Hello folks, this is the first time I've posted a project here; it's pretty rough at the moment and I'm looking for some help moving forward into texturing, which is something I'm a touch shaky at (at least when it comes to texturing for game assets).
The project is of a small game environment; in this case a subway station with a few moving trains and a single platform. Here are some quick shots of the scene as a whole, and a few details:
Scene OverviewHid the roof/tunnel geometryCloser view of the escalator/exitSection of trackTrain car
I've done my best to keep the number of polygons low, while keeping the flow clean so if I do need to go and add more detail later it shouldn't be that big of an issue.
My main problem right now is starting to texture everything; I have most of the pieces UVd, but it was mostly through basic planar mapping and I am unsure how well it's going to mesh with a tiled texturing approach. I see a lot of texture maps for game assets that are very condensed with sections of texture to be reused many times over (
example and
example two), and I don't have a clear idea in my head about how these go together with the laid out UVs, or where to start the process of creating the diffuse maps.
Note: A texture map laid out similar to
thismakes a lot more sense to me at base, since you can see where all the UVs sit on top of the texture; it's adding the concept of tiling that has me stymied, somewhat.
Replies
For the examples you put up, usually the texture is done first. Then when you go to do the UV layout of your meshes you just lay them on top of the image. Instead of unwrapping it like a prop and THEN texturing like the phone you put up.
Are you going to put this in an engine?
Yes, I'd like to drop this into Unity or UDK; most likely Unity, since the school already has it installed on all the machines. This is something of a "final project", and for once I'm trying to keep the modeling simple so I can get through texturing and, more important to me personally, lighting. I'm a lighting guy by experience (long time working in theaters) and am excited by getting to do it in a virtual environment, but unfortunately, I haven't had a lot of opportunity through school; suffice to say the program could use some differing emphasis.
Check this thread out, it's great http://www.polycount.com/forum/showthread.php?t=86588
Edit: Oh about the engine, I was going to say to import what you have into the engine and run around in it. Make sure the scale of everything feels and looks good. Don't want to get too into the detail and then realize everything is too small or big.
Edit: Or to phrase it differently, in order to use tiling textures across the piece I'm going to have to cut up the geometry into, essentially, tiled sections.
Non-griping, actual thought out response: how does the modular approach balance with keeping the number of polys in a scene low, for a game setting? Is it more a matter of minimizing poly-count for a modular block, and then the fact that the block's simply being instanced means that the engine only has to draw the thing a single time?
A lot of what the guides are saying in terms of working on a grid is starting to make a lot more sense; however, I didn't keep that in mind while modeling here. Would it be worth taking the time going through and breaking the scene walls (which are the only truly non-modular bits) down and re-aligning everything across the scene (re-scaling props, etc.), versus using a "prop" methodology for the scene walls here and Uving -> texturing them?
I'm reading through a PDF that seems to be explaining a lot of what I want to do; however, I'm a little confused because it doesn't appear (to my untrained eye) that his buildings here are modular in a vertical sense, but he's still talking about tiling the texture vertically. I'm looking specifically at the cylindrical tower in this case.
Games nowadays aren't too limiting on polycount, it's textures that's the problem. That's why it's better to cut your mesh a bit and reuse the texture then have a bunch of textures being rendered on screen.
Although his building wasn't necessarily made by lego peices, he still mapped it according to his textures. Instead of unwrapping it and then taking the UV's into photoshop and texturing it uniquely.
Imaging if it was a small city scene, with a bunch of different buildings and if they were all using unique textures it would start to use up too much memory.
Also, remember to give your modular pieces a second UV channel for the lightmap.
I understand that the tutorial isn't uniquely texturing the building(s); what has me confused is that the tower's main body (which is, so far as I can tell, a simple cylinder shape with maybe 16-20 vertical divisions) has the rock wall texture tiled three times vertically. If the rock wall was the only thing on the texture sheet I know how to achieve that effect (scale the UV of the wall up far enough that the texture repeats), but given the packed texture map where the rock wall texture only occupies one corner, I'm not sure how it was done.
Reading several other threads here, I'm beginning to suspect that taking this whole scene through to a game engine might have been biting off a touch more than I can chew, so I'm doing a bit of re-evaluating. And trying not to get thoroughly discouraged.
This is how I would've done it.
I would just add the cuts in the necessary places. Hopefully that image makes sense. It's technically 5 cylinders stacked up on-top of each other. You then weld the verts to make it one cylinder.
I think you should stick with it. If you think it's difficult that's even more of a reason to try it out.
It shouldn't be though. If you can model and unwrap something you should be able to do this.
Also, if the image I posted is not what you were talking about, my bad.
I've been trying to wrap my head around all this most of the day, so I'm going to cook dinner and relax with some environment painting. Tomorrow I'll push to get everything cut up and gridified and go from there.
In terms of props on the grid (for example the subway train, track, arches, etc.), I'm just looking to unify scale by making sure everything fits into integer multiples of the base 1x1 grid size, correct? For example, a wall section might be 3x3 "units", the arches 2x4 units, the train ZxYxQ units, etc. Is making the train itself modular (within each car) something to consider as well?
But for UDK 16 Maya Units is about a foot. Since their characters are around 96 units. 96 / 16 = 6.
So you usually want to keep your grid a bit big. So you're not moving pieces inch by inch, instead you're moving them by feet.
So it all depends on what Unity / Maya set up you have.
Also you want to keep your pieces in a power of two grid so they all line up with each other.
To be honest, I'm having a hell of a time trying to re-space and fit everything, particularly the wall which the tunnels pierce; each tunnel chunk is 512x512 at the square portion, but I'm ending up with odd-number wide gaps to fill and argh. There has to be a better way to deal with the larger arch shape of the roof, the arch of the tunnel ceilings, and how they all fit together, but I'm just not seeing it.
So, off I went into UV land:
I got the exterior split up, hopefully tomorrow I can take care of the interior and start texturing/arranging everything. My goal is to use a sheet for the interior and a sheet for the exterior.
I'm pretty proud, since I haven't done any UV packing before and I think it came out well. Relative scales are a touch off perhaps, but each individual shell is pretty close to undistorted, so hopefully textures will go smoothly.
Meant to get exteriors done today too, but alas. Tomorrow!