MergeChroma

[WarpEnterprises]

Has anybode else encountered problems with the built-in (in avisynth 2.0x)MergeChroma function?
The external plugin works ok, but the built-in gives an "illegal instruction" error when opened in VD.
(I've an celeron-850)

[sh0dan]

Seems like ATHLON is defined for this build, and it is doing prefetching. I don't have access to the CVS tree, so Dividee or Richard will have to remove them (just take them out completely, they don't help that much).

[sh0dan]

I just compiled a new binary for you to test:

Download from: here (http://cultact-server.novi.dk/kpo/avisynth/avs_202.html)

Changes:
- No athlon codes in merge. (not committed)
- Converttoyuy2() MMX optimized. (not committed)
- Fixes in greyscale() in RGB mode. (not commited)
- Latest CVS changes (compare, etc.)

[Richard Berg]

You've been added to the project. I think I tried to do this before but you weren't registered yet or something...anyway, you should be able to check-in files now.

[cjv]

@sh0dan
The new build works great (on my P4-m).
I couldn't, for the life of me, figure out why the MergeChroma was
causing an unhandled exception after upgrading to 2.0.2, but the Athlon thing sure would do it :)

[WarpEnterprises]

I can confirm, too - now it works.

But I did encounter a strange error (it reads "srcp==0") sometimes (even with the external plugin).
I couldn't find exactly what caused it, but converting to rgb and back to yuy solved it.
Maybe you find the error with the message above (it looks like a hardcoded one)

btw is there any principal difference between Layer(...,chroma=false)
and MergeLuma ?

[sh0dan]

Yes - the "srcp==0" is an internal error that should never be displayed (or happend). It should never happend when using the AVS 2.02 - only the external plugin.
An yuy->rgb->yuy is a nice temp solution, but not a good thing for quality.

Mergechroma/luma is just simpler than layer - that's the main advantage. I find layer very nice, but it's a bit hard for me to figure it out sometimes.