Exit
  • Global community
    • Language:
      • Deutsch
      • English
      • Español
      • Français
      • Português
  • 日本語コミュニティ
    Dedicated community for Japanese speakers
  • 한국 커뮤니티
    Dedicated community for Korean speakers
0

Spreading Adobe Encoder across multiple CPUs?

New Here ,
Jan 01, 2012 Jan 01, 2012

Is it possible to spread the encoding out across multiple CPUs like a rendering farm does for 3D movies and such? Is there any such encoder that allows this setup? We process a lot of videos and even with very powerful machines I feel we need a way to run our operation more sufficiently.

Idealy, I would like five workstations editing video then feeding it to this "encoding farm" to handle it.

Thanks in advance for any help,

Thomas

19.6K
Translate
Report
Community guidelines
Be kind and respectful, give credit to the original source of content, and search for duplicates before posting. Learn more
community guidelines
Community Expert ,
Jan 01, 2012 Jan 01, 2012

No for Adobe software

I have seen mention of specialized computers & software, I think starting at $50k per seat, but did not make note since I am not a Hollywood studio

Translate
Report
Community guidelines
Be kind and respectful, give credit to the original source of content, and search for duplicates before posting. Learn more
community guidelines
LEGEND ,
Jan 01, 2012 Jan 01, 2012

Adobe Media Encoder (AME) does not support automated distributed rendering, so neither does Premiere Pro. However, After Effects uses its own encoding engine and will happily do network rendering to still image sequences - see the AE help file for more info.

If you're working in Premiere with silent footage that has very complex effects, because AME can export to still images it is possible to manually set up a farmed render on a PP sequence; by copying the entire project to each workstation and setting different frame ranges for each one to export. You'd then bring all the image files back to a single workstation for encoding to video, but you're only saving time on the effects so the sequence has to be monumental for it to be worth the time and effort to set up each job.

In contrast, After Effects' effects can take a very long time to render and are always applied frame-by-frame, which is why it can support distribution. A single seat of AME with access to decent hardware (including CUDA GPUs) can convert those stills to a video file extremely quickly.

Distributed rendering  of video isn't a widely-used concept, as the compression processes  don't like frames arriving out of sync. There are a few solutions which can farm the video  encoding step, such as Promedia Carbon, but they're strictly the domain of studios (you're looking at tens of thousands of dollars in licenses, even for a small network). They're used for transcoding rather than rendering (e.g. converting tape library footage for the Web, telecine and pulldown).

Translate
Report
Community guidelines
Be kind and respectful, give credit to the original source of content, and search for duplicates before posting. Learn more
community guidelines
Engaged ,
Jan 01, 2012 Jan 01, 2012
LATEST

When performing distributed processing of video (Apple Compressor does this btw), there is an extra step where it has to piece back the separate parts into a single movie. This can sometimes take longer than the encoding process. But there can be an appreciable time saving for longer movies (30 mins+) so it's a feature I'd very much like to see in AME.

Translate
Report
Community guidelines
Be kind and respectful, give credit to the original source of content, and search for duplicates before posting. Learn more
community guidelines