[Fluxus] png diagonal oddness

[Kassen]

>
>
>>  added the warning and a note in the help string of load-texture.
> i'm still not sure if the multiple of 4 size requirement is valid. please
> let me know if you get a warning, but you can display the texture properly.
>
>
We may have worse issues. I just got a mail from Rob who claims that his
python script preserves the dimensions in pixels. This may mean that we
could have two png files with identical dimensions of which only one would
glitch. I asked him to provide two such files. I may already have 2; I'll
look in the morning.

Greg also wrote to me, in a mail that I take it was meant for the list;

Some explanations here:

http://forums.tigsource.com/index.php?topic=11908.0


Quickly:

- non power of 2 (NPOT) textures are available through an ARB extension
which is not supported by ATI/AMD

- there may be a performance hit if you use NPOT textures


Don't! :)


That's valuable info and I'll stick to it, from now on, but I'd also still
like to know what glitches and what doesn't, if only so we could glitch
on purpose if we'd like. I'm happy with this note though, as I'm fairly
certain I tend to max out both my GPU and CPU while having video memory to
spare. On top of that; I don't think we were great fans of ATI to start
with, but we may get more Fluxers with ATI cards as/when Windows support
improves.

For pixels->texture I did find that powers of 2 worked and nothing else
(that I can remember) but I strongly suspect more is going on here. If
textures from png files are somehow different from textures from pixel
primitives that's noteworthy in and of itself. I suspect they may be but I'm
not betting any farms on that before testing.

At any rate we're on to something.

Yours,
Kas.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.pawfal.org/pipermail/fluxus-pawfal.org/attachments/20110120/4689aa5c/attachment.html>