120 --stream-to=/userdata/test.yuv

imx219 + MIPI + CIM + ISP + GDC:

The v4l2 GDC application currently cannot generate config bin files from JSON files, so v4l2 GDC testing can only be performed using pre-generated config bin files.
Compared with the original v4l2 stream capture code, the v4l2 GDC stream capture code requires the following additional configurations:

# Need to add GDC input image width and height parameter configuration
if (TestContext[i].gdc_cfg) {
    TestContext[i].pic_width = 1920;
    TestContext[i].pic_height = 1080;
    TestContext[i].in_pic_width = 1920;  // Newly added input image width
    TestContext[i].in_pic_height = 1080; // Newly added input image height
}

# Need to add GDC config configuration
// Allocate memory for GDC config bin
int map_gdc_config_buffer(hb_mem_common_buf_t *hb_common_buf, uint32_t size)
{
    int64_t alloc_flags = 0;
    int ret;

    alloc_flags = HB_MEM_USAGE_PRIV_HEAP_2_RESERVED | HB_MEM_USAGE_CPU_READ_OFTEN | HB_MEM_USAGE_CPU_WRITE_OFTEN | HB_MEM_USAGE_CACHED;
    memset(hb_common_buf, 0, sizeof(hb_mem_common_buf_t));
    ret = hb_mem_alloc_com_buf(size, alloc_flags, hb_common_buf);
    if (ret < 0) {
        vio_gtest_err("hb_mem_alloc_com_buf size %u failed \n", size);
        return ret;
    }

    return 0;
}

// ioctl interface to send configuration to GDC v4l2 driver
int v4l2_set_ext_ctrl(int fd, uint32_t cmd, void *arg)
{
    int rc;
    struct v4l2_ext_controls ext_ctrl = {0};
    struct v4l2_ext_control ctrl = {0};

    ext_ctrl.controls = &ctrl;
    ext_ctrl.controls->id = cmd;
    ext_ctrl.controls->ptr = arg;
    ext_ctrl.count = 1;

    rc = ioctl(fd, VIDIOC_S_EXT_CTRLS, &ext_ctrl);
    if (rc < 0)
        vio_gtest_err("%s, cmd=%d, rc=%d\n", strerror(errno), cmd, rc);
    return rc;
}

int v4l2_gdc_init(vpm_test_context *ptc)
{
    int fd, ret;
    FILE *file = NULL;
    struct stat fileStat;
    hb_mem_common_buf_t hb_common_buf;
    gdc_config_t gdc_user_cfg;
    work_info_t *winfo = &ptc->work_info;

    if (!ptc->gdc_cfg || !winfo->priv_fd)
        return -1;

    file = fopen(ptc->gdc_cfg, "r");
    if (file == NULL) {
        perror("Error opening file\n");
        return -1;
    }
    // Get GDC config bin size
    ret = fstat(fileno(file), &fileStat);
    if (ret) {
        perror("Error getting file status");
        goto err;
    }

    vio_gtest_info("File size: %ld bytes\n", fileStat.st_size);
    // Allocate memory to store GDC config bin
    ret = map_gdc_config_buffer(&hb_common_buf, fileStat.st_size);
    if (ret)
        goto err;
    // Copy GDC config bin content into the allocated memory
    if (fread(hb_common_buf.virt_addr, 1, fileStat.st_size, file) != fileStat.st_size) {
        vio_gtest_err("failed to read gdc config file!\n");
        ret = -1;
        goto err;
    }
    vio_gtest_info("gdc config bin buffer phy_addr:%p virt_addr:%p size:%d\n",
        hb_common_buf.phys_addr, hb_common_buf.virt_addr, hb_common_buf.size);

    ret = hb_mem_flush_buf_with_vaddr((uint64_t)hb_common_buf.virt_addr, fileStat.st_size);
    if (ret) {
        vio_gtest_err("failed to hb_mem_flush_buf_with_vaddr!\n");
        goto err;
    }

    gpm[winfo->pipe_id].gdc_config.config_addr = (uint64_t)hb_common_buf.virt_addr;
    gpm[winfo->pipe_id].gdc_config.config_size = hb_common_buf.size;
    // GDC input image width and height
    gpm[winfo->pipe_id].gdc_config.output_width = ptc->pic_width;
    gpm[winfo->pipe_id].gdc_config.output_height = ptc->pic_height;
    gpm[winfo->pipe_id].gdc_config.output_stride = ALIGN_UP(ptc->pic_width, STRIDE_ALIGN);
    // GDC output image width and height
    gpm[winfo->pipe_id].gdc_config.input_width = ptc->in_pic_width;
    gpm[winfo->pipe_id].gdc_config.input_height = ptc->in_pic_height;
    gpm[winfo->pipe_id].gdc_config.input_stride = ALIGN_UP(ptc->in_pic_width, STRIDE_ALIGN);

    gpm[winfo->pipe_id].gdc_config.div_width = 0;
    gpm[winfo->pipe_id].gdc_config.div_height = 0;
    gpm[winfo->pipe_id].gdc_config.sequential_mode = 0;
    gpm[winfo->pipe_id].gdc_config.total_planes = 2;

    gpm[winfo->pipe_id].binary_ion_id = hb_common_buf.share_id;
    gpm[winfo->pipe_id].binary_offset = hb_common_buf.offset;

    gpm[winfo->pipe_id].magicNumber = 0x12345678;

    // Send configuration to GDC v4l2 driver
    ret = v4l2_set_ext_ctrl(winfo->priv_fd, V4L2_CID_DR_GDC_ATTR, &gpm[winfo->pipe_id]);
    if (ret) {
        vio_gtest_err("v4l2_set_ext_ctrl error!!!\n");
        goto err;
    }

err:
    fclose(file);
    return ret;

}

# Release GDC config bin
void v4l2_gdc_deinit (vpm_test_context *ptc)
{
    work_info_t *winfo = &ptc->work_info;
    hb_mem_free_buf_with_vaddr((uint64_t)gpm[winfo->pipe_id].gdc_config.config_addr);
}