View Single Post
Old
Mr Kidnapper
Senior Member
 
Status: Offline
Posts: 442
Join Date: Feb 2011
Default 02-17-2012, 08:49 PM

There's two ways I do texturing. That's in ZBrush and in Photoshop. Sometimes I do both. First I make a UV map. Making a good UV map is harder than you think. At least, eventually it will be harder than you think. You have to ask yourself a lot of questions before you start building. "Is my model going to be used for a game engine?" "Will it be fine if they just paint the polygons?" "Am I going to edit it in Photoshop?" "Did I make my cylinders really hard to UV like a douche?"
There are a few global things that go with making UVs: Make sure UV lines don't intersect, even if their faces are connected in the real model. This causes problems with the renderer in some situations. You might want to stack UVs on top of each other in the case of very common UVs. Make sure UV Islands have sufficient spacing. This is called padding. While giving them enough spacing, you also need to make sure you use the most space possible to ensure efficient use of your texture size. UV Islands are chunks of UV, for example if I took a high poly sphere, selected a random bunch of connected polygons and split them from the main group in the UV editor, that would then become a UV island. The rest of the main group as well as the main group before the split are also UV islands. They're probably named that for their variable shape, or lack of shape in some cases. The idea is to make sure each face has their own space in the texture to show up and each UV island has their own elbow room as well.
If your model is used in a game engine, you must do a few things: Prepare your model for normals to be made. Normals, or bump mapping (Technically different but the same for most people) are essential. They add details by changing lighting patterns on the polys without adding any. That being said normal maps have their own OCD-like kinks to them as well. You must separate UV islands and pad them. No exceptions, even if the object in question is essentially a bunch of clones. It will mess up the renderer and create seams(How to stack depth information? What do!?) There are also a few things that every model must go through: No chamfering, 90º angles are your friend. Chamfered edges make it harder to UV your model and they add unnecessary detail to your model that normal maps would have done anyway.
Make sure you make your high poly model have good indents. 90º angles on high polys don't bake and don't add detail.
If it's fine that they just edit the polygon's color then you've hit easy street. All you have to do is make sure every face gets a space of its own, and it's fine to stack clones. They probably aren't going to do anything fancy like bump maps either.
If you're editing it in Photoshop or some other image processor... make things easier for yourself. You should also make every face is properly to scale with each other compared to the real model. Every filter that goes through Photoshop affects the entire texture evenly and if you don't you might come out with a head that has a piece that looks as if it were stretched to hell.
If you did some silly things like tessellating only a part of an object then all you've done is make it hard for yourself to UV the object. You might suddenly come across a row of triangles in the middle of your squares and then life gets hard because it won't UV right. In fact, it's because of silly things like this that some people recommend you do the UVs while you model to avoid this.
In the event of actually making textures, more commonly I export a model into ZBrush and paint directly on it using many subdivision levels with millions of polys at once for extreme detail and then export the UVs as a texture to add onto my low poly. Needless to say I theoretically have infinite resolution. This is called Polypainting. It's also notable that I should say you need a graphics tablet to properly use ZBrush, unless you have extremely steady hands or something. For example in ZBrush I might import a gun and severely increase the polycount with subdivisions. I may use ZBrush's polypaint feature to add various scratches and grip patterns to weather it, so that I can see the effect of my additions in real time. This is most beneficial when used in addition to alpha masks, in which you use a semi-transparent texture to control your painting to specific areas. You might use alpha masks to paint an effect such as a "metal floor" pattern onto a steel box. It is notable to say that most of these alpha masks are made with Photoshop using Bezier (vector) curves and such things.
For models that aren't required to be so detailed, I use Photoshop or my favorite free alternative, Paint .NET. It's obvious why I would use Photoshop as it is the industry standard, but Paint .NET mostly because I dislike GIMP more than I do Blender.
Some artists are inclined to design their texture from scratch with nothing but a reference picture to stare at—not to take from—to know that they're going in the right direction. This is good for extremely high resolution textures—if your texture is made up of nothing but vectors, noise, and Gaussian blur then I imagine you could take your resolution into nearly five figures to later be scaled down. In fact, it's often best to make your work high resolution and then scale it down. For things such as a brushed metal texture they might take a gray color, add monochromatic Gaussian noise, add motion blur, and use the offset filter to make the texture tileable.
Other artists use real textures from real buildings, rocks, and cows for their work. There's nothing wrong with this, in fact you may be inclined or even forced to use this method to obtain extremely realistic results. They also stack these real textures to form even more real weathered objects to emphasize age and whatnot. After all, there's nothing more real than reality. The only problem with this method is that you may find it difficult to buy yourself a nice camera (And tripod) for good enough work. For everyone else it may just be that they don't happen to have Chernobyl down the street to take a picture of. Of course there's the internet for this, but who would pay someone for a picture that wouldn't be as good as one you made yourself?

Edit: Ohei post fits my screen's height. NOT ANYMORE.

Last edited by Mr Kidnapper; 02-17-2012 at 08:51 PM.
   
Reply With Quote