VP6 Codec Released! (was: this one is for doom9)

[slavickas]

Originally posted by On2Tech
Glad to hear it. We'll have a new version shortly. I think you'll find its two pass mode "greatly" improved.

will there be improvements in low-bitrate and(or) single-pass encodings? btw shouldn't be badframes.stt disabled in release build?
i don't think it's very useable for me :rolleyes:

[On2Tech]

Originally posted by slavickas
will there be improvements in low-bitrate and(or) single-pass encodings? btw shouldn't be badframes.stt disabled in release build?
i don't think it's very useable for me :rolleyes:

If you are getting badframes.stt it is indicative of a bitstream bug please private message me with the file / settings that cause it.

[SeeMoreDigital]

Originally posted by On2Tech
Glad to hear it. We'll have a new version shortly. I think you'll find its two pass mode "greatly" improved. Cool!

There's no doubt that the VP codec is able to generate excellent results with low bitrate 1pass encodes. If this quality can be improved further using 2pass encoding, then it's well worth waiting for!

Thanks you guys

[slavickas]

Originally posted by On2Tech
If you are getting badframes.stt it is indicative of a bitstream bug please private message me with the file / settings that cause it.

don't worry, i get it only (sometimes) when watching inet tv in winamp, i think it happens when bitstream is decoded not from keyframe

[h9903209]

I just download the code and want to try encode in virtualdub, but find there are two vp6, one is vp60r simple profile, another one is vp61r advanced profile, but the configure seems the same, just wondering what's the difference between these two?

also, do you mind explain a little bit more about the adjust quantizer? still don't quite understand how to choose the value even after reading the guide...

[On2Tech]

Originally posted by h9903209
I just download the code and want to try encode in virtualdub, but find there are two vp6, one is vp60r simple profile, another one is vp61r advanced profile, but the configure seems the same, just wondering what's the difference between these two?

also, do you mind explain a little bit more about the adjust quantizer? still don't quite understand how to choose the value even after reading the guide...


VP60 is effectively a simplified decoder version of VP61. It allows for faster playback on slower machines or higher def playback on faster machines. We use it to show hi-def on various set-top box dsps. Don't use VP60 on the PC unless you really can't play back fast enough. The quality isn't too much worse but it is a little bit worse.

The quantizer parameters allow you to control how the best possible and worst possible frames will look:

- adjust quantizer allows the compressor to change the quantizer ( or roughly speaking quality ) for the frames it encodes to try and hit the datarate you requested. If you don't check this setting each frame has the same quantizer ( the max quantizer value ).

- max quantizer controls how bad the worst and the best quality frames can be. The lower the max quantizer the better the worst frame will be.

- min quantizer controls how good the best frame will be. The higher the value the worse the best frame will be. ( this might seem like a silly parameter, but with the best quantizer set at 0 you might spend thousands of bits on what looks like a black frame but actually has slight variations of black).

[Sagittaire]

For high compressibility or/and high BPF (bit/pixel*fps) use [2-30] for interval quant ...

One2 Tech are agree with that i think ... ;-)

[On2Tech]

Originally posted by Sagittaire
For high compressibility or/and high BPF (bit/pixel*fps) use [2-30] for interval quant ...

One2 Tech are agree with that i think ... ;-)

I agree that in practice 2 - 30 will work just fine unless you're shooting for ridiculously low datarates.

In 2 pass mode the codec will automatically pick a max quantizer. As such 2 - 56 hopefully will work just as well as 2 - 30. And 2-30 on extremely difficult material will limit the lowest datarate you can hit.

For instance try shooting for 200 kilobits on that Harry Potter clip with a max quantizer of 30.. Of course I don't think you'll really like the results of your 200 kilobit encode but it should actually work.

[h9903209]

thx for explanation, because used to rmvb super simple setting, don't quite understand many technical terms...

In the guide for adjust quantizer, it says "Can be set lower if you prefer better quality frames and don’t mind dropping to a lower frame rate" what's bad if dropping to a lower frame rate? less smoothing? and if uncheck temporal resampling then would it still drop frame rate?

"For high compressibility or/and high BPF (bit/pixel*fps) use [2-30] for interval quant ..."
thx for your help, but I even don't quite understand what means by high compressibility or/and high BPF (bit/pixel*fps)... high compressibility/BPF = high datarate?

I usually encode using rmvb at one pass 450kbs vbr normal EHQ before, is that ok to set 2-30 in vp6 450kbs encode or need any change in order to achieve >= rmvb 450kbs quality? or should I use 4-40 as described in the example in the guide for 500kbs encode? in fact I don't quite care about whether the resulting datarate match the entered value, it's ok even it's higher.

also, I see you disable temporal & spatial resampling while in the guide it's enabled... just wondering whether the only effect of disabling them is better quality and the resulting datarate would be a bit higher than the entered value?

[Sagittaire]

For me High compressibility is 60% /quant2 encoding and high BPF are 0.25 for advanced codec ...

450 Kbps for usual encoding in 640*272*25 is weak (450000/640*272*25=0.1 BPF) and VP6 default interval quant is good ...

temporal & spatial resampling for VP6 are activing if the encoding condition are very hard I think (read VP6 doc for more precisions ...)