Rendering and Exporting and Compressing Question…

[Jeff Pulera]

Hi Tom,

XDCAM-EX (HQ) records at 35Mbps, using 4:2:0 color, and uses Long-GOP MPEG-2 encoding. Long-GOP means that to save space (compress the video more), rather than each frame being encoded individually (Intra-frame), Inter-frame recording is used. There may be just TWO full frames of data recorded per second, and the rest of the frames only record changes from one frame to the next basically. A little more involved than that, bi-directional stuff, but that’s it in a nutshell. But anything with a lot of detail and motion is going to suffer from compression artifacting. Think of a fountain or waterfall for example. Look at the still frames and the video can get very blocky.

So, do you want to take your edited work and create a “Master”, from which other formats will be encoded, and use an inferior codec to create that master?

Not saying your raw footage doesn’t look good, but we’re talking about generational loss, copies of copies – you want to maintain the quality as best you can.

ProRes is an Intra-Frame codec, so each frame is compressed individually, no borrowing from neighboring frames. It’s all there! Color is 4:2:2, not throwing away data. Why does this matter if the source was 4:2:0? Do you add any titles or graphics to your edits? Wouldn’t it be nice to at least keep those at full 4:2:2 fidelity rather than immediately dropping them to 4:2:0?

ProRes is not going to improve the quality of your source, but it will MAINTAIN the quality. ProRes uses 150 or 220Mbps (422 or HQ), but which to use is up to you. It may be overkill to convert HDV for instance to ProRes HQ – it may look no better than using 422 for mastering, depends on the source camera and codec quality.

And yes, it is standard practice to use ProRes as an intermediate codec/master file for video post production. Camera formats are almost always inferior since the camera makers want to increase record times by increasing compression as much as possible.

Thanks

Jeff Pulera
Safe Harbor Computers

[Tom Laughlin]

Jeff,

This is probably the BEST explanation I’ve seen or read on this topic and question for a few years now. I’ve always exported out in a “Same as Source” mentality, I think we adopted from FCP7 in the past, and though there may have been some better or more poor exporting going on in and out of FCP7, the mind-set that I had when I switched to Premiere, was that under the hood, Premiere, I assumed did the same, so, but that’s a different topic related, but what I’m saying is, the quality shift prevention measures, were somewhat handled that way in FCP7, so that’s what I’m saying, we export and match settings, in an effort to maintain quality, but in order to maintain quality, one NLE does this differently than another, which I fear, may have been contributing to poorer exports in the past, but not to drastic, but for some of those who are more anal about maintaining a more accurate and master quality, and from what you just explained, this ProRes as a safety and assurance, MAKES me FEEL so much more SURE of HOW to export and why, so thank you for that response and for everyone’s contributing input as well. I will now make the shift to ProRes masters for future project exports.

Thanks everyone!

Tom Laughlin
Producer/Editor
Digital Chop House
Salt Lake City, Utah
http://www.digitalchophouse.com

[Jeff Pulera]

Hi Tom,

I’ve never been an FCP editor, but my understanding was that pretty much all sources got transcoded at ingest, prior to editing, so sources were already bumped up to a better format. Then I believe there was also some smart-rendering going on, so that not everything in the timeline would get re-encoded upon export, it would just keep unchanged portions “as is” and copy those parts over without recompression, within the QuickTime environment.

So yes, definitely a different workflow with Premiere now, editing the raw camera files. Glad I could help clarify. You’re lucky to have ProRes to work with. PC guys have to install a third-party codec to have something comparable, no ProRes option (without jumping through some hoops at least) for PC editors.

Cheers,

Jeff Pulera
Safe Harbor Computers

[Tom Laughlin]

Jeff,

One last question – for those of us out there who are still editing in the more new and native environment.

With your reference to the QuickTime environment that FCP7 had been settled somewhat around, now into Premiere we go, does it EVEN matter now, to do the same before editing? In other words, is it BETTER to transcode your EX footage to PRO-RES ‘before’ editing, thus editing in a Pro-Res time-line, with a ProRes export? Are there any advantages to this, and are editors doing this? (If you have the time to transcode, that is, is it even beneficial, other than eating up MUCH more hard-drive space?)

In the past, yes, I’ve done transcoding from BPAV to .mov for FCP7 in the past, as you couldn’t edit raw .mp4 in the BPAV, so I’d normally use EXCAM Transfer or Clip Browser, from what I understand, all it did was ‘re-wrap’ the file in an “.mov wrapper” for editing in FCP7, so.

But I don’t recall ever “having to” transcode media via source footage from an EX3 or EX1 to Pro Res before an edit, not sure why…

That being said, yes, I ‘have’ worked with many many many shooters – who did shot on DSLRs (like the Canon 5D Mark II), who routinely would take their source media, and transcode it to “Pro Res 422” or “HQ” (using MPEG StreamClip to transcode RAW cards). So in that DSLR workflow, with the Canon 5D Mark II, editing in FCP7, this “Pro Res transcode before editing” workflow, to me, was more common, BUT not with source media from EX3 or EX1’s… Sorry for the long sentences.

So, it is better to transcode EX3 footage to Pro Res before edit, or not? I don’t see any advantages, with Premiere editing native, why would you want to transcode still?

Above is my last question, promise!!

Tom Laughlin
Producer/Editor
Digital Chop House
Salt Lake City, Utah
http://www.digitalchophouse.com

[Jeff Pulera]

As discussed earlier, transcoding your source clips into something else will never “improve” the quality – all we can do is “maintain” quality – so no, there’s no obvious reason to transcode prior to editing in most cases.

The exception is that some older machines may have issues decoding the newer AVCHD codecs and such, and converting to a “less compressed” format allows for smoother timeline playback. But newer machines should be just fine.

One more exception – I have seen certain kinds of GoPro footage bring Premiere to its knees even on a fast machine. GoPro provides a utility to convert native clips to QuickTime or AVI files and that is one case where I would typically go ahead and transcode prior to editing! Anyway, the utility also can remove the fisheye common to GoPro footage, so that makes it a no-brainer to pre-process most of those clips right away.

Jeff Pulera
Safe Harbor Computers

[Steve Siegel]

I probably shouldn’t waste anyone’s time on this really naïve question, but hopefully someone will take pity.

I’m trying to compare file sizes on my original XDCAM-EX clip, and on it’s Premiere Pro CS 5 (Windows 7) export. I assumed that with every step taken in import and export, some codec compression would have been applied, and the result would have been a smaller, less detailed, file. Not so.

The original file, out of the camera was 132 MB. Importing it gave a file of 43MB, pretty compressed, but what can you do? Then with no alterations whatsoever, I exported with the best quality selectable into H264. It was 442 MB. In PhotoJPEG it was 880. Where is all this extra data coming from? I know that PPro isn’t making my clips more substantial. The extra data must be instructions of some kind? Does file size really have no bearing on the quality of the image?

Thanks for a simple explanation with as little jargon as possible.

[Tero Ahlfors]

[steve siegel] “Where is all this extra data coming from?”

You are giving the encoder too much bitrate for what the footage would really need. Most standard codecs (Prores, XDCAM) have locked bitrates that you can’t change. Compressed formats like H264 have an option to set the bitrate in Mbps/s which is the data limit that the encoder has to work with. More bitrate means better quality and bigger files but smaller resolutions can deal with less bitrate.

[Steve Siegel]

The phenomenon happens even when the bitrate is fixed. Cineform is a good example. I put a 100 MB file into my NLE and export it out as
Cineform and it comes out as a 3.5 GB file. Thirty-five times bigger. What is all the extra stuff…you can’t blame it on bitrate.

[Tero Ahlfors]

[steve siegel] “I put a 100 MB file into my NLE and export it out as
Cineform and it comes out as a 3.5 GB file”

That happens because the set bitrate that Cineform uses is a lot higher than the compressed original file.