Juste a little question : is payback delay active when alt-cc is used ?

[HomiE FR]

Sorry, but I couldn't find the answer to this question when looking at the old threads about alt-cc (like "use alt-cc!" by Teegedeck and Foxer).

I try to understand how the alt-cc works, and I think that everything is ok except the question about payback delay : is it activated when alt-cc is used or not ?

Thanks.
And if this is the case, I personally think that developpers should put it out of the curve compression tab (even though it's closely related to it) because it makes confusion when we enable alt-cc. Sorry if the answer is known by everybody else on the board.

[Sirber]

For me alt-cc doesn't add delay. alt-cc is just a technique for bitrate calculation. b-frames add delay.

[unplugged]

AFAIK when alt-CC is used, all standard-CC params are discarded.

@Sirber: you misunderstood the post :)

[HomiE FR]

unplugged : I'm not totally sure that it's so simple. First I thought like you do but I changed my mind.
I'll try to show you that there is a problem if there is no use of payback delay when alt-cc is used (which equals to a payback delay set at infinite). If I choose Low distance at 0 (which means average frame size) and High distance at 2000 for instance (over the biggest frame) I think that there will be an oversize! Indeed, the frames below the average size will be used with the highest possible quality I think (quant = 2) and the minimum relative quality will not be used if the High distance is set so high. With a Low agression the codec will have a hard time trying to find bits to compensate the bits taken for the small size, if it can't take more bits from the big frames.
So unless there is (which is possible but I never read it) an internal process which prevents this situation from happening or compensates the bits used for the small frames by applying a stronger curve aggression than the one that is set in the tab, there will be a problem without a payback delay.

With an active payback delay, we can easily imagine that when the codec finds bits (positive or "negative") that are staying too long in the buffer, it just takes bits on every frame to compensate the lack of bits and avoid the oversize. It could take the control over the alt-curve and choose by itself the quants that need to be used, for instance using higher quants for small frames.

I'm really not sure that it works this way, but I would like to understand this part of the alt curve compression. Sorry if this seems obvious, but for me it's not!

Thanks for the answers!

[Sirber]

Sorry guys I read «P"l"ayback delay» lol.

:(