Megui quality settings, need some clarification

[llmercll]

I'm wondering what is the best way to acheive quality though megui.


it seems like its between automated 2pass bitrate mode and quantizer mode.

if i set the quantizer to about 26 it comes out to around the same size as a automated 2pass 700 bitrate file.

does this mean the quality is the same? or will the quantizer always come out on top?

i ask because im having trouble telling a difference, it's very hard and i cant seem to make up my mind so id like to ask what you guys would do.

the main things i will be dealing with is converting my xvid's to x264 for saving space, and the occasional higher quality dvd convert. I won't be dealing with blu ray too much. I just dont know which one is better at similar filesizes, and i really cant tell, but sure one is better than the other, it's just difficult to watch two movies at a time.

what im most interest in is a way to half the size of xvids without losing any quality. if someone can tell me how to do that id be ecstatic.

[manono]

if i set the quantizer to about 26 it comes out to around the same size as a automated 2pass 700 bitrate file.

That particular video will have similar quality, whether done by bitrate or by quantizer, all other settings being the same. It's not true that other videos done for the same quality will have the same filesize, or done for the same size will have the same quality. There's not much correlation between the 2. Different videos compress differently.

what im most interest in is a way to half the size of xvids without losing any quality.

You're joking, right? No matter how good x264 is, you'll still lose quality. It can't perform miracles.

[llmercll]

if i set the quantizer to about 26 it comes out to around the same size as a automated 2pass 700 bitrate file.

That particular video will have similar quality, whether done by bitrate or by quantizer, all other settings being the same. It's not true that other videos done for the same quality will have the same filesize, or done for the same size will have the same quality. There's not much correlation between the 2. Different videos compress differently.

what im most interest in is a way to half the size of xvids without losing any quality.

You're joking, right? No matter how good x264 is, you'll still lose quality. It can't perform miracles.

I can only wish right? =)

I'm not sure i understood your post. i know that quanitzer will not always = quality/size compared to bitrate, but if were to find the quantizer setting that would produce very similar filesize to the "bitrate" i want, would going with the bitrate produce better quality or would the quantizer produce better quality?

or were you saying, even in the case above, it's too general of a thing to ask?


I encoded an xvid at 1000kbps automated 2pass and couldn't tell any quality loss at all. it actually looked like it was smoother with less pixelation, don't' know if that's possible though, i doubt it. only thing is its not much smaller than the source xvid lol

[manono]

...but if were to find the quantizer setting that would produce very similar filesize to the "bitrate" i want, would going with the bitrate produce better quality or would the quantizer produce better quality?

Many people (myself included) believe that a quantizer based encode will produce better quality than a 2-pass encode for the same size. Why? because the quality is more even or very similar over the entire video, where a 2-pass bitrate encode will generally give a static scene better quality than it will a motion or action scene. The quants jump higher during the complex scenes. That kind of rate control can be justified by saying that the eye can't judge the quality of a motion scene as well as it can judge the quality of a static scene (unless you pause and study the video frame-by-frame). While true, I prefer to know that I'm getting even quality over the entire video. It's a matter of personal preference.

However, as I said in my previous reply, giving one video the bitrate to make it the same size as a quant-based encode for the same size doesn't mean that you can give the next video the same bitrate and expect the quality to be the same or similar to that of the first video. Again, different videos compress differentlyy. That second video may have lower quality than the first one; it may have higher quality.

The upshot is that if size (within reason) doesn't matter much, use quantizer based encodes. Find a quant that gives you the quality you like, and use that. The resulting sizes, even for videos of the same length, resolution, and other similar settings, can, and often do, wind up all over the place. And, of course, you get results much more quickly because you're only running the one pass. If you want the videos to all have the same size, use bitrate or size based 2-pass encoding. But the resulting quality might wind up all over the place.

[guns1inger]

I have a regular 30 minute TV show that I capture and re-encode for streaming. I use quality based encoding. This is a panel type game show, so mostly talking heads, very little action. The encoded varied in size across the course of a season between 480MB and 640MB, all based on the same quality value. This was using Divx and AutoGK.

Top Gear, using the same settings per episode in Xvid4PSP and H264 encoding has less variance, but still ranges from 540MB to over 600MB. That is over 10% variance.

[llmercll]

...but if were to find the quantizer setting that would produce very similar filesize to the "bitrate" i want, would going with the bitrate produce better quality or would the quantizer produce better quality?

Many people (myself included) believe that a quantizer based encode will produce better quality than a 2-pass encode for the same size. Why? because the quality is more even or very similar over the entire video, where a 2-pass bitrate encode will generally give a static scene better quality than it will a motion or action scene. The quants jump higher during the complex scenes. That kind of rate control can be justified by saying that the eye can't judge the quality of a motion scene as well as it can judge the quality of a static scene (unless you pause and study the video frame-by-frame). While true, I prefer to know that I'm getting even quality over the entire video. It's a matter of personal preference.

However, as I said in my previous reply, giving one video the bitrate to make it the same size as a quant-based encode for the same size doesn't mean that you can give the next video the same bitrate and expect the quality to be the same or similar to that of the first video. Again, different videos compress differentlyy. That second video may have lower quality than the first one; it may have higher quality.

The upshot is that if size (within reason) doesn't matter much, use quantizer based encodes. Find a quant that gives you the quality you like, and use that. The resulting sizes, even for videos of the same length, resolution, and other similar settings, can, and often do, wind up all over the place. And, of course, you get results much more quickly because you're only running the one pass. If you want the videos to all have the same size, use bitrate or size based 2-pass encoding. But the resulting quality might wind up all over the place.

thank you for this, it was very helpful. i too, like quantizer based, it always seems smoother and crisper to me, although it very well may all be in my head.

anyway, mind me asking what you use for certain sources? i was thinking between 13/16, i really cant tell any quality loss when i use this. Def 15/16 for dvds, and more like 13/14 for HD. the file sizes are an easy few gigs, but in this day and age i don't give a flip.

[manono]

Can't help there as I do next to no x264 encoding. Many say that below 20 or so you get more-or-less transparent encodes - results not materially different from the source.