Do we really need to be Uv mapping in 2015?
polycounter lvl 5
I`ve been thinking today whilst spending 2 hours uv`ing an fairly simple organic object in Maya, why do we still need to be doing this. I`ve been uv`ing for some 20+ years in games and i`m getting to the point where it feels like i`ll be doing it for another 20 years and I may just become a truck driver instead to save the grief.
I know its efficient to do so for games but surely someone clever can create some kind of PTEX mapping for game engines. Imagine how much time it would save us artists.
Thoughts?
I know its efficient to do so for games but surely someone clever can create some kind of PTEX mapping for game engines. Imagine how much time it would save us artists.
Thoughts?
Replies
I`ll check out iPackThat thanks.
It was a series of meshes I was mapping, spheres and cages that were boolean and to be mapped with a mesh texture. So it had to be clean looking and all the texture join up correctly. It was one of those jobs that you think will take 15 minutes and then by the time you get it into the engine without any errors (uv/lightmaps) ie unreal its 2 hours later.
Imagine if you could just model and then just texture in 3d and your good to go. And lets face it nobody wants to UV, everyone wants to model and texture.
what do you like about using 3d coat? I have it but never used it for uvs. I find uvs a minor inconvenience not a big problem.
UVLayout, iPackThat and 3DCoat
If you're still unwrapping in Maya, then that's your problem.
It's just easy to quickly and interactively cut UV seams (especially for hard surface).
I can do seams by hard edges (let's say 60 degrees), put a few more cuts in, click unwrap, check if there are any problems. That right there is like maybe 5 minutes at most. Then, I just export to max, UV seams to hard edges, do some test bakes real quick with substance for quick iteration. Then, when the test bakes are fine, I can do a proper pack in 3ds max/IpackThat.
Maybe there's a quicker way to do it but the method above takes maybe 1/8 - 1/4 of the time compared to a traditional workflow.
on top, for retopo you get results almost instantly and on the fly
I still wonder to myself it PTEX will be the next thing to insert itself into the workflow of what a real-time engine will be able to read in the coming future.
IPackThat can be super efficient and its about as good as a current procedural packer can get I feel, but its still not "clever".
Then again, I'm the rare guy who likes to dedicate a long time to his UV's, and enjoys it because I tend to get nice results I think.
Depends on the model, but usually. Unless it's something organic that has a very continuous surface and I don't want nasty seams, I almost always explode stuff or export separate chunks of a model and bake them individually.
I'd really like to have a smart baker that allows you to visually pick which parts of the high poly should bake to which parts of the low, and then calculate ray hits based on mesh continuity. Unfortunately, I'm too stupid to write that kind of tool on my own, and I don't really know anybody who has the necessary skills to make it happen.
Games with great performance are almost always the ones sporting the most carefully built assets. I think that's no coincidence
Substance Baker!!! I've baked some pretty complex hard surface stuff with no problems.
No exploding or cages! Just match by mesh name.
I also agree with Pior. Usually after my test bakes are finished I'll go back and really try to straighten as many shells/optimize UVs as much as possible. It's pretty fun with IPackThat because you can see the amount of space wasted so for every asset I try to get that percentage as low as possible.
So far I just make cage and bake out each item individually. #_# 20+ subtools and I'm dead.
Ah sweet! I'll have to give it a try.
As for if we need them, yeah. Even if some format or tech that didn't need Uv's took over, I think having the actual mapping would retain its usefullness. UV's are the source of many major resource/asset/time saving strategies.
Modular assets, texture variants, shared maps etc. Without UV's that stuff isn't going to be as easy or possible.
What you've mentioned you can do with Textools in 3ds Max so how is 3D-Coat any better?
Not much can go wrong if you know the tools.
That being said the current school project i'm on we aren't using UV's.
Anything that need a texture or a change in the material we handle with unreal's material editor, stuff like world space blending ect.
Our art style can support this kind of workflow though, not sure if its going to fit every art style.
I sincerely hate unwrapping complex mechanical hard surface stuff. You knw the kind of models that have extrusions and holes everywhere...
I would hate to UV map this for example
http://www.hdwallpapers.in/walls/bumblebee_in_transformers_4_age_of_extinction-wide.jpg
Uving for film is so different than Uving for games... I actually find film pipeline uvs easier than uv layout for games. In film it's pretty much just making sure your uvs have less stretching and are aligned for repeating textures and have equal texil density, and most places have really nice automated tools to help you out with those things. Game uvs have to be hand crafted to take advantage of mirroring properly (no visible seams is harder than you think!)
Well that's because you use the dark magic that is Modo's UV Toolset. With 901 it became even more stupid simple.
(https://youtu.be/7PHdNBxJMHk )
Organics is pretty easy, especially if you use Greg Brown's macro magic. Ends up being a one click solution most of the time.
http://www.polycount.com/forum/showthread.php?t=151832
Holy crap. I hadn't seen this video before. Guess it's time to upgrade again!
I don't care too much about the UDIM stuff, but literally everything else in that video is a laundry list of features I've been wanting for a while.
Aka I used to hate it but now I love it as my mental break. Automation tools for unwrapping is cool and all and I do use them, but I still sometimes like to do it a bit more manually/I try to have the islands created, oriented, and positioned in a manner that makes it easier for me to texture and quickly locate versus just being concerned about space. It's just so relaxing rotating and moving around
That said, I am curious to give 3D Coat a try. Do you think it's worth it to buy if I already have Max, Zbrush, and Substance Painter?
And I imagine if it had better packing code in like some plugins I've seen it would get even faster.
An example is UV layout. People get criticized when their UV map contains lots of empty space, or if they don't overlap/mirror certain UV islands. I get the logic behind avoiding this but in a video game, who cares?
You're never going to see how a person UV mapped a model while in game. You'll only see the final textures + shaders.
Technical limitations are still very much a thing. Everyone wants the most beautiful artwork and the game to run at a steady 60 fps as well!
But the people who play games aren't going to notice all that work that went into optimizing that UV island. They're only going to be impressed by what the final visuals look like.
Just like how, when I go to an art museum, the first thing I'm going to notice is what each painting looks like, and not whether the artist had used a ruler to measure every line.
but people will notice if it runs at a low fps, or hogs all there vram due to stuff taking up way more texture space than it is actually using.
But ... if you're efficient with your UVs you get better texture resolution for the same memory cost. Why, as a professional, would you not pursue that?
Your point is none, just look at the reception of Rage, arguably one of the best Looking games at that time. Players just complained about the texture resolution in closeup.
Which was a direct impact of megatexture a system for great unique texturing, but one that takes a hell lot of memory.
While here megatexture was a reason, a game with efficient UV usage will definitely have a crisper texture resolution than one where the artists didn't care enough to optimize properly.
What my complaint is that in terms of final presentation, you are never going to be confronted with an actual UV island in a game. Again, how many gamers or game reviews are out there praising "man, this game has nicely mirrored fruit textures"? It's this step I wouldn't mind seeing removed from the pipeline if it was possible.
Regarding memory usage, I believe diminishing returns plays a role here. The impact of having 100% efficient UV's is likely a small hit on visuals then it was perhaps 20 years ago (on consoles were 1mb of memory had to fit everything, compared to today's consoles that have 8GB of memory).
This is the definition of dated. Something that proved very beneficial in the past, but doesn't quite yield the same benefits today. Does not mean this techique is wrong, only that by today's standard, we should strive to replace it with something better.
It is still like that though, lots of optimization has to happen to run on the limited memory on consoles and mobile. The limited time good uv work takes is worth the gains by a pretty big margin. Why use 2k map if with good uv work you can accomplish the same in half or a quarter of the res. Most games have hundreds to thousands of textures so wasted texture space adds up fast.
Also what about cases where good UV work actually improves the art workflow, like mapping to existing trim or decal sheets. Or simply through creative UV working creating a new prop that only uses the texture space in a pre existing texture.
What they will see is, that the game runs smootg and is nice and sharp looking. And this goes directly back to proper UVing.
But it is not only about memory optimizations, but also about workflow. Unless you geberate your textures, painting will be a lot easier with nicely straightened UVs.
Also if I would not give a damn, i would possibly remove UV mapping from my skills in my resume :P