Log in

View Full Version : HC bitrate allocation :prog-interlaced


mikenadia
3rd March 2008, 15:55
I would like to encode, with HC, one progressive segment (bitrate A at CQ1) and one interlaced segment (bitrate B at CQ1) under total filesize constraint . I would like to make an educated guess regarding bitrate allocation among the two segments, knowing that interlaced segments require a higher bitrate for a given CQ and for a given visual quality (by how much?) .Both segments are part of a commercial DVD (Would like to know if interlaced segments from DV have to be treated differently).
Thx

Edit: Assuming segments 1 and 2 have the same number of frames, is there any relationship between SSIM(1+2), SSIM(1) and SSIM(2)?
It is my undertanding that PNSR(1+2)=WAVG(PNSR(1):PNSR(2)) regardless if 1 and/or 2 are progressive or interlaced.: If someone could confirm or infirm it.
If i have no other way, could I make the above-mentionned allocation on a PNSR or SSIM basis.Thx

manono
4th March 2008, 00:02
Hi-

When I have a problem like this I'll run a 1-pass constant quant encode on them. From that I can figure the percentage of the final size to give to each, the final size of each, and from that (if needed) the bitrate to assign to each.

I don't know anything about your edit question.

mikenadia
4th March 2008, 00:55
Thx for replying. Can I deduct from your answer that your allocation will be the same if the two segments were progressive. The question I am wondering is because the bitrate of the interlaced segment for a given CQ will be higher (compared to the same source if it were progressive) if I should not give its bitrate a smaller weight (and if yes, how to define it) to have the best combined visual quality.The purpose is to allocate the bitrates of Extras (some progressive, some interlaced) in a DVD. Should I take the same Q for interlaced and progressive segments to make the allocation.Thx

manono
4th March 2008, 04:38
Hi,
Can I deduct from your answer that your allocation will be the same if the two segments were progressive.
No, not unless they're the same video. It's for 2 different progressive sources that I use the method I outlined so I can put them on the DVD with roughly the same quality. Different sources, even if both are progressive, compress differently. But I may have slightly misunderstood your question.
...if I should not give its bitrate a smaller weight
I don't know. I just told you how I'd do it. If you wish to give the interlaced one lower quality than the progressive one, that's up to you.
The purpose is to allocate the bitrates of Extras (some progressive, some interlaced) in a DVD.
Oh, it's for extras. I give them a defined Q (I use 35 in CCE - I don't know how that translates to HCEnc usage) with a lower bitrate quantisation matrix. Interlaced or progressive, it's all the same to me. And I won't reencode using 2-passes for a defined size. I'll keep the 1-pass CQ encodes for those extras. The final sizes will become whatever it takes for the quality I define. As long as I can encode the main movie to the quality I want, that's the main thing. I may also do a 1-pass CQ encode of the movie at the same time to see how the final sizes turn out

mikenadia
4th March 2008, 13:19
Thx a lot. That is (specially the last paragraph) all I wanted to know and I like the answer.
Sorry for not getting clearer in the first post, but I was told to treat interlaced and progressive separately and therefore was confused on how to handle the issue. Not anymore. Thx again.

manono
5th March 2008, 00:45
Well, it's an interesting question, and my method, while it works fine for me, is by no means the definitive answer. How do you decide what extras can be kept with an acceptable minimum quality, versus the movie with also a minimum (but much higher) acceptable quality?

I all the time see statements like, "I make all my extras for a 2000 bitrate". But statements like that betray a complete lack of understanding of even the basics of video compression (and/or no care at all for quality). Given the same quality output for my extras, I might have a talking head intro look fine at or below 1500, but an old grainy cartoon or an action packed deleted scene might need 4000 or more for the same quality. As you know, an interlaced extra will need a much higher bitrate than an equivalent but progressive extra for the same quality (up to 40-50% higher in some cases). If I find that the extras and movie done for these minimum qualities go over the DVD5 size, I might try again with the extras done for a 45 or 50 Q in CCE, but it may just mean that I have to start removing stuff. This method is a good one for helping to decide if I can even include everything I might want to include, given my particular standards.