UV maps, where to start?

[leigh]

Colour.

Because the colour map sets the look for the entire surface, it's the most logical place to start.

[Dreamwave]

doh, forgot to look at my own thread :rolleyes:
not many replies anyway...

sounds logical leigh...but why do so much people start with bumpmap then? it must have some advantages then? :shrug:

[leigh]

I have no idea why people like to do their bump maps first :shrug:

I cannot see any logical advantages of it.

[ThirdEye]

Bump is the only channel that modifies (it's actually an illusion, not a real 3D displacement) the geometry, probably this is the reason they use it first. It's something similar to model wrinkles, add details...

[leigh]

That is all very well, Third, but what people forget is that anything in the bump channel should be included in the colour map anyway. So you might as well start with colour, because those details will have to be added no matter which map you do first. All details affect ALL maps.

[Threedeemer]

Im no texture expert (my main weakness in 3d), but I've heard the bump map is the most important to get accurate and detailed, then you can tweak and colorize it to produce the other maps such as color and specular. Its sorta like oil painting where you start with a charcoal sketch then from that you create either a black and white or Sepia underpainting to layout the details, then you simply tint it with color washes. After I create a bumpmap base I lower its contrast and colorize it in Photoshop with the Variations filter, then tweak its saturation, hue, etc. in certain areas, also using adjustment layers can be helpful.

[leigh]

I was chatting to Lildragon about this last night, and he told me that the reason he does his bump maps first is because they create the illusion of adding more to the model itself, so when he then begins his colour map, it's kinda like pouring paint into a mould.

An interesting process, I must say :)

I still think that colour is the most logical place to start, but I guess it really is up to personal preference.

Originally posted by Threedeemer
but I've heard the bump map is the most important to get accurate and detailed

All maps should be EQUALLY accurate and detailed. Otherwise it would be a sloppy texturing job :)
No particular map is more important than any other, since it is the combination of all the different maps that gives you the final appearance of the surface.

[ThirdEye]

"...he told me that the reason he does his bump maps first is because they create the illusion of adding more to the model itself, so when he then begins his colour map, it's kinda like pouring paint into a mould..."


That was exactly what i tried to say :p

[lildragon]

Heheh yeah exactly ;) here's the process used on the Cave troll from L.o.T.R

http://www.lordoftherings.gr/movies/sfx_article.asp?articleID=45

speaking of which, how the hell does one paint bump maps like this?

Bascially how would you hand paint a pattern like this?


http://www.lordoftherings.gr/movies/images/troll3.jpg

I mean I'm still kinda new to the whole "painting" the map thing :)

salud

[ThirdEye]

well that's a displacement map, i'd use procedurals since the pattern is quite regular... :shrug:

[Stroker]

About a month ago, I was doing a low-poly humanoid. Well under 1000 tris, but he was suppossed to be sinewy and it just wasn't in the model.

For this particular texturing, I decided to go with the bump map first. I spent quite a bit of time on it. Once I got it done, it was very easy to derive the rest of the maps from that. A little tweak here, a little tweak there, and not much brushing. I was *very* satisfied.

In this particular case it was much easier to start with the bump. I doubt this is a general catch-all technique, though.

[Dreamwave]

Originally posted by lildragon
Heheh yeah exactly ;) here's the process used on the Cave troll from L.o.T.R

http://www.lordoftherings.gr/movies/sfx_article.asp?articleID=45

speaking of which, how the hell does one paint bump maps like this?

Bascially how would you hand paint a pattern like this?


http://www.lordoftherings.gr/movies/images/troll3.jpg

I mean I'm still kinda new to the whole "painting" the map thing :)

salud

nuhu, on the extended DVD edition (dunnow which cd of the four) there is a small piece about the Cave Troll (damn this is getting really off topic but ah, who cares:shrug: )

The troll was modeled in clay and then in wax and painted, the wax model was scanned and then textures, so there are no actual bumpmaps on that...everything is in the model because it was scanned...

based on the painted model the colourmap was made and off course all the others...AND the muscles of course so there is still enough work to do...

I don't know what the fases are but I know really sure all was scanned...

[Crazzy Legs]

It seems to me that asking one of the individuals who worked on it would be the best bet. The people at WETA have been more then helpful on many website forums or pages that I've been to.

I know I heard and read the process behind the troll, it's one of the cinefex magazines if I remember correctly. But I must say that many movies I don't see a reason for not using Bump maps on really small detail (example inperfections on the trolls skin.) The big stuff is definitly displacement maps but the smaller they get, the higher the possibility it is a bump map. Most definitly there are plenty of discplacement maps. I saw an image from DragonHeart where you could see the bump maps in the lips and the side of the face. I imagine now that lighting has improved far beyond DragonHeart, so bump maps could be a little harder to discern, so using them would be just fine.

Find that article about the troll and read about it, it was really cool, and made lots of good sense. I could tell you what I think I remember, but I don't want to lie when finding the correct information is a couple of mouse clicks away.

[jeremybirn]

"so there are no actual bumpmaps on that...everything is in the model because it was scanned..."

Often, after a model is scanned, the difference between the point-cloud data and the surfaces you wrap around it gets solved into a displacement map (Point Cloud Data: lots of info from scanner, impossible to rig. Final character: finite number of points, animatable, needs maps for small details.)

Bumpmaps are less popular with Renderman than with many other renderers, because in Renderman you'd need the same Shading Rate to get equal amounts of detail from a bump map as a displacement map, so there's isn't as much difference in computational expense (sub-pixel-sized details can be displaced rather easily.) Bump mapped details don't self-shadow, so the shading is better with displacement, even for front-facing details that could be bump mapped without hurting your profile. (On the other hand, bump maps can be cheated between passes without alignment errors, and don't require a higher res shadow map as displacement sometimes does.)

-jeremy

[CGTalk Moderation]

This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.