Converting H.264 Files Before Editing

[wcn_dave]

My camera produces HD avchd files using h264 and after editing I then render out files for YouTube which use the same codec.

Scrubbing and editing has been painfully slow and I've started using proxy files instead. Those files use QuickTime and the gopro cineform format.

They are super fast when editing. In fact I use the same resolution for those files.  I Believe It's all down to the native codec I'm using having a lot of compression.

R Neil Haugen also mentioned that h.264 used to be often transcoded before use.

This led me to think why don't I just convert all the files into a new format before I start my project.

If I do that what's the best tool to use and the best format to use to avoid any loss of quality?

I could create the previews in same res like now, but I am not sure if I am losing information.

I would prefer some quick batch process like handbrake, but the same question arises.

If I did do this, I'd discard the original files, and only keep the newly converted, as otherwise I'd run out of space.

The other benefit to conversion would be that I could go Ryzen with my next upgrade, however I'd have to be sure it's a lossless conversion.

You're going to tell me Google is my friend, however I am also curious about a workflow that suite video editors well.

Thanks!

[Kevin-Monahan]

Hi WCN_Dave,

The first part really helps, as I see I can have it auto convert on import, and also on drag into project files.

Glad it helped!

The second part sounds like export is fast, as your previews use the same format, however, the link doesn't work

OK, I just fixed it. Yes, your previews have to be in the same format in order for this workflow to function properly. However, if you do not have any render files attached to a clip, smart rendered transcodes with zero effects work the same way. I hope that is clear.

That's why a hybrid "smart rendering" workflow exists where you edit with proxies (or the camera originals at a very low resolution) and at the very end, change back to the H.264 masters and render the entire timeline to ProRes, Cineform, or DNxHD/HR—then export.

I generally want to export either for home video use, to display at events, or for YouTube, so I've always exported in H.264, however that won't work with smart rendering as it's the slow format.

Right, H.264 doesn't work with smart rendering, but wouldn't that be wonderful if it did? Perhaps in the future. That said, who wants to edit with H.264 anyway, not me!

In any case, at least now I can edit with ease!

Awesome.

Regards,
Kevin

[wcn_dave]

So, I've not managed to get this to work.  I tried with both prores and gopro cineform.

In both cases the render sequence time, and the output time are exactly the same, so there's no copying, it's doing it again.

Also, the output file is the exact same size as the transcoded footage, so it _could_ have copied (I was doing with no effects).

The transcoded file is slightly smaller, as there's no audio in it, assume that's correct.

There's more details in my post towards the end.

Re: Converting H.264 Files Before Editing

[wcn_dave]

I have figured out what was going on.

One gotcha I encountered with transcodes combined with smart rendering was that the time to render timeline and render output file was the same.  So I thought smart rendering was not working. After all, it's not doing any encoding, so should be faster right?  What was actually occurring I believe, was that given the transcoded file was also in the same format, the timeline rendering was also not encoding, so basically the timeline was "smart rendering" too... each step was doing the same thing.

That is an excerpt from a blog page I created to document everything that I did, as I wanted to consolidate and organise all my findings and thoughts.

The page is here, and I'd be happy to receive feedback.

[JuanNavedo2080]

I know this is an old thread but my goodness is all this an eye opener. I didn't know H.264 was so "damaging" to Pr workflow. I too got a sweet editing rig and was wondering why it was still hard to edit/playback footage from my A7riii. It seems transcoding is the way to go. Here's the thing...

I am a stickler for wanting to work with the highest quality video, so when I transcode, which one stays truer to the original file without causing the, what I'm now calling, "H.264 Lag", ProRes or DNx?

Also, would it be better to transcode and work with those files as if they came from the camera, on their own without attaching them to the originals or should I attached them to the originals? 

[R Neil Haugen]

It depends on the data amount/rate of your original files.

For most 8/10 bit H.264 out of DSLR and mirrorless cameras, the ProRes LT will be fine. The Cineform codec is also wondrous to work with.

Use MediaInfo to find the data rate ... install that free utility, drag/drop a clip onto that app's icon on your desktop, go into that app's Tree view, and check the Mbps. Do this with both the original media and a couple options for t-coding. Make sure your data-rate is at/above the original.

Neil

[rodneyb56060189]

So complicated. I'm starting to lose track. If you record via atomos via hdmi or sdi those ARE your original files. There's no SD card getting the lower H264 files.. it's ONLY on atomos SSD. So, that's your original.

??

that means that you got the stuff without a transcode or proxy or whatever you want to call it... it's the source.

After THAT, once you move stuff from recorder to internal HD ( or SSD, whatever ) it's just the usual stuff. Back it up if you want, on another drive or whatever. Who cares. I don't worry about backing up anything until I know it's worth backing up, and to be honest most of what I've been shooting or editing lately is basically useless. No client, no big deal if I lost the source. I'm an idiot, so you'd have to be even dumber than me to give me a project to edit.

[rodneyb56060189]

am watching 'westside story' on broadcast TV … is 4:3.. black bars on sides ( on a 16:9 TV). Basically the TV broadcaster is buying right to show movie from some entity who already scanned and panned and reduced movie stuff to TV stuff ( NTSC) at 4:3...

The world is the culture we deserve, as Barzun would say.

[rodneyb56060189]

made me cry

[wcn_dave]

I've failed to get smart rendering to work.  I am also in a position where I have to make a 2 second tweak, and then wait for the whole file to render out, so smart rendering looks good, even if the final file is large.

I have screenshots, however won't upload them unless needed, hopefully my description and numbers will make sense.

I have an MTS file, 1920x1080 50fps which is 1.03mins and 212MB

It transcoded using "Match source - GoPro CineForm YUV 10bit" as ingest setting to a 2.8GB mov file in 38 seconds

I set sequence settings to custom, and saw two choices.  Format: GoPro, codec: Quicktime, or Format Quicktime and codec: gopro.  I tried both, one created a 2.8GB preview file, slightly smaller, however with no audio.  The other created a 9GB file, so decided that was not what I needed.  The 2.8 file took 38 seconds again.

(There are no effects anywhere)

Exported as Quicktime, gopro, and then did "use previews", the export took 38 seconds, it did therefore not seem to use the preview.

I exported as quicktime,  preset "match source", and it took 38 seconds

I exported as quicktime,  preset "match source", and then selected "use previews", it took 38 seconds....

Doesn't seem that any of the options actually used the preview file, as the render to create with or without preview file is 38 seconds, and the time to created the preview is also 38 seconds....

Can't see what I'm doing wrong!!

[MyerPj]

Not sure if you've seen Kevin-Monahan​'s blog post regards smart rendering, it could be helpful:

Smart Rendering in Premiere Pro CS6 (6.0.1, and later), & Premiere Pro CC

Also this thread:

FAQ: How do I speed up rendering, exporting, or encoding?

[wcn_dave]

Yes, I followed his page to do it.