Render optimization and further conversion in ffmpeg

[elijahk7811267]

Hello,

I have faced with some questions concerning a possibility of render optimization for some projects. There's no 3D. The main part of the projects are some footages, photos and some shapes. The main problems for rendering are blur effects and 3D-layers transformation with DoF.

The main question is render efficiency increasing with hardware.

As I've learned it's useless trying to increase efficiency via GPU acceleration, isn't it? So the main question is efficiency of multi machine rendering. Is it worth of money or the best way is to use maximum from one machine. It's very important to get the final video as quick as possible. So what is the best and the fastest way: using one machine and render the final video directly or using multi machine render and then combining frames with audio in ffmpeg?

Also using AME is restricted because it doesn't have console version.

By the way, if someone know this stuff too:

The final result is HLS video in h.264 codec. What is the best render format to use it in further conversion in HLS video via ffmpeg? What is the best way to include audio in final video if the render result is image sequence?

Regards,

Ilia

[Mylenium · Accepted solution]

These issues have been discussed at length in as many threads here on this forum as our galaxy has stars. Just buying more computers or GPUs won't do anything, either way. None of that does you any good if the source materials are stored on slow drives, are clip-based, may be compressed and need decoding to temp files and a million other things. Conversely it will matter if the rest of the machine can keep up with a fast CPU down to the speed at which your PCI lanes shuffle data around and whether or not specific effects and features are involved that always work single-threaded and slow down AE. You have to take a much broader view here. There simply is no simple, dumb way of resolving every AE performance issue and even the best hardware in teh world can go to waste if only the project is complex enough or uses a quirky, slow feature. That and of course it's all relative. Even with all the optimizations rendering a single frame in AE can still take forever. Well, whatever, you'd have to be much more specific about your projects and your expections. Just saying that you want the fastest possible way is not useful. Also many of those limitations will also apply to processing outside of AE. It may not at all be of much use to use an external encoder on a slow server when your actual bottleneck is network bandwidth, i.e. the fast render machines' frames getting stuck in traffic. Again, you have to think much bigger.

Mylenium