Incomplete stats file support - wrong Luma Noise?

[LigH]

Dear developers (-h, Koepi...): When I load stats files created by XviD into GKnot and watch the graph, the blue Luma Noise curve looks very strange, it seems to have always nearly the same low value (e.g. 0.008) - this ruins the display of this "curve" (it looks like random blue needles instead).

Do you think that you will once be able to support this part correctly? Not only that GKnot will display a beautiful curve, but also because this value may be used for optimising the bitrate calculation.

[-h]

What does the lumi noise value actually mean? If you can supply a calculation (i.e. weighted average of macroblock DC coefficients?), it might be able to make its way into the codec.

At the moment, no such value is returned by the codec's core.

-h

[LigH]

Don't ask me what exactly "luma noise" means; I remember it is reported somewhere, I read it once (it's about the difference of luminance over the whole frame, right?). Nandub developers shall know it, because this value is obviously used in the stats file - Koepi shall be a good address I hope.

[Koepi]

Uh, crossposting over different forums is really odd - I answered the same question over at www.videocoding.de, stating that it's a dummy yet :)

Regards,
Koepi