I am working thru a lower poly asset (current texture is thoroughly temp!):
The model is relatively simple, but I’m finding unwrapping this to be relatively difficult. I’m really unclear how to get a good map to work from. There are enough different shapes and, tho’ regular, nothing is square or easily projected. I’ve tried a few different approaches, and most of the tutorials are merely showing me how to unwrap a cube (or an organic shape…). This seems like it should be relatively simple, technically, as most of the shapes are regular, but keeping the uv’d textures looking good, not stretched and integrated is proving … difficult.
I’d be interested in any advice or pointers to any tutorials to follow as this has proven to be a bit of a road bump to my successful Blender-an-dizing.
My trade-off for advice would be that if I can get something to work well, I’ll make a tutorial out of it and post it here!
In this case this method will work very bad, since it’s not a simple pipe. This will result in having a very stretched UWMap. I am not even talking about texturing it.
Unwrapping is one of such things in 3d that consumes a lot of time (especially in case when it’s required to get a highly effective UWMap for mobile games, for example).
As always, there are no strict rules of how to Unwrap and it’s really depends on final look of a model.
There are several tips, hovewer, that can help to understand some of concepts:
Even a very complex model consists of a lot of simple objects. For example, in our case, we have a sphere at the very bottom, cylinder on top of it and so on.
It’s always good to think of which parts will have the same repeative pattern and, well, repeat their texture space. I’ve illustrated this on image below (so we’ll do UV once and just repeat over the object where’s required).
The less seams we have the easier it is for us to hand paint texture and notice seams in game. Quite obvious.
Can’t say anything about slight stratching since a lot of professional uses it, but I am not a fan of it.
i saw his prototype and i know (or think that i know) that his Outpost wouldn’t be seen very close up. The “unique” unwrap + repeating textures will pump up the vert-count massively. Stretched UVs are not that much of a problem. The texture work will be harder i agree with that. So i guess best bet for this objective would be a mix of some kind of “cylindric” unwrap and unique islands.
First: Why would a “unique” unwrap + repeating textures will pump up the vert-count? As unwrapping doesn’t effect the model’s geometry, is this UV verts? (I’m not sure even what I’m talking about…) [See edit below]
Second: If I approach this in a more “tiled” solution, I could imagine projecting from view if I could get the camera perpendicular to the face of the UV and then find a way to copy and paste the UV information per tile… Does this make any sense as an approach?
Or am I just trying to avoid adjusting the UVs by hand? I I must just get a base UV for a face and then hand tweak each face, then I’ll just have to realize this and jump in.
And closeness? You can scrape it:
Ultimately, once I get the concepts under my belt, I’ll need to texture the ship properly as well…
[edit]
Cross post re: the vert count! I’m starting to understand now. Doubles the vert count on the model’s mesh? Or just the UV vert count? (Again, I’m not sure exactly what I’m talking about… )
Any vertices in your original model which fall on a smoothing seam, or a UV seam, has to be duplicated in the mesh data. This is due to the way in which the data is sent to the graphics card. So, the more smoothing seams and UV seams you have, the more vertices will be reported.
credit goes to Duck.
so “smoothing seam” you may better understand as a sharp edge in blender. would mean that every vert that is on a hard edge or on a UV seam will double.
So using the repeating texture system/style would mean many doubled verts.
for “projection from view”…meh, i totally dislike stuff like that. your UVmap would be even more streched and distorted than/then (<—what is the right one here?? ) with the method shown in the link i posted before.
Yes, ultimately a “project from view” would be horribly distorted unless one could place the view camera perpendicular to the face procedurally… But my experimentation shows that simply “unwrapping” one face on it’s own may do just this internally. A single face unwrap seems to preserve the correct shape… but now I understand your worry about vert count… Hum!
Thinking.
Won’t get to experiment until tomorrow at the earliest.
Okay, engine doubles vertex count automatically. Doesn’t do any harm. Target platform is not Nintendo DS, so no one really count every vertice.
Yes, it’s possible to manually adjust UV positions for tiling or map UV information from same parts (in case if they’re sort of equal: vertex count etc.). But usually the fastest way is to divide object for parts (and delete ones that should be repeated), unwrap them and then duplicate to get a model again. Of course, after this action merge mesh and vertices to get model as it was before but unwrapped.
At least that’s how most of professionals are doing it.
You know you could always use overlapping uv’s. Figure out what needs to be unique, what doesn’t. Areas that can be overlapped can be scaled up/down to maximize uv space.
Rather then looking at this as one object, look at it in sections, you have cylinders and spheres, and I’m not entirely sure how blenders uv mapping works, but I suspect you could project the shape if a cylinder in separated parts (break up the object if necessaries to make this happen, sometimes this is also the easiest way to get your uv’s correct on complex models) once you start looking at the core shapes separated, you can break the uv’s down in layers that make sense.
my first advice would be…re-model it sounds bad, eh?
you started your model with a 14 side cylinder and just extruded and scaled it. that means that the big body(radius 16 meter) has the same faces as the tiny poles (50cm)
my advice would be to like Larv said, seperate parts. the big body…go for a 18-20 sided cylinder, the poles 6-8 sided. your polycount wouldn’t increase that much, maybe you would even be able to reduce it.
The higher polycount on the big parts would mean less distortion.
Yes, Schlumpfsack, as you guessed - the basic shape started with a cylinder that was extruded up and down, tapering/scaling the current circle/ring of verts.
Remodeling wouldn’t be that hard. It’s not a complex shape.
I won’t get much of a chance to futz with this until the w/e, but I’ll give it another go.
slapped something really fast together. I hope you see how i modelled and unwrapped it. While it isn’t perfect (distortions on some curved areas still) i hope it gives you an idea.
) with the method shown in the link i posted before.
sounds bad, eh?