xjhsxjh / uvc-gadget Goto Github PK
View Code? Open in Web Editor NEWThis project forked from jdonald/uvc-gadget
Laurent Pinchart's repo with required patches as directed by gadget-testing.txt, made functional on Raspberry Pi Zero
This project forked from jdonald/uvc-gadget
Laurent Pinchart's repo with required patches as directed by gadget-testing.txt, made functional on Raspberry Pi Zero
UVC gadget test application Readme Copyright (C) 2010 Ideas on board SPRL Copyright (C) 2013 ST Microelectronics Ltd. Contacts: Laurent Pinchart <laurent.pinchart@xxxxxxxxxxxxxxxx> Bhupesh Sharma <bhupesh.sharma@xxxxxx> Introduction ============ This README file documents the UVC gadget test application available here git://git.ideasonboard.org/uvc-gadget.git, which can be used to test the UVC webcam gadget driver located under drivers/usb/gadget directory. Possible Use-Case scenarios =========================== There can be a number of scenarios under which the UVC webcam gadget can be used and each use-case can vary in the following ways: - Video streaming endpoint choice: -------------------------------- As per UVC specifications (Revision 1.1), either Bulk or Isochronous USB endpoints can be used are for implementing UVC Video streaming interface. Both have there unique challenges as: Isochronous video streaming interface will have two alt-settings: - Alt-setting 0 for zero-bandwidth mode, and - Alt-setting 1 for full-bandwidth mode. This requires us to ACK status stage of SET_ALT(alt-setting = 1), only when we have video frames available which can be queued to UVC domain. Bulk video streaming interface, on the other hand will have only one alt-setting. This requires us to prepare UVC gadget for streaming when we receive a a UVC_VS_COMMIT_CONTROL command from USB host and actually start streaming as soon as we have video frames available which can be queued to UVC domain. - Availability of a video capture device: --------------------------------------- To emulate a real UVC webcam gadget (capturing live video from a video capture source and sending it over USB bus), it is possible to integrate a V4L2 based video-capture device driver with the UVC gadget test application acting as an interface between the UVC based video-output device and V4L2 based video-capture device. Accordingly, there can be the following scenarios possible: - No video capture device available: In this case, there is no video capture device available and the UVC webcam gadget is working in an standalone environment. It is possible in this case to stream a dummy test pattern from the UVC webcam gadget and view the same on the Linux/Windows USB Host machine using standard video capture utilities like CHEESE. - VIVI (Virtual Video driver) acting as a video capture source: In case we don't have a real video capture hardware (e.g. sensor) available, it is still possible to exercise the complete path from the V4L2 based video-capture domain to UVC based video-output domain, using the Virtual Video driver (VIVI) available here: drivers/media/platform/vivi.c It is possible in this case to stream changing vertical color bar pattern pattern from the VIVI capture device and send it to UVC webcam gadget, which will eventually send these frames over USB bus. Similar to the case above, it is possible to view the frames on the Linux/Windows USB Host machine using standard video capture utilities like CHEESE. - Real video capture hardware and corresponding V4L2 driver available: In this case, we have a real video capture hardware (e.g. CCD / CMOS sensor) available. Here, it is possible to exercise the complete path from the V4L2 based video-capture driver domain to UVC based video-output domain, using the UVC gadget test application. It is possible in this case to stream real video/still images from capture source and send it to UVC webcam gadget, which will eventually send these frames over USB bus. - IO methods supported: -------------------- V4L2 framework supports a number of IO methods and to achieve zero-memcpy of video frames as they pass from V4L2 video-capture domain to UVC video-output domain, it is important to select appropriate IO methods for working on videobuffers. Accordingly we can have the following scenarios possible: - UVC webcam standalone, supporting IO_METHOD_MMAP method: In this case, there is no video capture device available and the UVC webcam gadget is working in an standalone environment. It is supporting IO_MMAP method, which implies that the videobuffers are allocated by the UVC webcam gadget itself and they are filled with dummy test pattern by UVC gadget test application. - UVC webcam standalone, supporting IO_METHOD_USERPTR method: In this case, there is no video capture device available and the UVC webcam gadget is working in an standalone environment. It is supporting IO_METHOD_USERPTR method, which implies that the videobuffers are allocated and filled with dummy test pattern by the UVC gadget test application and UVC webcam gadget operates on a USERPTR of these videobuffers. - UVC webcam integrated with V4L2 video-capture driver, with V4L2 video-capture driver supporting IO_METHOD_MMAP method and UVC webcam supporting IO_METHOD_USERPTR: In this case, either VIVI or another V4L2 based driver (working on real HW) is available as a video capture device and is integrated with the UVC webcam gadget via UVC gadget test application, with V4L2 video-capture driver supporting IO_METHOD_MMAP method and UVC webcam supporting IO_METHOD_USERPTR. Here, the videobuffers are allocated in the V4L2 video-capture domain and UVC gadget test application acts as a means to pass a USERPTR of these videobuffers to the UVC webcam gadget domain, without requiring any memcpy from the CPU. - UVC webcam integrated with V4L2 video-capture driver, with V4L2 video-capture driver supporting IO_METHOD_USERPTR method and UVC webcam supporting IO_METHOD_MMAP: In this case, either VIVI or another V4L2 based driver (working on real HW) is available as a video capture device and is integrated with the UVC webcam gadget via UVC gadget test application, with V4L2 video-capture driver supporting IO_METHOD_USERPTR method and UVC webcam supporting IO_METHOD_MMAP. Here, the videobuffers are allocated in the UVC webcam gadget domain and UVC gadget test application acts as a means to pass a USERPTR of these videobuffers to the V4L2 video-capture domain, without requiring any memcpy from the CPU. Modifying UVC gadget test application ===================================== The UVC webcam gadget kernel driver located under drivers/usb/gadget directory and compilable as g_webcam.ko kernel module, supports 360p and 720p frame resolutions and YUV4:2:2 and MJPEG frame formats by default. So, the video capture device must be aligned to support similar frame format and resolutions. In case the video capture device supports other resolutions and frame formats, we need to replace the default values used in the UVC gadget test application with the same and also update the UVC webcam gadget kernel driver. Keeping USBCV's UVC test-suite happy ==================================== The UVC webcam gadget kernel driver located under drivers/usb/gadget directory and compilable as g_webcam.ko kernel module, supports changing the Brightness attribute of the Processing Unit (PU) by default. If the underlying video capture device supports changing the Brightness attribute of the image being acquired (like the Virtual Video, VIVI driver does), then we should route this UVC request to the respective video capture device. Incase, there is no actual video capture device associated with the UVC gadget and we wish to use this application as the final destination of the UVC specific requests then we should return pre-cooked (static) responses to GET_CUR(BRIGHTNESS) and SET_CUR(BRIGHTNESS) commands to keep command verifier test tools like UVC class specific test suite of USBCV, happy. Note that the values used in UVC gadget test application are in sync with the VIVI driver and must be changed for your specific video capture device. These values also work well in case there in no actual video capture device.
A declarative, efficient, and flexible JavaScript library for building user interfaces.
๐ Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. ๐๐๐
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google โค๏ธ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.