What sort of quality loss does re rendering result in?

[Norman Black]

There is no real answer except that there is always loss when using a lossy compression algorithm. How much is the question, but that cannot be answered as that depends on the source material, the encoder used and what options were used for the encoder. By encoder I don’t mean mpeg-2 verses mpeg-4. I mean the actual encoder implementation. Like one AVC encoder versus another.

Loss effects are not necessarily linearly accumulative. When you encode to a lower level than source, you lose detail. This allows the lower bitrate of the lesser encode. Encoding again is not likely to lose the same amount of detail as the first generation because the encode options are a better fit for the material that is about to be encoded. That said you might think something can stabilize, but the encoders do not make the exact same compression decisions, regarding macroblocks, each time you do an encode and this will continue to soften the image.

If you intend to do generational renders then use an intermediate codec and/or keep your bitrates high. Intermediate codecs are best for this.

Do a test yourself. Your eyes are the best judge out there.

Then remember that pixel peeping a frame is different than watching video. Easier to notice differences.

[Steve Rhoden]

Thanks Norman, was gonna go into full details, but you covered it
well.

Steve Rhoden
(Cow Leader)
Film Editor & Compositor.
Filmex Creative Media.
https://www.facebook.com/FilmexCreativeMedia
1-876-461-9019

[Scott Simpson]

It’s like making photocopies of photocopies. Even on an excellent photocopy, if you make copies of copies for long enough, the artifacts are going to add up.

Obviously you have to render at least once, right? If you want anyone to see your product, it has to be rendered.

If you’re talking about pre-rendering elements as intermediate steps, sure, there are good reasons to do that from time to time. If you can use original source material on your time line instead of intermediates, you’ll get better quality.

But there are alternatives. Rendering to an uncompressed format (leaving aside super-techy things like color space conversions) will be roughly perfect.

What you need to watch out for is lossy compression. Just like converting a wav file to mp3 throws out a lot of information to make a smaller file, rendering a video in mpeg2 or AVC throws away information so you can have a smaller file.

So, you’re always negotiating a tradeoff: if you want the detail left intact, you’ll have a bigger file. If you need a smaller file, you’ll eventually end up with artifacts and lost data.

If you have the HD space, you can render to a lossless format or even a high-bitrate compressed format. If you can live with a little quality loss, you can compress a little further.

What’s the quality loss look like? Maybe some fuzziness on sharp edges, maybe some noise, maybe blocky parts. Sometimes that stuff matters. Sometimes it’s not such a big deal.

For me, if I’m pre-rendering something that’s really important for the project, I’ve been rendering it to DNxHD or Quicktime Animation. For slightly less mission-critical things, I think I’ve used Cineform. For no-big-deal stuff, I’d squashed it down to MP4 and not worried about it.

Hope that’s some help.
—
edit: re-reading your question: Re-rending something that’s already been compressed — eg. editing a DVD rip or, worse, something grabbed from YouTube — will be more obvious. The old computer truism is “garbage in, garbage out.” With compression, it becomes “garbage in, stinky garbage out.” You can try to minimize damage by re-rendering to a high-bitrate format, but you’ll never get those lost details back.

—
Radio guy in a TV world. Bigasssuperstar.com

[Norman Black]

I did a little quickie test of my own. I took a 20 seconds of 35Mbps AVC 1080p30 GoPro Hero 3 Black clip with lots of super fine detail. Leaves, twigs, sandy surface.

I rendered 4 times with Sony AVC at its max bitrate, about 26Mbps, high profile. The fourth generation was not all that good. A fair amount of softening.

I did the same with Mainconcept AVC, one pass variable bitrate at 35Mbps average and 45 max, high profile. The fourth generation was nearly as good as the original. You could not tell without pixel peeping at 100% and flicking back and forth between the two samples.

To be fair to Sony AVC since it cannot use the high bitrates of my source, I tried Mainconcept at 25Mbps average and 35 max. The fourth generation was really very good. I still needed static pixel peeping to tell but there was more loss.

This is an example from my statement about so many factors involved including the specific encoder use. Sony AVC did not do as well in this circumstance.

Intermediate codecs should be visually lossless over generations. For intermediates in Vegas without installing anything you have HDCAM SR variants and XAVC Intra.