Unfortunately AB does not address that. It is currently decode-only for ProRes and ProRes RAW. The Jonathan Morrison video does show how even ProRes can become CPU-limited for decode if using enough streams at high resolution, and AB benefits this.
In theory an FPGA can be programmed to decode H.264, and there are some academic papers written on that. However Quick Sync and T2 are already pretty fast at that. The question is could an FPGA be any faster.
It's tempting to say H.264/HEVC is for amateurs, but some pro cameras like the Panasonic EVA-1 use HEVC and even the $11,000 Sony FX9 uses XAVC-L Long GOP.
So there is a valid argument for multi-stream, super-fast, hardware-assisted decoding of H.264/HEVC.
You could argue that the amateurs can wait a day to transcode, it's the professionals who need to work immediately on a wide array of submitted 4k Long GOP material, and would be willing to pay handsomely for this capability.
Hm Hm Hm.
It says that the AB helps with transcoding of prores, not encoding.
Encoding means encoding ProRes from uncompressed Video (wherever that occurs).
Transcoding means changing from one compressed format to another. So it COULD mean it helps with rendering ProRes files from h264s. I‘m absolutely aware that the AB right now doesn’t decode h264, but maybe it can take half the load of such a transcoding operation. 50% for CPU and GPU for the decoding/50% AB for the actual transcoding part. Or am I wrong here?
I dont know your Definition of professorial, but If I look at two typical examples that I had to deal with in the past 6 weeks, one was a film for Mercedes Benz about a brand new car and it was shot partially on sony cameras in some h264 mxf format, and the other one was a facebook series with some influencer, That also had some red prores files and some h264 Sony mxfs and GoPro h264 material.
And both shows were multicam 4K, sometimes up to 4 cameras At once.
On the trashcan I definitely had to use proxies, will be super interesting how these projects behave on my new machine.
But I really would like to edit stuff like this where I encounter mixed source footage in native resolution at least. If a show already has 7 TB of footage I‘m not even sure if encoding everything to ProRes 422 HQ first is an option..
Still even if only some camera streams natively are ProRes, I think the editing could benefit from the Afterburner because the CPUs could focus entirely on the none prores files.