Texturing... ( i dont like UVW's )

Dodo

Tank
Joined
Aug 30, 2004
Messages
2,140
Reaction score
0
Texturing a home grown self created model is some difficult sh*t !! for instance... i dont even like saying it but.. oh god... the UVW !! i hate it ! soo thank god, i found this program "DEEP PAINT 3D" ( http://www.righthemisphere.com/products/dp3d/ <---- fantastic !! ) but i dont know if there will be any problems... has anyone of you ever tried this program ?! if you do, plz tell me if it works like they say it does ? (they say its god-like) :D
 
That app is only good up to a point. To be honest UV mapping isnt that scary unless your mesh is a mess or you havent read any tutorials on unwrapping and texturing your uv's.

There are hundreds of tuts out there.

Can you post a wireframe of your model?
 
Deep UV/Deep Paint are pretty limited in scope. You're far better off (in the long run) biting the bullet and unwrapping. Yes - it's tedious, but it's necessary.
 
well im working on a MOD (and sorry, i cant tell the name... it's secret... dont want to put it on the internet without any media... we've been doing nothing else then thinking up gameplay aspects, a making concept art for 3 months... so we have a good idea of the player model... and we have a 2 good moddelers on the team, soo, im doing the texturing.) And i've never textured a single polygon in my 16 years that if lived on this planet ! but i can draw and paint like hell, soo the texturing in deep paint looks easy... sooo NO I DONT HAVE A PLAYER MODEL THAT I CAN SHOW YOU GUYS but thanks anyway :D
 
Mr. Super Moderator, can a ninja kill a Zombie ?! :D
 
Check out 3dbuzz.com nice vid tut on how to unwrap under the unreal section somwhere abouts.
 
deep uv is just a uv mapping tool, no painting.
deep paint 3d is a piece of crap. i dunno what app u use for painting but its horrible.
 
Pi Mu Rho said:
Deep UV/Deep Paint are pretty limited in scope. You're far better off (in the long run) biting the bullet and unwrapping. Yes - it's tedious, but it's necessary.
Need to create proper UV maps with Deep Paint anyway, if its to be used properly. I find Deep Paint handy for getting positions correct and checking for problems with stretching, then filling in the basics or cleaning up already baked maps.

-

UV's are basically unavoidable Dodo, you could use automatically generated ones, but they have their own problems and are in most cases best avoided unless you know exactly what your doing. So as Pi says, best to just jump in and do them properly and by hand, only way to be perfectly sure.

Don't worry though, UV's in XSI are insanely easy to setup and actually fun to do.
 
thanks... (I DONT WANT TO UNWRAP) still isnt deeppaint a good way of texturing a model... whats the difference !? the only thing is you can paint on a 3d model... take an other brush and paint the bump. it sounds great. so whats better about unwrapping (manualy :p)

But ok... i'll try UVW unwrapping, dark elf..
 
Dodo said:
thanks... (I DONT WANT TO UNWRAP) still isnt deeppaint a good way of texturing a model... whats the difference !? the only thing is you can paint on a 3d model... take an other brush and paint the bump. it sounds great. so whats better about unwrapping (manualy :p)

Control


I've used both apps, Deep UVs and Deep Paint and I can tell you that unwraping your model in your 3d package is a better idea. Deep paint can be usefull if you want to add extra details fast (like blood splatters or scratches) or remove seams but other than that... There is not magic in texturing 3d models, no programs will do it for you, you'd better learn how to unwrap correctly and then paint on it, you'll get better results.
 
Dodo said:
thanks... (I DONT WANT TO UNWRAP) still isnt deeppaint a good way of texturing a model... whats the difference !? the only thing is you can paint on a 3d model... take an other brush and paint the bump. it sounds great. so whats better about unwrapping (manualy :p)

But ok... i'll try UVW unwrapping, dark elf..
It sounds great, and in theory it is (I happen to really like Deep Paint and use it often) But forget trying to get really fine details with it, and to use it to its best, you want to be painting on the original model you should be working with, not the low poly version, which means you'll need a powerful setup or it'll chug along. You could always cut the model up to work on it, but that can get messy. And UV mapping a high poly object is suicide in many cases.

Doing it manually, you know _exactly_ where things will be, automatically you can't say if a UV map will be placed correctly or not. Also automatic isn't very optimised compared to manually create the entire UV map piece by piece. The software wont know which area's should use more UV space than others.

Really its not all that complicated once you get used to it. Every model can be brought to its basic shape. a head is a cylinder for example. So you'd create the basic cylinder UV map, then move points about until you've a much cleaner map.

There's also two kinds, Discontinuous and Continuous. Discontinuous is basically every polygon is flat on the UV map, so there's never any stretching. But none of the polygons are joined, hence the name. Also its next to impossible to paint on in photoshop or similar, and if your not super careful you'll get artifacts at the edge of every polygon.

Continuous is basically you could draw along a line of polygons in a UV map and it would be, well, continuous. There's no breaks between polys, unfortunately automatically this will always generate stretched area's, unless you flatten the model first and prepare it for mapping (the flatter the polygons in the UV map are, the less stretching and warping and the better quality the final map will be.)

Also with normal maps, discontinuous map errors will show up far more.

So really, your competing with people who are making their models as high poly versions, texturing them and baking the data onto low poly versions. Taking the easy way out is going to show in your work and they wont have a chance against those done properly. There's really no quick fix or simple way of getting good quailty, thats why people doing it for a living get paid so well. It's an art form in itself and the automatic methods just don't cut it anymore.
 
ok.. well i understand, i wont use the deep paint to unwrap the model's UVW but i'll try to learn unwrapping, thanks again dark elf ! :D
 
how long does it take to texture a model... (for example) the middle dystopia model on (the blue one with the "glow-in-th-dark" eyes) ? a week ?
 
Dodo: Please quite double posting. I've warned you once, if you continue to do it, you'll be warned again.
 
uh.... ok.. sorry. sorry. sorry. I wont do it again! "SUPER MODERATOR !"
 
Dodo said:
how long does it take to texture a model... (for example) the middle dystopia model on (the blue one with the "glow-in-th-dark" eyes) ? a week ?
Entirely depends on the method and the talent, basically one of the following..

method 1 (recommended)
Build high poly model, texture it. Build low poly cage and UV map it, bake high poly texutres, maps and normals onto UV mapped low poly cage. Done.

method 2 (old game method)
build low poly game model, UV map it, texture it by hand

method 2 can look brilliant in the hands of a very talented texture artist and if the model is well made.

method 1 has lots of uses, will give very cool results if the effort is put into it. But can take longer, though obviously worthwhile in the end.
 
It's worthy of note that method 1 is better employed for organic models, where the extra level of detail can't really be achieved with texturing. For more mechanical stuff, though, method 2 still works well.
 
For me, it takes the same amount of time to texture the model (unwrap/photoshop) than to build the 3d model.
 
Tell me about it. That's why it's such a pain in the arse to do, and you don't feel much of a tangible benefit afterwards.
 
Well, you get a fully textured model, that's a nice reward :)
 
method 1 (recommended)
Build high poly model, texture it. Build low poly cage and UV map it, bake high poly texutres, maps and normals onto UV mapped low poly cage. Done.

method 2 (old game method)
build low poly game model, UV map it, texture it by hand

ur method 1 is flawed, u wont get a decent lookin model from that. You still have to paint good texture maps, especialy for the source engine since it renders normal maps like crap (you wont see em half of the time) and for engines like ue3 and doom3 u still need to create nice color maps. Note that uving both the highpoly and lowpoly model is a huge waste of time. Baking color maps off a highpoly is pretty stupid, because you will need to paint color and bump for the highpoly, which is much more annoying to uv map then a low poly (because a decent control mesh is fairly high poly). The dystopia model could take 2 days to make if u work on it for a long time each day (model/uvs first day, texture 2nd)

either way ur stuck learning uv mapping. deeppaint is a horrible texturing program. Only useful for cleaning up seams and stuff like that. If you know how to paint then use pshop or painter.
 
thereisaspoon said:
ur method 1 is flawed, u wont get a decent lookin model from that. You still have to paint good texture maps, especialy for the source engine since it renders normal maps like crap (you wont see em half of the time) and for engines like ue3 and doom3 u still need to create nice color maps. Note that uving both the highpoly and lowpoly model is a huge waste of time. Baking color maps off a highpoly is pretty stupid, because you will need to paint color and bump for the highpoly, which is much more annoying to uv map then a low poly (because a decent control mesh is fairly high poly). The dystopia model could take 2 days to make if u work on it for a long time each day (model/uvs first day, texture 2nd)
Clearly you've never tried method 1 or you'd know it works.

a) build the high poly model

b) texture it, far easier than painting everything by hand

c) bake it all to a low poly model

done

I fail to see how its flawed, Source's normal maps are fine and you DO see them, you just don't notice them (try turning bumpmaps on and off a bit, you'll notice the difference instantly)

You need to read up on it because obviously you've never done it before, and you think you UV map the high poly model hehehe, no you don't, you never need to. As I said earlier that would be suicide. You _only_ ever UV map the low poly object, there's simply no need to with the high poly one as you can use all other methods of texturing to work on that, you don't need to use UV's unless you want to.
 
ok, well I, what method did valve use for the Counter-Terrorist ?
 
why don't you want to unwrap? Get someone else to do it for you then. It should be done...
 
our modding team, isnt THAT big (3 people :D), and we dont want people over the the internet texturing our models, or kwowing anything of our gameplay elements that we want to put into our MOD. And we need more than one texturers (cause a mod cant be textured by 1 person !), and i was interested in texturing, but now after all what you guys posted I understand that UVW mapping isnt that bad is i thought, and its something you need in the game biz... now the big question is how long will it take for me to learn unwrapping ? (im a fast learner)
 
The principle is easy, but unwrapping is a time-consuming process. You're basically looking at seperating your model into it's most basic shapes, and flattening them out or unfolding them.
 
it helps to have a mind geared towards puzzles, as sometimes thats how it turns out :p its not hard, it just takes a while, like Pi said.

Tip: look at each section individually and work out in youre head how it will unfold before diving straight in and you'll be fine :thumbs:
 
The Dark Elf said:
Clearly you've never tried method 1 or you'd know it works.

a) build the high poly model

b) texture it, far easier than painting everything by hand

c) bake it all to a low poly model

done

I fail to see how its flawed, Source's normal maps are fine and you DO see them, you just don't notice them (try turning bumpmaps on and off a bit, you'll notice the difference instantly)

You need to read up on it because obviously you've never done it before, and you think you UV map the high poly model hehehe, no you don't, you never need to. As I said earlier that would be suicide. You _only_ ever UV map the low poly object, there's simply no need to with the high poly one as you can use all other methods of texturing to work on that, you don't need to use UV's unless you want to.


if i were to do that, where would i get my color info from for the lowpoly? if u bake it off the highpoly model u still need to create a uv for the high poly in order to give it any sort of decent shading. And u wont get decent results using prodecural textures on a high poly without a uv map. Unless im missing something, you either paint the color maps by hand or u bake em off a highpoly model (and unelss you use color, sss maps and other stuff, it will most likely look like crap using procedural textures)
 
thereisaspoon said:
if i were to do that, where would i get my color info from for the lowpoly? if u bake it off the highpoly model u still need to create a uv for the high poly in order to give it any sort of decent shading. And u wont get decent results using prodecural textures on a high poly without a uv map. Unless im missing something, you either paint the color maps by hand or u bake em off a highpoly model (and unelss you use color, sss maps and other stuff, it will most likely look like crap using procedural textures)

You don't need UV maps on the high poly model to create the texture information. You use whatever method you want that gets the results your after. You only NEED a UV map on the low poly object, otherwise there's nothing to bake the data to. So as for the high poly model, you use any method you want, projection, planar, cylindrical, procedural, decal and so on.

Once you've got that, you'll have likely also created all the other channels, specular, luminosity, diffuse, bump. You give it a decent ambient lighting, no direct light or shadow (brings out the details just as you would do when drawing by hand) Then all those are baked at the same time to individual image files, along with the normal map data.

Providing the UV map on the low polygon model is correctly setup, and the baking process goes as planned. Your low poly model will then have all the images it requires.

Also another use to it, which you can't successfully do with painting it by hand is the normal maps aren't just used for details, they are also used for smooth surfaces. A high poly head is going to be far smoother because of the number of polygons it has than the low poly version. This data is carried over through the normal map, so when its on the low poly model and in game, the realtime lighting on the models has a better quality because of the normal map tricking it into thinking there's more normals than there is. Thats why the model lighting looks nicer in dX9 than it does in dx7, or dx9 with and without normal/bump maps.
 
The raising the bar book mentions that Valve found Zbrush worked well for creating normal maps, you still need to unwrap UVW's but honestly I've found worst casing it a day or two and you should be done, there are lots of little annoying things, unfortunately games require more optimizations than other media forms, or I'd recommend you try out something like Bodypaint 3D.

Zbrush:
http://pixologic.com/home/home.shtml

Bodypaint 3D:
http://www.maxon.net/index_e.html
 
my point is, it will look like shit if simply bake everything off the highpoly without giving the highpoly proper shaders, and in order to give it proper shaders, it would require proper texture maps, then it will require uvs. Of course u can just bake a basic shaded model, but it will look horrible.
 
thereisaspoon said:
my point is, it will look like shit if simply bake everything off the highpoly without giving the highpoly proper shaders, and in order to give it proper shaders, it would require proper texture maps, then it will require uvs. Of course u can just bake a basic shaded model, but it will look horrible.
hmm. I think your getting confused as to how you texture a high poly model. This thing wouldn't be used in realtime, or animation, ever. It's a pre-render job only.

UV's are still pretty new, the realtime side had them before the pre-rendered for once. Artists got on just fine without being able to use UV maps in film/tv/pre-rendered CG.

UV maps primary use is realtime. You can use them in pre-rendered and they are useful in many ways. But the fact is, its simply a shortcut in that sense.

Realtime your limited to how you can map a model. Pre-rendered has no such limits, you use whatever method you want. To say you need to use UV maps to get good quality is taking the piss out of years of work by pro's that created flawless imagery without having UV maps handy.

I think your also underestimating just how powerful procedural textures can be too. Yes in the old days you were pretty much limited to doing fractal cloud effects or grid patterns and that was that. But now in software you can mix them about, use other types of mathematical effects to create a lot of surface types.

The times you need to use images, why limit yourself to UV's when you can use all manner of projection methods. You build up the textures piece by piece, using many layers and surface types to get what you want. In such a case there's simply no need to worry about optimizing it, its never gonna get animated, it'll probably only be rendered once at full detail for when you bake all the data.

You name and part groups of polygons as you build it so you know what their supposed to be, give them unique surface names and you work on them one at a time.

eg: the character has two arms, you'll lay down the base surface for both arms, and then work on the details for each, avoiding them being an obvious copy of the other. a single crust procedural texture with its settings turned way down, given a brownish tint, used in the color channel will give you random moles or freckles. An inverted crumple procedural, with its size set right down and used on a low bump setting in the bump channel will give you a basic skin appearance. Create a wavey grayscale image and project that onto the elbow as a turbulence layer to effect the procedural skin layer. Then copy it and use it as an additive bump layer above it will give the appearance of the skin on the elbow.

So you don't just do it all in one single image, never do that if your doing things properly. You build things up piece by piece, working on each channel and surface type at a time until things look how they should. Then you bake it to a high res image, which spits out all the types of surfaces for the entire model, setup automatically to accurately match the low poly's UV map, and thats it.

Being honest, your arguing with years and years of methods that no pro will agree with you on. UV's are great, they are useful, but to think you need UV's to get good quality results is like telling an artist they need the most expensive pencil and paper to do amazing drawings.
 
On The subject on unwrapping I have finished my latest model and I have been worrying about which application to unwrap it in. Basiclly I have a 7005 triangle model and it will have about 5 LOD levels, if I unwrapped it inside of 3dsmax7 as soon as I edit a single polgon of the 7005 it will totally reset my UVW is there any application I can use that will simpily work out the correct change from polgon to polgon so technially I won't have to unwrap it again?
 
IchI said:
On The subject on unwrapping I have finished my latest model and I have been worrying about which application to unwrap it in. Basiclly I have a 7005 triangle model and it will have about 5 LOD levels, if I unwrapped it inside of 3dsmax7 as soon as I edit a single polgon of the 7005 it will totally reset my UVW is there any application I can use that will simpily work out the correct change from polgon to polgon so technially I won't have to unwrap it again?
Lightwave can handle editing a mesh without destroying the UV map, including splitting polygons and such. Though will still need a bit of editing afterwards just to clean up in some cases. But most of the time its best to avoid editing a mesh once the UV map is setup.
 
uv maps arent really shortcuts, since polygons require uvs to display textures, sure do it half assed with a basic projection. True nurbs dont because the surface is in uv space, so no handwork must be done (other then the pain in the ass texturing since you cant modify uvs in a nurbs surface) so yes most old school pros have mainly worked with nurbs which requires no uving, in that case i would use procedural textures since its a lot ezer then painting on a horribly stretched uv map, but for polygons, you would get a lot better results by using 3-4 different maps then creating a gigantic render tree for a skin shader which produces random results, the tones and wrinkles in the human face aren't random.

ichi, im pretty sure max dosent reset uvswhen u add in a edge, as long as u work under the uvw unwrap modifier. XSI automaticaly adds in edges that are cut or subdivided in, i'm positive maya does it.
 
thereisaspoon said:
uv maps arent really shortcuts, since polygons require uvs to display textures, sure do it half assed with a basic projection. True nurbs dont because the surface is in uv space, so no handwork must be done (other then the pain in the ass texturing since you cant modify uvs in a nurbs surface) so yes most old school pros have mainly worked with nurbs which requires no uving, in that case i would use procedural textures since its a lot ezer then painting on a horribly stretched uv map, but for polygons, you would get a lot better results by using 3-4 different maps then creating a gigantic render tree for a skin shader which produces random results, the tones and wrinkles in the human face aren't random.
Fine yeah ok whatever. Your not even willing to listen to this other method thats been proven time and time again to work, you've got it in your head your right and thats that. Well fine, I cba trying to discuss it like a grown up with you, I've ignored your down the nose comments until now, now its just boring and im wasting my time trying to help you. Heck you even contradicted yourself complaining about UV's on NURBS surfaces. Then go on to call projection mapping details "half arsed" and completely ignoring other points I made, prefering to focus entirely on procedurals.

You go do things your way with your little freebie copy of XSI and keep thinking you know what your doing while the pro's do it their way.

Oh and for future reference, you generally use more than one rendertree on an object not one huge one, only newbies would do something like that.

Also, you might want to study the human body, I think you'll find that wrinkles, pores, moles, scares, birthmarks, coloration etc. of the skin don't match on different bodies and are for the purpose of re-creating it digitally, random looking.

Toodles.
 
heh, have you ever experienced what you are suggesting? made a highpoly model and completly textured it using procedurals?
Facial features, from person to person, of course wont look the same, but the tone in the forehead is different then the tone in the nose, then the tone in the cheeks. you have little control in that in say a sss map that is procedural. texture maps are all about control, and control leads to better results. As for a giant shader tree, you would most likely end up with one doing evertyhing with fractals. i never tried projection mapping before so i cant comment on it, or how useful it is in the baking process.
ps iv been using max for 4 years, maya for 2 and xsi for <1, and i havent touched max or maya for more then 10 mins since last year, im sure the same can be said by you. Since you know so much i'm sure your way is the pro way, but ill stick with the "noob" way since it yieldsbetter results.
 
thereisaspoon said:
heh, have you ever experienced what you are suggesting? made a highpoly model and completly textured it using procedurals?
Facial features, from person to person, of course wont look the same, but the tone in the forehead is different then the tone in the nose, then the tone in the cheeks. you have little control in that in say a sss map that is procedural. texture maps are all about control, and control leads to better results. As for a giant shader tree, you would most likely end up with one doing evertyhing with fractals. i never tried projection mapping before so i cant comment on it, or how useful it is in the baking process.
ps iv been using max for 4 years, maya for 2 and xsi for <1, and i havent touched max or maya for more then 10 mins since last year, im sure the same can be said by you. Since you know so much i'm sure your way is the pro way, but ill stick with the "noob" way since it yieldsbetter results.
again with the "completely with procedurals" I never said completely use procedurals, your not limited to JUST procedurals

*yawn*

As for projection mapping. You project an IMAGE onto a surface from the view of the camera, allowing you to project it from any angle, like you would a spotlight with a projection. a good example of using this would be the face maps, you just line it up and there it is, viewing the result in a realtime render of it (XSI's render region or Lightwaves FPrime)

As for experience. mmkay then

relevant 3D apps
Lightwave since the Amiga days, around 1989/1990 - say 10 years to be fair as I haven't used it _every_ day.
Softimage - 2 1/2 years
SideFX Houdini - 2 years
Maya - 3 years
3DS Max - 1 year (split up over years as I've never liked it much and only ever used it because I had to, certainly not by choice)
ZBrush - 1 year (still getting familiar with it)

relevant 2D apps
Photoshop since v2
Painter since v5

Taught CG design and Lightwave course at Bradford University for a short period. Mostly because I wanted the practice of speaking infront of large groups.

+10 years in the Film/TV/games industries, inc. freelancing, short stint freelancing for ILM.

over 30 games, mostly educational titles though so nothing to write home about there lol, some other games, Cannon Fodder, Alien Breed, Sensible Soccer. Was all a long time ago, the games industry wasn't as profitable back then as it is now.

Used to do commission works, held an exhibition in Edinburgh years ago for charity, sold all but six or seven paintings (which were shit so im not surprised nobody wanted those lol :p)

Ran a couple of CG related companies until I packed it all in last April to take a break from the industry in general for a few years, which is what I'm doing at the moment and loving having a break.

While I'm certainly not one of the best artists by a very long shot, and would never dare to assume so, I have had the pleasure of working with some of the best talent out there, learning from them and watching how they work, what they do, the tricks they've picked up over the years and years they've been doing it. Some of them were coming up with how to pull off various effects before computers could even manage basic 2D graphics.

I think its safe to say THEY know what they are talking about, right?

Or do you still think the guys and girls at EdenFX, Weta, ILM, Dreamworks, DigitalDomain, Blur, ARK etc. and so on don't know what their doing? I reckon the awards those lot have between them for outstanding work and the amount they can now charge for the simplist job prove otherwise.
 
UVW mapping can be pretty obnoxious at first, but once you really learn how to think about how it will map as you model it, it's not too bad.
 
Back
Top