Enjay wrote:I meant to attach this earlier and just didn't.
http://www.mediafire.com/download.php?9klk7ja169bljkr
It may be useful for illustrating the scaling thing that Nash and I were discussing.
Its a map and a simple box model. The box, according to the model editor, is 128x128x128. It has been used for two actors. In one actor (using the Baron marble face texture as a skin) it has a scale ratio of 1.0 1.0 1.0. The same model is used again with a ratio of 1.0 1.0 1.2 and uses the other (archvile?) marble face. In the map there is a box made from sectors that is 128x128x128 Doom units and has the pentagram marble face on it. In the map you can see that both models match the width and length of the Doom sector boxes but the 1.0 scaled Baron one is shorter whereas the 1.2 scaled archvile(?) one is the same height as the sector box.
The model-based actors have been left as non-solid so that you can wander around more freely looking at stuff. You can also see some nice z-fighting on the inside of the box. :P
Yes, and the code responsible for the short Baron box is - as demonstrated a few posts back - inside gl_models.h. I don't know why it's doing that, but that's the way it is. So to counter that, a Z ratio of 1.2 has to be applied in the actor's MODELDEF.
Interestingly enough, if you mod the GZDoom viewport renderer to draw square pixels (gl_scene.cpp), the 128x128x128 sectors (the one with the pentagram) will look perfectly square/cube, BUT the Baron box model will STILL be squashed.
So the conclusion here is that the code that draws models in GZDoom is working independantly of the viewport renderer.