Painting with Mudbox - Stretched Projections, How To Avoid?

mortalhuman

Well, I was watching Wayne Robson's Mudbox stuff and he painted a pillar in moments and it was a pretty well looking asset.

Surely, this wouldn't be useful on EVERYTHING, but it is too much of a workflow increase for me to ignore. Since I have been learning more about materials in Mudbox, I have begun painting on models in a per-element basis to make painting easier, but it is hard to paint the desired faces and not end up painting stretched or broken data into neighboring, hidden faces.

Any tips to get the most out of this technique? Face selection and locking, or something else? Thanks in advance Polycount

danshewan

I haven't used Mudbox extensively, so I could be wrong (some images would be helpful), but it sounds like the mesh you're trying to paint on doesn't have enough geometry.

Depending on how your base mesh is constructed, you might see surface stretching towards the center of flat or larger surfaces as opposed to the corners because of the subdivision operation if the base mesh isn't dense enough, or if the quads aren't equally distributed. Try to even out the quad distribution across all surfaces of the mesh as equally as you can prior to importing it into Mudbox.

Piflik

Since Mudbox does real Texture Painting as opposed to Zbrush's Polypainting, the Quad distribution is completely irrelevant in this case....but I also have limited experience with Mud, so the only Advice I have is to stay away from Faces that are perpendicular to the camera while painting...

mortalhuman

Fair enough. Let me ask another then while we see if any other tips roll in (sounds right to me, but you never know if there is a setting tucked somewhere). Some of my team points out seams and are wondering "how can you get seams on your uv borders in mudbox?!" and I say "well, seams are on every piece of art in the world, you just don't notice them because they are hidden, or just not noticed among all the other art in the scene"

Is that the case, or am I holding back our project either at the engine or artistic level myself?

Here is an example, and it happens in and out of mudbox, max hi to lo bake does it too:

A rock, for example, needs to be split into two UV islands minimum to get a decent mapping, but in-game, and in renders, although my texture is painted on seamlessly and uses ample padding, the UV's come out in renders and in game as if - and this is the best example I can give - they come out at is the top of the rock is a jar lid, and the bottom is the jar, and instead of a tight shut lid that lines up, it is SLIGHTLY turned, you can ONLY see it from points of observation that this rock will Never get in a game, so I wonder, is this unavoidable and my guys are being way too picky (I have never worried about seams before or noticed them so much/hid them well or on hard edges) or is this a problem with A: perhaps my process or B: settings C: maybe something needs to be done in code or shader?

Again, the seam shows in max too, so I think it's not the problem "every program shows them different", because I can get seams on any UV seam anywhere I display any piece of art.

Thanks, this has been bugging our team for a long time, and I have been non chalant about it thinking they are just a fact of 3d?

cryrid

Is the seam noticeable on the diffuse map, or just when you bake it and apply the normal map?

mortalhuman

in this case, both, I suppose, here is an example of the diffuse map only with all other supporting maps disabled:

http://dl.dropbox.com/u/1483473/images/diff_example.png

This is on a half-sphere, or the end of a propane/gas capsule, and the seam is where the UV seam runs through of course, but this helps to see the glare and the shape of what you're looking at, I didn't think it over very well when I cropped it :P

With the normal map on, the shift moves through the daycycle, but that's another issue and I think that since it is a diff->disp->normal based map, it is simply my own problem with rotated UVs. The diffuse shouldn't have this seam though with the problematic normal disabled, right?

It's sort of a case study where I've always had seams, but hid them well, and they were never noticed unless you were looking right at them.

It is probably also important to mention that if I back up a little bit, like 10 meters, the seam vanishes, but I have until now figured it was just less pixel information to define against itself on screen and thus lucky?

edit: there could be the chance that post sharpening did it prior to dropping from 2048 to 1024? i dont usually have anything this severe, but it is characteristic of other things I encounter, so, trying to put as much info as I can lol.

oglu

under optimal conditions there should be no seams in mudbox...
if there are big faces its hard for mud to find the coresponding area on the other uv shell...
really long faces are also bad...
if you are painting a higher res version of your mesh you will get less seams...
i subdivide my lowres meshes to get a better control...
if your mesh will distorted cause of the smooth subD there is a option for subdividing without smoothing...
mudbox wont smooth the uvs so its no problem to do that...
or if some shells are in a different scale youi will also get seams...

mostly i start with a flat version of my mesh and cleanup the seams afterwards...
[ame]http://www.youtube.com/watch?v=RkH5goz-RO4[/ame]


for different materials i split up my uvs in tiles...
like here...
[ame]http://www.youtube.com/watch?v=Ro_Ruhu-SGs[/ame]


here an other useage of uv tiles...
[ame]http://www.youtube.com/watch?v=MPXJUK3ekgE[/ame]

if there are really bad areas i use the Ps link to fix them...
[ame]http://www.youtube.com/watch?v=RxoGn1qPhYA[/ame]

mortalhuman

wow, thank you so much for taking the time! A lot of reassurance in some things I've been trying, and some new tips, means a lot to me! Thanks so much everyone for insights, I was able to be more careful with a better sub d version for the mesh and it closed things up. Not perfectly, but much much better results (although nothing to show - it was better to test these techniques quickly)

Thanks again!! All of my gratitude.