06-18-2018, 09:55 AM
Hello,
Thank you for the link.
This project is interesting, However, on this project the resolution processed is 320x240 where I need to process 1920x1080 in my first prototype and 3840x2160 in "production". Due to resolution authors can use a raw video data connection. With 4k resolution we must compress and uncompress data what uses a good part of GPU.
The only way to know it can really work is to test in real conditions on the Rock64 board. For that I need to make OpenCV working on GPU to decode H.264 and send images to OpenGL to apply simple filters (through shaders). OpenCV is not used to process the image like motion detection or tracking etc.
Does someone can help me make OpenCV working on GPU or maybe know an other way to decode H.264 (and MjPEG) and inject images into OpenGL buffer to efficiently applly sharders using OpenGL Library ?
Just for information I develop in C/C++.
Thank you for the link.
This project is interesting, However, on this project the resolution processed is 320x240 where I need to process 1920x1080 in my first prototype and 3840x2160 in "production". Due to resolution authors can use a raw video data connection. With 4k resolution we must compress and uncompress data what uses a good part of GPU.
The only way to know it can really work is to test in real conditions on the Rock64 board. For that I need to make OpenCV working on GPU to decode H.264 and send images to OpenGL to apply simple filters (through shaders). OpenCV is not used to process the image like motion detection or tracking etc.
Does someone can help me make OpenCV working on GPU or maybe know an other way to decode H.264 (and MjPEG) and inject images into OpenGL buffer to efficiently applly sharders using OpenGL Library ?
Just for information I develop in C/C++.