[PD] Gem pix_snap2tex bug?
cgc at humboldtblvd.com
Sun Sep 25 00:27:20 CEST 2005
There are a couple of issues I see with snap2tex:
1) It apparently only does power of two textures as currently written.
You can try using power of two window dimensions for cleaner edges
right now, but it's to be expected that non-power of two sizes will
have somewhat undefined behaviors. I have already changed things to
support sub texturing, which has reduced edge artifacts.
2) The object really needs rectangle texture support. I'll have a
look into doing this soon.
3) There might be a small bug with the way ATI GPUs handle texture
borders. This can sometimes be seen along the edges of a cube for
example. It used to plague rectangle textures quite a bit, but it
comes and goes with 'updates' (you know how those go with Apple).
There are a couple of opportunities for additional features like
texcoord manipulation and using repeating textures, etc, that might be
worth a look. Also, one of the goals is to move to multiple render
targets, which will make this snap2tex slightly obsolete. I have no
idea when that will be done...
On 9/23/05, vade <doktorp at mac.com> wrote:
> Ive been having an odd issue that I cant seem to figure out. Im using
> pix_snap2 to capture the complete rendered frame of my openGL scene. My
> objects are drawn in the center, and never go out of the openGL clipping
> volume, or the 2D clipping plane of the window.
> However, pix_snap2tex seems to be picking up a one or two pixel border
> around the captured frame.
> Im running OS X 10.4.2, powerbook, 128Mb VRam, GEM 0.90, Hans's PD
> Here is a screenshot to illustrate
> Now, I was able to minimize these lines to only the bottom left by changing
> the capture resolution to snap2pix to 798x598 (chopping off two pixels from
> my gemwin resolution), but I cannot completely get rid of those lines..
> anyone know? Idly curious..
> btw, I tried mailing this to the list with an attachment, and it never went
> through... whats up with that ?
> v a d e //
> doktorp at mac.com
More information about the Pd-list