GK Compressibility Check

[gm4it]

I have difficult in in the comprension about the compressibility check criteria of gordian konot.

In a Forum in Doom9.org I only found the following about the functionality:
"The test compares what the maximum b/p*f value could possibly be, against the bitrate you've currently configured, usually by selecting a file size. It is that ratio, expressed as a percentage of maximum bitrate, that is important."

But is it correct calculate the RATIO compareing the average value with the maximum?

- If we suppose that only a frame in the output has a value b/p*f equal to the maximum value, the overall loss of information is insignificant, depending on the total number of frames checked and the percentage reported by the compressibilty check (roughly): RATIO * 1/F.
- if we suppose that all the frames in the output have a value b/p*f equal to the maximum value, the loss is (roughly): RATIO * 1.

A better ratio should be weighted, with the number of frames that loss information and the entity of loss. For example a better ratio could be estimated in this way:
1.should be calculated the ratio for each frame checked,
2.the ratios summed,
3.and divided by the number of frames checked.

Regards.

Giorgio

[LordIntruder]

Regards.

Giorgio

If you change the mathematical way to calculate the ratio or whatever, at the end you got a figure. This figure gives you an idea of what the quality will be if you encode the whole movie with this bitrate and video size. Encodes after encodes you know, thanks to this figure, what quality you are going to have, you know that below a limit the video will look worse, etc...

So if you change the way that figure is computed, you are going to move references thousand of people got. It's useless, except on a mathematical point of view I agree, :) but if this is at the end to replace the current reference that works very well by another and have to learn again what are the new limits, I see absolutely no interest on a practical point of view for encoders except to bore them.

Now on a theoretical point of view, it's totally different. :)

[gm4it]

OK. :-)

Could you suggest me a method for a self-analisys of a compressibility check run? All information are in the vdenc.log file?

Thanks.
Giorgio