I would say this is an 80% software and 20% hardware project — as is so often the case, the software work can't really be captured in photos. I have written a custom video encoder/decoder that is optimized for this exact task: low-latency, parallel-processable video encoding/decoding.

So far, it is possible to stream some Linux distros and Win11. Compression/encoding takes place as close to the source as possible, directly on the GPU, i.e. hardware accelerated. The encoder is written in DirectX on Win11 and OpenGL on Linux, plus a C++ server app.

Things I think are very cool: with this custom encoder, I am fully GPU-vendor-agnostic. I don't need a special GPU with dedicated encoder blocks using a video format that then forces me to find decoder ICs on the client side that have everything I need.
Instead, I'm fully flexible in rewriting my FPGA code and can drive whatever display I want: parallel RGB, LVDS, MIPI, etc.