CQ Quality Setting

[Greyhair]

I've discovered something, I think, that may be quite helpful in TMPGEnc. Maybe other's already know this and this isn't news, but I've not seen it anywhere else, so what the heck, here it is.

I think the 2 pass vbr offers slightly higher quality than CQ, but I don't like how long it takes. OTOH, CQ gives me good enough quality but I can never know the ultimate files size. I think I've figured out a way to determine ultimate file size using CQ.

Here's what I'm doing. Let's assume I want a max bitrate of 8000, and a min. bitrate of 2000. Next, let's assume I want an average bitrate of 2700 (it's a really long movie). If you look at the spread between min and max, which is 6000, and divide that into the desired average bitrate you get .45 or 45%. If you then setting the CQ settings to the above max/min, and set the slider at "45", the file seems to come out to around 4.3 gigs.

I've only tried this a time or two, but so far so good. Just something to try.

[banjazzer]

I would not agree that TMPG's CQ is inferior to it's 2-pass VBR. If anything, I would suggest the opposite. However, used properly both modes give excellent results.

I also think you have not discovered a way to predict filesize. The way that CQ works means it is entirely dependant on the nature of the video to be encoded. Try encoding a predominantly dark sample with little movement, and compare with the same-sized sample with daylight scenes and plenty of movement! Then compare filesize.

Also, I don't know whether the figures you used were purely for illustration, but I presume you must be using one of the DVD templates? I can't speak for DVD, but with SVCD I would hesitate to use any quality setting less than 75. I would suspect it would be similar for DVD.

[Greyhair]

This is not meant to be a cq vs. vbr discussion......yet again.

Yes, I am using a dvd template. But the above numbers are just for illustration. Yes, I'm aware that final exact file size is dependent on the video itself. But, isn't it possible such a formula puts you close enough in the ballpark to be able to predictably use cq?

I'm postulating that the cq slider is just another way of setting average bitrate but based on the min/max's you enter, as opposed to some absolute quality indicator.

FWIW

[banjazzer]

I've been working with CQ for a while now, and I don't think there is any reliable way to estimate filesize, short of doing reasonable-sized test encodes, and just hoping that the sample is typical of the movie itself. I do 2x10 minute samples at quality 75 and 85, and work out what the final filesize is likely to be by scaling up the sizes of the 2 samples. Hopefully I can choose somewhere between 75 and 85. If it looks like being lower, then I think about going for an extra disc.

I don't see how your method can take any account of the compressibility of the source material, which is why using CQ is so unpredictable. You will get the same setting for a high-motion, difficult-to-compress movie as you would for something like Eraserhead (the most compressible movie I have ever seen) 8) And if I could predict the filesize (which I don't think you can for the reasons I have stated) I would not be prepared to encode to a quality setting less than 75 anyway. 8)