[XviD-devel] lumimasking problem

[Radoslaw 'sysKin' Czyz]

> I was dumping quantizer data at every interaction of motion estimation
> and I discovered ome very strange things.

Hi, I think i found the problem: I was wrong from the beginning :) The
thing is that while ME is performed, pMB->quant is not initialized yet
at all. It's being initialized later, after MC.
The output I showed you is probably with one-frame delay - dquant is
current, quant is delayed.

> .... Anyway, I have to assume
> that my ME doesn't work with lumimasking at all, because quant data
> are different then I had expected:

And this is 100% correct.

I'll fix it (in a simple way - I'll just copy "pMB->quant=" code from
encoding loop to ME loop (leaving everithing in coding loop, too) and
see if it fixes the problem.
If I succeed within next 28 minutes, I'll check if my new CVS account
works and try to commit it. If I won't, I'll try tommorow ;> (my
dialup will hang up in 28 minutes)

Radek