Quark / map2dif texturing trouble
by J. Alan Atherton · in Artist Corner · 07/10/2004 (7:04 pm) · 12 replies
I've built this "sawblade" ramp thing... dunno what to call it. Art. Yeah, that's it. Anyway, the texture looks great in quark, but when I bring it into torque, it's all... wonky. Take a peek:
The left is how it appears in torque. The right is how it appears in quark. I would much rather not fix it for each block, as that would take hours (lotsa blocks). Also, I've done this with a radial duplicator. If someone wants to try to replicate this problem, just create a radial duplicator, put a value in the "spiral" entry to make it climb like a staircase, then mess with the "linear" values to get it to twist a little bit. Then put your favorite texture on, and see if you get the same results. I've even tried dissociating the duplicator images, so each block was independent, and it looks exactly the same in torque.
Hmm... just had an idea... I'm not sure that I'm using the most up-to-date map2dif. In fact I'm sure I'm not. I'll try this again soon to see if I can't figure something out with a more recent map2dif.
The left is how it appears in torque. The right is how it appears in quark. I would much rather not fix it for each block, as that would take hours (lotsa blocks). Also, I've done this with a radial duplicator. If someone wants to try to replicate this problem, just create a radial duplicator, put a value in the "spiral" entry to make it climb like a staircase, then mess with the "linear" values to get it to twist a little bit. Then put your favorite texture on, and see if you get the same results. I've even tried dissociating the duplicator images, so each block was independent, and it looks exactly the same in torque.Hmm... just had an idea... I'm not sure that I'm using the most up-to-date map2dif. In fact I'm sure I'm not. I'll try this again soon to see if I can't figure something out with a more recent map2dif.
#2
07/12/2004 (3:47 am)
Yep, 256x256 and 64x64. Also tried the most up-to-date map2dif... no difference at all. 
#3
07/12/2004 (6:21 am)
John, could you send me the qkm and textures for that set of shapes? I'll look it over.
#4
I'd much rather have map2dif translate these linear transforms correctly though. I'd probably be able to figure out how to program it eventually, but it'll probably take too long relative to the timeline of my project. So I'm not going to touch it right now. Essentially, the way I see it, is map2dif isn't quite properly handling texturing. It works just fine for most cases, except for the linear transformations (pity... they make for some cool geometry).
07/12/2004 (7:03 pm)
Ok, I've discovered at least part of the problem. The top face works when I take the "linear" transformations off of the radial duplicator. (Actually I had to make the object from scratch, then not apply the linear transforms) The problem then is the side faces (on the inside of the corkscrew) aren't mapped quite correctly, though it could be good enough for my purposes. The spiral value (which makes the ramp go up) is what causes these maps to fail. I'd much rather have map2dif translate these linear transforms correctly though. I'd probably be able to figure out how to program it eventually, but it'll probably take too long relative to the timeline of my project. So I'm not going to touch it right now. Essentially, the way I see it, is map2dif isn't quite properly handling texturing. It works just fine for most cases, except for the linear transformations (pity... they make for some cool geometry).
#5
07/13/2004 (12:25 am)
Not sure if this will help or not, but do you have floating point turned on?
#6
07/13/2004 (9:53 am)
I have seen this sort of problem on other surfaces which were rotated along multiple axis. It doesn't alwasy seem to happpen though. I think that there is some problem in translating the texture coords of oddly-rotated faces into DIF... but I am not sure if it is related. Does is screw up on a flat (non-twisted) staircase as well?
#7
Alex,
I mentioned the linear transforms above... that's what causes the twisting (both of the geometry and the textures). The only reason I need the linear transforms to twist is that the blocks don't quite line up perfectly unless a little bit of twist is added to each one. They go around the circle and even upwards just fine (so the ramp looks more or less the same, except for a little bit of "uneven sidewalk" syndrome). I tried lining them up by moving vertices, but quark wants to move the wrong pair of vertices relative to what I need (if that makes any sense). In the case without twisting the texturing works fine. In other words, if it's a flat circle it works without a hitch. If I make it a staircase, then the top texture works fine, but the side textures are bad (imagine the staircase as a cylinder, then the texture is mapped with repeating to the inside of the cylinder, then you take the stair pieces out). One thing I don't really understand is that I can dissociate the duplicated brushes (so each becomes a unique, separate brush), but it still doesn't work. It would seem to me as though it would "bake" the texture transforms in, but it doesn't.
As I said, I can work with it as it is for my current project, but it would be nice to have. It's one of those "won't die without it" kind of things. I guess this could be classified as a feature request. I can make up a couple of example maps where the texturing fails if it would help somebody to investigate the problem.
07/13/2004 (11:33 am)
Floating point is on, yes.Alex,
I mentioned the linear transforms above... that's what causes the twisting (both of the geometry and the textures). The only reason I need the linear transforms to twist is that the blocks don't quite line up perfectly unless a little bit of twist is added to each one. They go around the circle and even upwards just fine (so the ramp looks more or less the same, except for a little bit of "uneven sidewalk" syndrome). I tried lining them up by moving vertices, but quark wants to move the wrong pair of vertices relative to what I need (if that makes any sense). In the case without twisting the texturing works fine. In other words, if it's a flat circle it works without a hitch. If I make it a staircase, then the top texture works fine, but the side textures are bad (imagine the staircase as a cylinder, then the texture is mapped with repeating to the inside of the cylinder, then you take the stair pieces out). One thing I don't really understand is that I can dissociate the duplicated brushes (so each becomes a unique, separate brush), but it still doesn't work. It would seem to me as though it would "bake" the texture transforms in, but it doesn't.
As I said, I can work with it as it is for my current project, but it would be nice to have. It's one of those "won't die without it" kind of things. I guess this could be classified as a feature request. I can make up a couple of example maps where the texturing fails if it would help somebody to investigate the problem.
#8
07/13/2004 (4:14 pm)
John, in QuArK, check out the texture options (under the options tab). You have "default texture movement", "sticky textures", and "axis stuicky textures". I normally use the default but in this situation some experimentation with axis sticky might be the trick.
#9
Thanks for that... I can't believe I didn't even notice those. I searched through the configuration menus rather closely, but didn't even notice these. I played with them, and it seems axis sticky textures is the most promising... it almost gets it right but is still a little off (the first block is right, and it gets a little bit off at each one, so by the end the "blade" portion is off of the cube). None of the settings handle the sides of the cube at all... those are always way off. Hopefully these settings will work in some other situations, though.
I appreciate you looking into this.
07/13/2004 (4:53 pm)
Desmond, Thanks for that... I can't believe I didn't even notice those. I searched through the configuration menus rather closely, but didn't even notice these. I played with them, and it seems axis sticky textures is the most promising... it almost gets it right but is still a little off (the first block is right, and it gets a little bit off at each one, so by the end the "blade" portion is off of the cube). None of the settings handle the sides of the cube at all... those are always way off. Hopefully these settings will work in some other situations, though.
I appreciate you looking into this.
#10
Hmmm, just reread above where you said you did that. I had no problem...what version of QuArK are you using? What version of map2dif?
07/13/2004 (5:01 pm)
John, forget the sticky textures...just dissassociate the radial duplicator into its sub elements and compile:Hmmm, just reread above where you said you did that. I had no problem...what version of QuArK are you using? What version of map2dif?
#11
I am using the most current map2dif from head (freshly compiled). I am using QuArK 6.3c. That's another thing I'd like to ask... can QuArK 6.4 be setup for torque? There are a few features (tagging, gluing) that looked pretty good.
Hmm... I'll have to fish out the version I sent to you... since mine currently has a bad brush ( < 3 points issue) that I can't seem to find. Once I do that I'll post again with my results.
Thanks again.
07/13/2004 (5:28 pm)
Awww man! Why does it work for you? Hmm... I may just need to make it again from scratch and try again. This is probably the same as when I tried taking off the linear transforms from the radial duplicator (removing the twist), and nothing changed, but when I created the spiral from scratch, the texture was good at every stage up until I dropped in the linear transforms for the twist.I am using the most current map2dif from head (freshly compiled). I am using QuArK 6.3c. That's another thing I'd like to ask... can QuArK 6.4 be setup for torque? There are a few features (tagging, gluing) that looked pretty good.
Hmm... I'll have to fish out the version I sent to you... since mine currently has a bad brush ( < 3 points issue) that I can't seem to find. Once I do that I'll post again with my results.
Thanks again.
#12
Also, I got no bad brush message when compiling your full map
07/14/2004 (7:04 am)
Get the QuArK 6.3 (final) ---> there are several updates to 6.3c that fix textures etc. if you can't find it at planetQuake, let me know. Or, DL the zip (change extension): holodeck.st.usm.edu/vrcomputing/vrc_t/tools/quark/software/QuArK630.zm9Also, I got no bad brush message when compiling your full map
Associate Matt Fairfax
Night Heron Games