H.264 export times are long, looking to clear up render settings misconceptions

[Ryan Holmes]

Let me see if I can condense what you just stated:

You are shooting 3 cameras at different frame rates, shooting to some various flavors of h.264, editing natively, adding effects and color correction, then exporting back out to h.264 on a relatively lightweight (spec-wise) laptop with slow hard drives.

Yeah…when looked at that way having to encode something for 8 hours may actually be fast. Your demanding a lot of things be done to your timeline, and that’s not wrong. Just remember for all those requests you’re adding time onto your final export. Things like color correction is computationally intense, conforming frame rates is very resource intensive, doing a 2-Pass VBR will take longer than a CBR or 1-pass VBR.

So here’s my $0.02 on the subject. When you shoot all cameras should be set at the same frame rate, either 24 or 30. Shooting multiple frame rates is a mistake made by people new to the profession….no problem. Just don’t do it again! 🙂 It’s a pain to fix, 2:3 pulldown footage doesn’t look good IMHO (what you’re doing when you convert 24 to 30), and it adds time to either ingest or export. It’s a simple setting in the camera. Take 5 minutes before you hit record and verify your settings on each camera – frame rate, ISO, shutter speed, and white balance. Every time. Doing this alone will save you time when you hit the computer.

Since you’re uploading to YouTube what’s the benefit of VBR? Variable bit rates exist to attempt to maximize the use of a data to define a scene. So the encoder is smart enough to know that the talking head portion of the video won’t need as many bits to define the picture as your intense car chase scene. So it’ll use less bits during the talking-head portion and more bits during the fast-action car chase scene. Thereby making your overall file size smaller. Smart right?! But these “passes” take time as it has to analyze the scene either 1-time (1-pass) or 2-times (2-pass) before applying the data. But it only matters if you’re concerned with your overall file size. Since you’re uploading to YouTube there really isn’t much need to be concerned. If you have a verified account with Youtube you can upload a 20GB file. For my YouTube uploads I set the encoder to do a CBR (constant bit rate) and forget about it. VBR settings require the encoder to analyze the scene before encoding it. Thereby making it smarter. However, a CBR is in a sense a “dumber” encode whereby it just throws all the bits at the scene indiscriminately. So you’re talking head gets just as much data as your car chase scene. Not the most economical use of data, but if you don’t care how big the ending file is and you want/need a faster encode, then go with a CBR.

As to your other questions – rendering to your external drive is probably better, though they are both slow at 5400RPM. Where does your media live? If it’s on the same drive you’re exporting to, that’ll slow things down further as you’re asking the disk to both read and write at the same time (not a strength of traditional spinning hard drives).

720 vs 1080 won’t greatly change your encoding time, depends more on the the things discussed above.

Rendering straight from PPro vs. handing off to Media Encoder….in my experience the export times are about the same. The difference is that sometimes that handoff, using Dynamic Link, can go sideways on you and freeze your export. Provided neither method hangs, in my experience, I don’t see a huge time savings in one over the other.

Ryan Holmes
http://www.ryanholmes.me
@CutColorPost

[Josh Peterson]

Thanks so much for taking on the wall of text and providing such a thorough answer, Ryan.