GstCUDA - cudadebayer
This page describes in detail the cudadebayer ad-on element of the GstCUDA plugin.
Description
Cudadebayer is a one input/output pad video filter GStreamer element, that executes bayer to RGB conversion on the GPU through a CUDA algorithm. This element offers a de-bayering GPU based solution alternative.
This element has the bayer to RGB conversion CUDA based algorithm integrated into it. Because of that, in contrast with the cudafilter element that could execute a custom CUDA algorithm, the cudadebayer only executes the de-bayering algorithm.
Cudadebayer is a complete example showing the GstCUDA framework in action. It is an element that uses the GstCUDA framework to easily interface GStreamer with CUDA. All the complexity of the NVMM memory types buffers handling and interface with zero memory copies between GStreamer and CUDA is absorbed by the GstCuda framework provided classes and functions.
This is a clear example that demonstrate how the GstCUDA framework could be used to generate a specific custom element that executes a specific CUDA algorithm integrated into the element itself (for example a final product GStreamer element). The flow process of creating a final product GStreamer element with GstCUDA framework maybe something like that:
- First validate and debug the CUDA algorithm using the cudafilter element.
- Then when the prototyping stage was done, generate the custom final product GStreamer element based on the GstCUDA framework.
Key features
- De-bayering GPU based solution alternative.
- High performance, due to zero memory copies interface between CUDA and GStreamer.
- Directly handle of NVMM memory type buffers.
Element properties description
Under construction
Element inspect
Under construction