Does re-rendering cause a drop in quality?

[Scotness]

Does each time you re-render a file in Premiere drop it's quality slightly due to compression? - I'm using avi files compressed with microsoft dv pal compressor - although it says in the final export settings it is uncompressed - but I think there's a slight drop so have been trying to avoid it as much as I can, but sometimes with a number of filters or plug ins to be applied it is unavoidable -- what do people think?

Scot

[blackangus1]

Absolutely correct. Every render causes an uncompress/recompress cycle. DV is a "lossy" compression scheme (unlike LZW, for example, which is lossless -- but lousy for video). So anytime you invoke a lossy compressor, you will be losing a bit of picture quality. Most DV algorithms can go six or seven times before the degradation becomes noticeable.

[Brian Maier]

I am going to disagree with the degration in quality. This is not analoge. You dont compress and uncompress when you render files. Scottess render as many times as you wish your quality will be the same.

Brian

[blackangus1]

<BLOCKQUOTE><font size=2 face="Verdana, Helvetica, sans-serif">quote:</font><table border="0" width="90%" bgcolor="#333333" cellspacing="1" cellpadding="0"><tr><td width="100%"><table border="0" width="100%" cellspacing="0" cellpadding="2" bgcolor="#FF9900"><tr><td width="100%" bgcolor="#DDDDDD"><font size=2 face="Verdana, Helvetica, sans-serif">Originally posted by Brian Maier:
I am going to disagree with the degration in quality. This is not analoge. You dont compress and uncompress when you render files. Scottess render as many times as you wish your quality will be the same.

Brian</font></td></tr></table></td></tr></table></BLOCKQUOTE>

Hmmm, perhaps some clarification is in order. If you are referring to rendering items on the timeline or something, and NOT KEEPING THE OUTPUT, then yes, there is no cumulative degradation. If, however, you sent the footage through multiple renderings, one on top of the other, there is a definite degradation. Example: let's say you decided to color-correct a clip, so you apply a filter and then render the footage out and discard the original, so you only have the new color-corrected footage. That footage will have gone through an uncompress/modify/recompress cycle. Then, if you decided that this color-corrected clip still wasn't right, and you wanted to maybe boost the contrast a bit and render that out, the new contrast-boosted footage would go through another uncompress/modify/recompress cycle. And on and on. Every time you modify it, you introduce another lossy compression into it. It does add up.

See http://www.adamwilt.com/EBU-DV.html for a thorough discussion, especially section C.4.5. There it quantifies the image degradation that occurs between 1, 4, and 7 generations of compression.

[Matt Pacini]

Scott, I noticed a little of this when I used Premier to edit Lost Tribes.
However, where Premier ABSOLUTELY causes degradation, is in audio.

DO NOT apply any of Premier's audio filters or FX if you want your audio to sound decent.
I ended up exporing to SoundForge for all my audio editing needs (other than MIDI stuff, which I used Cakewalk Pro Audio for), and SoundForge did an excellent job.
Really great program for filmmakers.


Matt pacini

[MovieStuff]

Hmmmm. Well, I think the title of this thread should actually changed to "....cause a NOTICABLE drop in quality?".

I use Premier all the time for both compositing and for audio. I re-render out video clips all the time and I've never noticed a problem with image degeneration. So, while I'm not versed enough in the specifics of HOW the computer is dealing with the files, I can only say that I've gone quite a few generations and have seen no difference in the image quality, visually.

Regarding audio, I often will export a copy of the time line audio as a single wave file which I then insert on a separate version of the project for mixdown purposes. I do this to simplify the audio timeline into "Sound Effects", "Ambience" "Music" and "Voice". Again, I have never noticed any drop in quality, even when applying filters. I can't speak for everyone else's set up but on my Pinnacle Real Time board with Premier 5.1c, it works just great and is pristine from first generation to last on both sound and video.

Roger

[Brian Maier]

Alright i guess i missunderstood what you were geting at. But as Roger said the noticable difference is minimal at best, but if there is a section of video that you think is suffering from degration just take down the changes you made to it, delete that section of timeline and load the original and apply the changes again all at once. While this may be a pain it would work would it not?

Brian

[blackangus1]

Yes, that would work perfectly. It just gets a little slow on the timeline, having to re-render everything, especially multiple layers. But that's the tradeoff: quality vs. re-render time.

As Roger points out, the degradation is not large, depending on the codec used. Most DV codecs can handle 4 generations with basically no visible degradation, and you can usually go seven generations before there's any noticeable effect.

So, in summary, you should generally try to avoid re-rendering when possible, but if you have to, it's no big deal. Just keep in mind that it's not "free", so if you do it a lot the cumulative effect will start to show.

[Scotness]

It's funny you know - the audio was one thing I was feeling good about but the video I felt was droppping -- typically this is what I have to do

1. Super 8 telecined by Rank to miniDv

2. Import miniDV and synch with audio - render to one combined avi for that take FIRST RENDER

3. Construct a scene on the timeline out of the various synched takes - render that out SECOND RENDER

but it's normally not as simple as that

- alot of the tight shots have vignetting (so I have to crop them)

- about 10% have that freaking jitter (which I must say that dynapel software is doing an excellent job of removing - it completely removes it in about two thirds of the cases and minimises it in the really bad ones) but each of those gets croppped automatically after the stabilisation process to create consistent edges

- and some shots are going to need grading/colour correction etc

Now most of these can be done on the timeline as I'm rendering the final version of that scene - but if it's a long jitter thing in a big scene the computer can lock up, and you can't cross fade two clips with the anti-jitter filter on them - so occasionally I have to do a render of a take and then re-import it into the timeline - so some bits are on their THIRD RENDER

SO when I bring all of those edited scenes onto a new time line which is the rough cut of the film I'm loathe to render that again to make one big avi of the film, but rather just play it off the time line to capture it to miniDV tape again

I think the croppping (both directly and from the anitjitter prog) might be what causes alot of the drop in quality - and you can start to see a faint grid pattern - I spose it's the limits of the minDV format -- but I've got a friend who's done about 7 rerenders on a short (which he doesn't need to do - but anyway) and the same grid pattern is showing up - but he has used no cropping etc -- so I think it's pretty well how you summed it up Blackangus - avoid it if you can!

Scot
Here endeth this thread I suppose :-)

[blackangus1]

<BLOCKQUOTE><font size=2 face="Verdana, Helvetica, sans-serif">quote:</font><table border="0" width="90%" bgcolor="#333333" cellspacing="1" cellpadding="0"><tr><td width="100%"><table border="0" width="100%" cellspacing="0" cellpadding="2" bgcolor="#FF9900"><tr><td width="100%" bgcolor="#DDDDDD"><font size=2 face="Verdana, Helvetica, sans-serif">Originally posted by Scotness:

2. Import miniDV and synch with audio - render to one combined avi for that take FIRST RENDER
</font></td></tr></table></td></tr></table></BLOCKQUOTE>
No, this should be lossless. If you haven't touched the image, you're just adding audio, then it should just copy the data directly over and interleave the audio in -- so no loss in quality.

<BLOCKQUOTE><font size=2 face="Verdana, Helvetica, sans-serif">quote:</font><table border="0" width="90%" bgcolor="#333333" cellspacing="1" cellpadding="0"><tr><td width="100%"><table border="0" width="100%" cellspacing="0" cellpadding="2" bgcolor="#FF9900"><tr><td width="100%" bgcolor="#DDDDDD"><font size=2 face="Verdana, Helvetica, sans-serif">Originally posted by Scotness:

3. Construct a scene on the timeline out of the various synched takes - render that out SECOND RENDER
</font></td></tr></table></td></tr></table></BLOCKQUOTE>

Again, no, this shouldn't cause a render. If it's cuts-only, stringing pieces of video after each other, then there's no manipulation of the image and once again no rendering. The only rendering (which would cause loss of quality) would be cross-dissolves, picture-in-picture, something that involves changing the image). So again, no render here either.

<BLOCKQUOTE><font size=2 face="Verdana, Helvetica, sans-serif">quote:</font><table border="0" width="90%" bgcolor="#333333" cellspacing="1" cellpadding="0"><tr><td width="100%"><table border="0" width="100%" cellspacing="0" cellpadding="2" bgcolor="#FF9900"><tr><td width="100%" bgcolor="#DDDDDD"><font size=2 face="Verdana, Helvetica, sans-serif">Originally posted by Scotness:


- alot of the tight shots have vignetting (so I have to crop them)

- about 10% have that freaking jitter

- and some shots are going to need grading/colour correction etc
</font></td></tr></table></td></tr></table></BLOCKQUOTE>
Yes, all of those will involve rendering, which necessitates a recompression.

<BLOCKQUOTE><font size=2 face="Verdana, Helvetica, sans-serif">quote:</font><table border="0" width="90%" bgcolor="#333333" cellspacing="1" cellpadding="0"><tr><td width="100%"><table border="0" width="100%" cellspacing="0" cellpadding="2" bgcolor="#FF9900"><tr><td width="100%" bgcolor="#DDDDDD"><font size=2 face="Verdana, Helvetica, sans-serif">Originally posted by Scotness:


SO when I bring all of those edited scenes onto a new time line which is the rough cut of the film I'm loathe to render that again to make one big avi of the film, but rather just play it off the time line to capture it to miniDV tape again
</font></td></tr></table></td></tr></table></BLOCKQUOTE>

Don't fret, creating a new file off the timeline (that doesn't involve manipulation of the picture) should be a simple frame copy, not a render. No loss of quality here.

Copying frames (whether from tape to the hard disk, from file to file, or from file to tape) is lossless, and that's the beauty of DV. The only part you have to worry about is stuff that involves uncompressing and recompressing.