quality difference for multiple reincodes

[resloan]

Would there be any difference in quality of a video that was reencoded several times compared to a video reencoded once if both end up with the same bitrate, size, resolution? For example, I have a movie that was ripped and compressed with DVDShrink to about half its original size. Later I use AGK on it and compress it to XVID to about half that. Not considering the extra time needed, would there be any difference in quality If I had just used AGK on the original and compressed it to the same size, bitrate, etc?

[poisondeathray]

Would there be any difference in quality of a video that was reencoded several times compared to a video reencoded once if both end up with the same bitrate, size, resolution? For example, I have a movie that was ripped and compressed with DVDShrink to about half its original size. Later I use AGK on it and compress it to XVID to about half that. Not considering the extra time needed, would there be any difference in quality If I had just used AGK on the original and compressed it to the same size, bitrate, etc?

yes, using the original vs. the dvd shrunk version as input into agk will give better results (less quality loss)

[resloan]

Thanks for your reply. I take it that the quality, even though the specs are the same on both versions, is relative to the source. Also I am curious as to the bit/pixel ratio which is referred to as a "quality" factor. How reliable is this? Is there any way to predict what the perceptual quality of a movie will be, ie, artifacts, jerkiness, blockiness, etc?

[jagabo]

Also I am curious as to the bit/pixel ratio which is referred to as a "quality" factor. How reliable is this?

Not very reliable. Look at the QF in these samples:
https://forum.videohelp.com/threads/295672-A-problem-for-video-experts?p=1811057&viewfu...=1#post1811057

[resloan]

I see. All the QF would mean in these cases is how close you are getting to the quality of the source. Most of my sources are commercial DVDs, which are good quality. I use DVDShrink when I want to keep the menus, and XVID for just the movies. Usually I can run a video thru shrink a couple times with maximum compression and still looks decent. - Shrink on repeated use appears to just lower the bitrate. I ran across a chart on the net that showed quality vs bitrate for several resolutions ( http://www.mediachance.com/dvdlab/tutorial/bitrate.html) for MPEG. Is there anything like this for other formats like XVID, etc?

[jagabo]

I see. All the QF would mean in these cases is how close you are getting to the quality of the source.

Only if the two videos came from the same source, were treated the same way, and were encoded with the same codec and similar settings. If you DVD Shrink a movie in half, then encode with h.264 to shrink that in half again, it will look worse than if you started with the original DVD and compressed to 1/4 (the same final size) with h.264.

Usually I can run a video thru shrink a couple times with maximum compression and still looks decent. - Shrink on repeated use appears to just lower the bitrate.

That is one of the worst ways of reducing the bitrate. If you enlarge the frames (watch on a big screen TV for example) you will see a lot of blocky artifacts.

I ran across a chart on the net that showed quality vs bitrate for several resolutions ( http://www.mediachance.com/dvdlab/tutorial/bitrate.html) for MPEG.

That type of chart should only be used as a very rough guide. Many videos will fall well outside the graph.

Is there anything like this for other formats like XVID, etc?

I've never seen a chart, but very rougly, you can cut the bitrate in half with Divx/Xvid, to 1/3 or 1/4 with h.264.

[tomlee59]

Would there be any difference in quality of a video that was reencoded several times compared to a video reencoded once if both end up with the same bitrate, size, resolution? For example, I have a movie that was ripped and compressed with DVDShrink to about half its original size. Later I use AGK on it and compress it to XVID to about half that. Not considering the extra time needed, would there be any difference in quality If I had just used AGK on the original and compressed it to the same size, bitrate, etc?

As the other replies have said, the process is as important as the target size, bitrate, and so forth. Each transcode operation is lossy, so the more operations you perform, the greater the loss. It's not only a function of the bitrate. It's a bit analogous to serial photocopying. A copy of a copy of a copy will look worse than a copy of a copy, even though the optical resolution, etc. was kept constant throughout.

So the general rule is: Avoid performing more than one conversion, if preserving the maximum quality is important to you.

[resloan]

Thanks all for your responses, I am enlightened.

[resloan]

One other quick question: how do "quantizer" values relate to CQ percentage? In other posts, I've seen quantizer numbers referred to, but In AGK a percentage slider is used.

[poisondeathray]

One other quick question: how do "quantizer" values relate to CQ percentage? In other posts, I've seen quantizer numbers referred to, but In AGK a percentage slider is used.

Which program are you referring to in terms of quantizer, CQ ?

[resloan]

I'm not sure now, I just read a post where Jagabo I believe was saying he used quantizer 3 for watchable video, and 2 for more important things. I took it that this was referring to video quality in general in single pass mode.

[manono]

4.8 In the Quality 1pass, what are the correspondances between quality percentage and DRF?
- See table below for average results; due to B-frames usage, it is difficult to get accurate results between 80 and 100%

2.5   = 80 %
2.67  = 75 %
2.99  = 67 %
3.51  = 57 %
4.0   = 50 %

http://forum.doom9.org/showthread.php?t=72679

And this is a very rough guideline. Also, it applies much more to XviD than it does to modern DivX codecs.

[resloan]

DRF? compression factor?

[manono]

DRF stands for Detail Removal Factor. It's analogous to the Quant. The higher the DRF or Quantizer, the more detail is removed. The default 75% in AutoGK should give you a very good result.

[resloan]

Many thanks!