HowTo:FullUnwrapBlender

• Introduction
• Clean up the mesh
• Materialize it
• More cleaning up
• Think Hard First
• Placing Seams
• Reorienting the ship for unwrap
• Entering Potato Mode

Placing Seams

It's always easier to start planning from the biggest parts to the smallest. Our biggest part is the fat fuselage. The fact that it is almost round, tells me that the best way to go with it will be using "Cylindrical Unwrap. That is, once we get to UV unwrapping, we'll look at it from the front, and from the Unwrap menu choose Cylindrical. This will roll the hull out like a tape. Of course, this wouldn't work too well for the tower and the wings, so we're going to "cut them off", texturing-wise, by placing seams right along their welds. To place a seam, you have to be in Edit mode, Line Select; so you select a line along edges (Alt-right-click and Shift-Alt-right-click are often useful here) and Ctrl-E -> Mark Seam. By default, seams look brown in the edit window.

Note that you don't need to place seams along split edges. Split edges already "imply" a seam. So, if you did a thorough job of splitting your smooth groups, you won't need too many seams.

So, let's start by splitting off the wing. (Note: I'm assuming you know about the Mirror attribute, and that your model is mirrored in X so that you only have to work on one side of the ship, and the other side automatically mirrors what you do. If you don't know what I'm talking about, you really need to do a bunch of beginners' Blender tutorials first...). Okay, let's see... Problem! The engines are in the way! How are we going to draw the seam?

If the engines were a separate object, we could move them to another layer, but now the whole ship is one object! Well, there are other ways. This should be obvious, but I'll spell it out just in case. We made the material of our engines "Stainless", so we can just do a selection by material...

... and press H to Hide it...

Avra Kadavra! The way to Reveal things we hide will be the subject of a future tutorial... :D

Allright, so let's get started: Top edge of the wing's weld:

You might ask "why do we pick *that* line, and not, say, the one below?". Well, there's a reason: the strip of quads below the line is the actual 45 degree bevel strip. I prefer to leave that bevel strip as part of the wing, rather than as part of the hull. Why? Because, either way, we'll have to LSCM the strip so its projection doesn't get compressed. So the narrow projection of the strip will kind of flare out a bit as we flatten it. But while this is okay to do at the edge of the wing's unwrap; it's not so clean if it's part of the hull, because when we take the wing out of the hull it leaves a rectangular hole; but if we have bent edges around the hole that need to stretch out, they'll actually stretche "in" --into the hole in the hull, which means that their lengths will have to decrease for their widths to increase. Often LSCM goes nuts when you try to make it do that. So, try to only catch polys that will map fairly straight, when you're seaming the boundaries of a big piece by cutting smaller parts out of it. Leave the troublesome parts for later. Procrastinate.

Here's my cut at the front end of the wing:

And NOW I realise I have too many bloody polygons around this area... We'll place the seams first; I'll come back to this later...

Here's my cut under the wing:

And here's my cut at the back of the wing:

Now, verify in transparent mode that the loop is closed, then Ctrl-E -> Mark Seam:

Same thing for the tower:

Now we cut out the front-facing part of the hull:

We have to, as a cylindrical unwrap of the main hull would miss a front-facing plane entirely. But you might ask, didn't I say I didn't want to catch 45 degree strips into the main hull? Well, in the case of the wings and the towers that is so, because when we cut them out, we leave holes in the hull unwrap. But in the case of the front plane we don't have to worry about strips flaring inwards into a hole, because the edges adjacent to the front plane will all be on the left side of the texture, lining up vertically. And I'd rather have those 45 degree strips with the hull, because I will probably put impacts and scraches there that should smear or fudge or fade backwards along the sides of the hull; and if I put those bevels elsewhere in the texture it would be pretty hard to match a scratch begining in one island and continuing in another.

Additionally, if you look at the paint, you'll see that my seam goes along the edges of the paint. This will allow me to better separate and group islands by material.

And what are those extra seams making V-signs at the corners?

Those are there so that when the bevel strips flare out they don't have to stretch like contortionists. They are "relief cuts".

Let's do the bottom tower:

Note that the towers and the front, not to speak of the wings, will need further seams. We'll add more seams when we get to them; right now we're not working on the towers or wings or anything other than the main hull. One thing at a time... Now the back:

Done with the main hull. What?! Yes; we don't need to put seams for the boxes at the top and bottom because they are already split. We don't need seams for the inside surfaces of the intakes, because they are also split already. Smooth groups, remember? Ahh, but we do need a couple more seams; hold on... Here:

The reason we need those there is that the surface going in there is continuous with the hull. Only the upside-down surfaces are split. So, those parts inside the "vents" need to be put elsewhere on the texture; otherwise they'd overlap with the outside of the vent, in the cylindrical unwrap.

Now we're really done with the hull. What's next? We said we'd start with the biggest parts first. I'm not sure what's the next-biggest part... Probably the wings.

Different people have different ways of doing things. Most Blenderheads swear by LSCM. Me, I don't like LSCM, --at all. Why? Because I like my things aligned, sort of. Specially the front-to-back direction on the ship, I want it to align horizontally in the texture. LSCM is an "obsessed algorithm", IMO; that only cares for surface equalization, and for which everything else is secondary. For me, front-back alignment is a primary concern; and surface area equalization is kind of the last item in my long list of priorities. The way I work is I use Unwrap -> From View. Naturally, therefore, that's how we'll do this tutorial. If I tried to teach the way someone else works, I probably wouldn't do it right, anyways.

There are 6 views that interest us: Front, back, top, bottom, left and right. As such, prepairing the seams is easy. Here I've seamed up the wing:

Ehmm... Yes, if it looks like I've made a major clean-up of the mesh, it's because I have; and I'm not quite done with it either. Expect to see more polygon reduction crackdowns as we proceed... And, by the way, what I did to display titanium parts only, was to select them (by material), then menu Select -> Inverse, so that everything that's NOT made of titanium is selected, and then H to hide. And as I did that I discovered many little polygons floating in the middle of nowhere... Facets that I meant to assign a different material to them, bot missed them. Cleaning up the mesh is a job that's never done... I'll fix that next, won't re-upload the pic above; better for the garbage to show there, for the record. So, I realize now that this is something that should be done in a systematic way. Select one material at a time, inverse selection, hide; and then visually verify that everything that's there should be there. Otherwise select the mis-materialized pieces, assign the correct materia to them, and hide them. Then Reveal everything, and do the same in the next material in the list. Alright, let me do just that...

Okay, done cleaning up the titanium family. I even found floating edges without faces "made of titanium". Meshes are like white socks; --never perfectly clean. I'll do the other materials later.

Anyways, back to the seams in the wing: There's always the question of which view gets the bevel strips. I gave the top-bottom views the top priority; side-view second; and front/back views last. Why? Front and back views should always have the last priority, for the reason I stated earlier: Part of the texturing will be to represent impacts and scrathes. Some of those impacts should begin at a partially front-facing bevel and then fade towards the back. To represent the fading of such impacts, we can use linear blurs in Gimp, but Gimp knows nothing about the UV-mapping; --i.e.: it couldn't begin a blur in one island and continue it on another. So what we want is for such bevels to be already on the left edge, and part of, an island in the texture. Then an impact is basically a dot within the bevel strip, in the texture, and we can just motion-blur or smudge the dot towards the right, and the smudge will proceed towards the right in the texture; --over polys that follow behind, in the mesh. As for whether the top/bottom views should get priority vis a vis the side view, or viceversa, over which gets the longitudinal bevel strips, this is a matter of taste. I usually give that priority to whichever are the larger pieces, but it's just a matter of personal preference, and more for the sake of having a rule I follow, than any technical consideration.

Still proceeding from largest to smallest pieces, next in the list would be the dockings. Here we have a bit of descision to make: What do we do with the inclined surfaces facing towards the engines? If we unwrap them from top or bottom view, the rectangles for those areas in the UV-map will be compressed. If we unwrap them from side-views, the'll be compressed also. If we LSCM them, they will get deformed, like parallelograms probably; and we'll have to fix them by hand. Or else we could turn the view until we see them flat, and unwrap them from that view... I'm going to unwrap them from top and bottom views; because, in this case at least, I don't care about a texture stretch at a place that's so hidden from view and probably featureless anyhow. So, same general method as with the wing:

And as with titanium, selecting the green paint parts uncovered things that were green that shouldn't be, and another clean-up took place. I've been fixing so many things, from blotched normals to unneeded vertices to wrong material assignments, that I couldn't hel make a new render to see how those fixes so far affect the final result...

Looking nicer and cleaner, as expected.

Proceeding from largest to smallest parts, I think what comes next is the front and back facing sections of the hull.

The front is just a matter of separating the battering ram. I ate my words and jumped to the occasion to seam up the ram itself. No seams are needed around the opening because that edge is already split.

The back actually doesn't need almost anything done to it, because it's already all split up by smooth-grouping, including the Apollo/Soyuz dock. All that needs to be done to it is a few cuts to be able to unroll the cylindrical items...

...

(Last step: Align vertices near x=0 to x=0, apply X-axis mirroring, and place the central seam.)

Optimizing ship orientation for UV unwrapping

(Z is "up", Y is "forward"? Got to experiment first...).

Worry not, though; we'll put it back to "Z is forward" after the unwrap. It's just that Blender shows the top, bottom, and side views vertically, when your forward axis is z, which is too bad if you're unwrapping "from view"... because we want the forward axis horizontal and pointing to the left in the texture...

"Starting Big" in Potato Mode

Do the biggest parts first.

(2 be continued; this is a WIP --Work in Progress)