Doom3world

continuing on from the buggy 3rd page of my wip topic

thanks blackdog, and I could ul the model`s .blend file if you have blender or perhaps export it to a format your modeling app can use?

either way I would appreciate seeing how you or anyone else would do the seams on models like this, as I accept I`ve still things to learn with modeling

Sure, .blend is fine for me. If you want anybody to be able to see it, .obj is a better choice.

I'll be a few hours though, gotta grab a few zs.

thanks and I`ll sort the files out in a few mins, and no problem if you can`t work on it tonight. I`ll just get on with the mesh editing for the other sections of my rock bridge.

I`ve been offline for a while but have since set up an account at pcgamemods.com

unfortunately I couldn`t export to .obj from blender without getting this error, does anyone know what I`m doing wrong?



but here is the zipped up .blend file. I cleaned out all the other models, materials, textures etc that I could find, and packed my rock uv test image seen in the blender images.

ratty_terrain05_03.zip

zombie13, sorry it took so long, but I`ve just tested inline 1 on a couple of my .ase models.

if there is any vertex colors like the black to white fades we use to control texture blends, then it changes the some of the white to greens or blues, which I can`t see where they are coming from since my rock texture and the d3 dirt I was using don`t appear to have either of those colors in them.

I also tested the "textures/common/collision" material on some of my .ase models and it doesn`t seem to clip the player or their weapon projectiles, but some of d3`s clip materials did work, also doing a window context search of the original d3 mapobjects resulted in only .lwo models to use the collision material.

"Native" exporters break all the time in blender. This seems to be a more reliable .obj export/import script for now:

http://members.westnet.com.au/GeoffGole ... dif232b.py

At any rate, I have the mesh and I'll ul an unwrapped version in a bit.

understood and thanks for you help black dog

OK, I had some serious trouble with that. After spending much more time wrestling with blender than pushing uvs, here we go:

http://members.westnet.com.au/GeoffGole ... PPED.blend

I suggest that you append the object from this file rather than just opening the .blend, since it's not likely you want all my wierd settings. (Putting user interface settings in the same file as mesh data is pretty stupid). Some stuff has been ripped out to make it easier for me to work with the mesh, but don't worry about that - just take a look at the seam placement and uv layout.

First thing to notice is that I don't just use long, continuous seams but also some short one-ended cuts. That allows lcsm to push uvs over those seams, which helps relax stretching. In a few spots that results in mapping that is not 1 to 1, but that shouldn't be a problem when you are using an existing world texture (it's a big nono if your uvs will be used for baking/lightmapping/painting though, keep that in mind). I pushed those spots to where I thought they would be difficult to see - without knowing what your lighting and mesh position is going to be in game it's a bit difficult to know, so you'll want to make that kind of decision for yourself.

Secondly, the unwrapping here is pin-free and based entirely on the cuts that I've made. The only change is that the resulting uvset has been rotated and scaled a bit. If you want to play with the seam layout, just add more and unwrap again.

Lastly, if I was modelling this myself I'd give some thought to other methods of getting rid of seams. There's a big nasty one down the middle of the bottom of this mesh, maybe I'd move that one to where I know there'll be shadows or model a boulder to go on top of it. You pointed out yourself that you could arrange for seams to go where you've blended in the dirt texture, that would be worth looking at. You could also add another crack on the other side of the bridge, as those are perfect spots to hide seams.

thanks black dog and this is very interesting

firstly I didn`t know we could have seams that weren`t continuous edge loops. also great idea to hide the seams in cracks, I think I can do that quite easily from now on, and a few more beveled cracks will make the rocks look more interesting. and as you also suggested, covering them up with other models like boulders.

I might still use the odd pin, but I think your method will save me a lot of time in the future, as my uv maps were getting messed up a lot each time I added in more polys like the cracks, because the excessive pinning I was using was restricting the uv map from automatically shifting around while doing a lscm to compensate for the new polys uv`s.

so was this the first time you`ve used blender or was it just my .blend file that caused you problems? either way, I very much appreciate the time you spent on helping me with this

did you know we can use the file browser and deselect the "load ui" button at the bottom of the file browser?

this means that each blender user can retain their user prefs but load in other users project data (mesh, textures, materials etc) although some of the global prefs, like the paths blender looks for textures, are saved into the .b.blend file. although even that can be backed up or over written with global prefs for different projects we might be working on.

I do agree though it did seem stupid at first that blender saves some user prefs into the .blend project files, but I`ve got used to it.

which is the main modeling app you use?

I use blender extensively, but rather narrowly. In other words, I stick to a certain subset of features that I know very well and ignore the rest. Loading your .blend caused a few issues that I wasn't familiar with, so figuring them out took some time.

I did know that there was an option not to load the ui, just couldn't remember where the hell it was.



That's the main disadvantage of pinning. From what I can tell it's a more useful feature when doing things like unwrapping a character model, but I rarely use it on world geometry.

understood black dog, and I hope I didn`t sound patronizing when telling you about "load ui" button?

I`m now using blender more then any other editor, but every week I learn something new about it or realize the potential of a feature I vaguely knew about previously. if users can get past the very steep learning curve of its interface, then it`s amazing how feature packed it is, would you agree?

I`ll try to not use pinning for my world models, but at least it is something I can fall back on if I struggle with the seams.

am I right in thinking that as long as they are well placed and hidden we could have many seams on a model, especially the one ended type, and the result would still be neat looking uv map, so instead of adding in more pins as I have been doing I add in more seams, yes?

I haven`t done any of the new uv mapping, so this image is just of blender`s solid view mode, but it shows the rock bridge meshes spanning across the main cave.

Patronising, no. Please never hesitate to tell me useful stuff. Maybe I already know it, maybe not.

Blender is certainly full of stuff to learn. The thing is, I have no need for most of it and I'm more interested in learning to make assets than learning Blender. My focus is entirely on modelling and unwrapping and as such dealing with the complexities of the rendering or materials systems just isn't appealing. Out of all the clever things that got added in 2.42, for instance, the only ones I have real use for are the uvlayout export (decent at last!) and the very useful merge to target vertex. Now they just need to make the knife tool snap to the grid.

While it's possible to have many seams on a model, I don't know that more is always better. The idea is simply to have the surface look as contiguous as possible without stretching, ideally using as few seams as possible. Learning how to do that effectively is really just a matter of unwrapping a few models and seeing what works.

Looking Good Ratty.

When it comes to UV-mapping in Blender I suggest you try the "Live UnWrap Transform" feature. This allows the skin to change as you move pinned faces making it very flexible to modify skins effectively. Although it wont move pinned faces so its best to have maybe 4 or 5 pins.

Depending on the size you could just leave the standard uv-map (Each quad/tris uses the entire texture) although if its not a repeating texture it will look silly..

Cant wait to see more.

cool and same here

although designing and make assets for games is my favorite type of work, I have another large, non game related project, where I`ve had to learn with help from kit89 the material, lighting, animation and rendering side of blender. although most users wanting to make games assets can ignore those extra features in blender.

over this year I`ve grown a lot in confidence with building meshes in blender and love using it for that purpose, as in comparison it makes all versions of radiant I`ve tried of the years, a pain to build game levels in, with the exception of straight walls which radiant is still quicker and easier to work with in my opinion, then blender.

but learning to optimize models for games, and as you`ve just helped me with understanding better the uv mapping, is still something I need a lot of practice with.

I think we can probably help each other and certainly other users in here, with the tips on using the blender mesh tools, but as I`ve still not upgraded blender from 2.41 to 2.42, I`m not familiar with "merge to target vertex" is it different from "merge verts to cursor" available in older versions of blender?

regarding the (edge cutting) knife tool, I always snap the resulting cuts manually to grid using the "shift+s" popup menu, but I first have to either zoom in the 3d window to a small enough grid, or keep another 3d window on screen with a small grid size, so the snapped verts are fine enough, other wise the model comes out looking to blocky. is this similar to how you use the knife tool?



understood and thanks



thanks kit89, and I had already been using "lscm live transform" which is a lot better then lscm on it`s own, if we intend to then tweak the uv`s by hand, but I now think black dog`s suggestion of first making more efficient seams on these game world models, will be the most time saving method, although I`ll still be using lscm live and pinning on non world models.

thinking about it, I`ll probably still use just a couple of pins on the game world models but only so I don`t have to scale, rotate and align the uv maps each time I use lscm to unwrap the model, but I don`t intend to tweak their uv`s by hand.

I'd suggest looking at this tutorial. It doesnt go into the basics of uv-mapping but it does show you how to get fairly seamless skins effectively. You may find it usefull.

Link to Tutorial

thanks kit89, that is a good tutorial, although it still involves manually editing the uv`s which is what I`m trying now to avoid on this project, because of the amount of tweaking of the meshes, particularly after viewing them in the d3 engine with it`s lighting and texture blending.

still though, I`d be interested to see any other uv tutorials you find, as it all helps us better learn what is possible

In my experience, UV mapping can and generally is just as time consuming as modeling itself.

this seems to be the case, and it`s almost as if we build the model once in 3d then again in 2d for the uv mapping.

black dog, I`m getting on well with using your method of placing seams onto the rocks

but I wonder how you would unwrap the "horns" or saber type "teeth" that are in the original hell maps and my version of them, like in this image.