Okay, so I got a little frustrated with the available UV mapping programs. I've been using UVLayout a bit - more specifically the free demo version - which is actually quite nice. I think the guys at Headus have pretty much nailed the workflow. But there are several little things that don't quite work the way I'd like them, too, and some things it simply refuses to let me do. We can talk about details later.

It so happens that I'm not completely clueless in the kind of algorithms that one would use to optimize UV maps. So I've really been tempted lately to go ahead and try my hand at UV mapping software myself. For starters, I was thinking of some simple tools that would let me modify and optimize regions of an existing map. But if there was a big enough market, I would definitely consider turning those into sellable products and maybe given time even extending them into a full-fledged UV mapping/layout suite.

So I thought I'd just ask around what features people are missing in the UV software they are currently using and specifically which ones they think could be handled by a UV "postprocessing" tool.

Cheers,
O.

the nicest thing I would like to see is a tool that matches seam edges in the texture (kind of blur or clone thing but that is aware of the UV seams run. I have no issue with creating seams, then making a decent layout of the UVs (in fact Blenders UV layout is remarkably good if you have used good seams), it's just that seam matching bit on the 2d texture which is the pain.

Hope that explains what I mean!

Ooooppp a follow up thought as blender is open source, you could play with your algorythms in your own local build or publish for others to play with if you so decide (depends on whether your aspirations are commercial tool or not.)

UVLayout have great way for matching the seams.

Tihomir

Flat maps, with less distortion.  All of the UV apps I have used tend to localize the pinched areas (red) into smaller very distorted spots, like the tip of the nose etc. If it could somehow be made to spread the distortion gradually over a larger area and still keep things fairly square. I always end up using smudge tools and spending hours and days to spread it by hand all the while constantly battling my UV layout app.  Also a tool to manipulate the UV's directly on the model itself, bodypaint has one but a rather crappy one.

Those two items for now, but I have many more wishes :lol:

cheers,

Mike.

Many thanks for the input so far!**

Gog**: As far as I'm concerned, UV mapping and texturing are two separate cans of worms. There are a number of 3d painting programs such as ZBrush, 3D-Coat etc that supposedly let one paint over seams. Some of them have clone tools. So I would imagine one could do the bulk of the texturing in Photoshop or Gimp etc, then fix the seams in one of those 3d programs. I haven't really done any serious texturing yet, so I don't know how practical that idea is.

Don't get me wrong, having a UV-aware photoshop plugin or somesuch that could deal with seams directly in UV space is still a very neat idea. Not exactly what I had in mind originally, but maybe something I could keep in the back of my head.

Regarding the Blender thing: I was considering doing a plugin or extension for Blender or Wings3D first. But that would require me to immerse myself in the respective API and inner workings and, in the case of Wings3D, a new programming language. If I do a standalone program, I don't have the infra-structure that comes with an existing sophisticated modeling package, but on the other hand I can dive straight into coding without having to learn any new tricks. It's a bit of a tradeoff, and I admit I am not completely decided on which way to go. But I guess I'll try a simple standalone first and see how it goes.

The other aspect of course is that if I'm intending to turn this into profit at some point, a standalone might be easier to handle.

Quote - UVLayout have great way for matching the seams.

Tihomir

Not sure if I understand you correctly, but I think you are probably referring to the match seams function via the 'm' key. In UVLayout, one can mark edges for welding, then hit 'm' to align those with the matching edges on another island/chart, and finally hit 'Enter' to do the actual welding. Is that what you meant?

Gog was talking about matching textures across seams. Completely different thing. I don't think one can make textures in UVLayout, but maybe I'm wrong.

Oh, and just to avoid misunderstandings: I'm absolutely not trying to badmouth UVLayout or Headus in any way. I think they have an amazing product, and if it turns out that I can in fact do some useful things that are not yet possible with UVLayout, I'd be much happier to eventually work with them then against them.

Quote - Flat maps, with less distortion.  All of the UV apps I have used tend to localize the pinched areas (red) into smaller very distorted spots, like the tip of the nose etc. If it could somehow be made to spread the distortion gradually over a larger area and still keep things fairly square. I always end up using smudge tools and spending hours and days to spread it by hand all the while constantly battling my UV layout app.  Also a tool to manipulate the UV's directly on the model itself, bodypaint has one but a rather crappy one.

Those two items for now, but I have many more wishes :lol:

cheers,

Mike.

I like those suggestions.

Regarding the first one: all the applications I've seen so far basically have one - or sometimes two - general unwrapping method that does its thing with not many ways for the user to influence the overall behavior. Usually the options seem to be either take it as it comes or otherwise do it yourself. I think it should be possible to change the cost function in such a way that unavoidable distortion would be spread out rather than localized.

Just to mention one other possibility: I'd find it very neat if one could restrict the UVs in certain areas in such a way that, say, constant y in model space would always correspond to constant v in UV space. That would be great for making things like socks and stockings easy to texture.

The second suggestion probably speaks for itself. I imagine the really tricky part would be to come up with a good user interface for that kind of thing. Could you tell me a bit more about how the tool in bodypaint works, and what's wrong with it from your point of view?

odf: The UV tools in Body paint allow you to have your model in one frame and your UV's laid out in another side by side, you can move your UV's around in the UV side and wach what they do on the model in the other. But you can't actually move them on the model itself. (You can do much the same thing in Max) What makes The body paint tools sort of crappy in this case is they display very strange on the model and sometimes you can't tell the back of the model from the front The model sometimes becomes transparent or inverted. The other thing is the move tool itself does not fallof very smoothly no matter what setting. I would like it to fallof smoothly like the smudge tool in Maya. And I would like to be able to actually move the UV's on the model itself not in another window.

Zbrush has a way to roll your UV's horizontally or vertically in order to help line up the seams, but still will not align the seams like youare talking about for using shaders for stockings and clothing. I have always had to do that by hand with smudge tools and guesswork, bodypaint, and UV mapper to move the individual edge vertices to align perfectly once I get everything worked out somewhere close in the other apps. So that would be a fantastic addition for a UV app. I have always thought this might be very good for making textures in photoshop because you could actually just overlap on one side of a seam and then cut the excess off and move it to the other side of the seam and it would line up perfectly vertice to vertice up and down.

Quote - odf: The UV tools in Body paint allow you to have your model in one frame and your UV's laid out in another side by side, you can move your UV's around in the UV side and wach what they do on the model in the other. But you can't actually move them on the model itself. (You can do much the same thing in Max) What makes The body paint tools sort of crappy in this case is they display very strange on the model and sometimes you can't tell the back of the model from the front The model sometimes becomes transparent or inverted. The other thing is the move tool itself does not fallof very smoothly no matter what setting. I would like it to fallof smoothly like the smudge tool in Maya. And I would like to be able to actually move the UV's on the model itself not in another window.

I see. You can actually do the side-by-side thing quite nicely in Wings3D. It displays okay, and Wings has a "magnet" option, so you can move whole regions around. I don't use it much, though, so I don't know how smooth the falloff is. I'd actually be not too unhappy with the combination of UVLayout and Wings if the former wasn't so erratic about importing existing UV maps.

Quote -
Zbrush has a way to roll your UV's horizontally or vertically in order to help line up the seams, but still will not align the seams like youare talking about for using shaders for stockings and clothing. I have always had to do that by hand with smudge tools and guesswork, bodypaint, and UV mapper to move the individual edge vertices to align perfectly once I get everything worked out somewhere close in the other apps. So that would be a fantastic addition for a UV app. I have always thought this might be very good for making textures in photoshop because you could actually just overlap on one side of a seam and then cut the excess off and move it to the other side of the seam and it would line up perfectly vertice to vertice up and down.

Just to clarify: how would a cylindrical mapping do in those cases? Let's say you make a shirt. If you mapped the arms and the body from below the armpits down cylindrically in y and x, respectively, and let the program find a nice smooth transition for the shoulder area, would that solve your problem?

You're talking with a center seam down the back? If you could write up something that would give a good seam for the shoulder area that would be impressive by itself. I think the worst areas on mapping human shaped figures are the arm pits and the crotch. I also spend alot of time lining up the leg seams. I line them up so that the edge vertices on the leg seams line up perfectly from left to right  on a grid pattern so that you could draw a straight line horizontally from one vertice on one side of the seam to it's mate on the other side of the seam. I am thinking that this will allow much easier matching of the texture along the seam, since each vertice and its mate is straight across from eachother. And the distance vertically along the seam from edge vertice to edge vertice is exactly the same on each seam edge. I do this same thing for the arms only horizontally instead of vertically. If there was a way to do this automatically in the unwrapping process it would save enormous amounts of effort and time.
I think you are doing much the same thing with Antonia?

Center seams always match since I just chop the model in half along the center and mirror and weld , and flip that half in UV mapper so my UV's always match on each side.

This next idea might be crazy, but do you think there might be a way to create a UV tool that would integrate with Poser that could automatically adjust the UV's for heavily morphed figures? Something so a user could morph their character with dials or self made morphs and then in Poser click on the tool and say adjust the uv's to compensate for the stretch of the morphs? It would keep the same seam edges but just sort of smooth the UV's. Just a thought. :)

Quote - I think you are doing much the same thing with Antonia?

Not really. I'm much too lazy to do so much work by hand. Once I had rotated the charts for the legs the right way, UVLayout did a pretty decent job at aligning the seams just using the standard method. Then I stitched them back to the body and relaxed the transition area. Not perfect, but good enough for lazy me. Legs are hard to model, but not that hard to map.

I believe the pro version of UVLayout has a way to constrain vertices along a seam to be at the same u or v coordinates as their matching partners.

Quote -
This next idea might be crazy, but do you think there might be a way to create a UV tool that would integrate with Poser that could automatically adjust the UV's for heavily morphed figures? Something so a user could morph their character with dials or self made morphs and then in Poser click on the tool and say adjust the uv's to compensate for the stretch of the morphs? It would keep the same seam edges but just sort of smooth the UV's. Just a thought. :)

Yep, that would be nice. But it sounds very hard, even for my inflated ego.

I didn't know UVlayout could do that, I looked at it but the price is pretty high for me right now.

I did some experimenting along the lines of compensating for the stretch in the shoulders and buttocks in the bends. And I came up with something that worked pretty good for much of it. It involved a two prong attack on the problem. The first part was to bend the body parts and export the model posed and then to relax the UV's again to take out some of the red in the stretched areas, not completely but a good balance between the stretched areas and the non stretched areas so as not to ruin the areas that don't get stretched. And then the second part of the fix is to spread the poly stretch out over a larger area with the JCM's, but if the UV part could be done in Poser with a UV plug it would be much better. That would probably be a pretty ambitious project to create a UV plug like that, but I do think it would be popular.

Maybe another thing to roll around in the back of my head for a few weeks...

I'm not completely sure about that UVLayout feature, but I think I saw it described in the manual. If you download the demo, that gives you access to the pro features for testing, too, although then you have to sacrifice the ability to save your mesh.

Anyway, I think it should be possibly to do more than that. There's always a trade-off, of course. Nicer seams for texturing produce more distortion. So in the end it's up to the user to decide. Ideally, they would be able to assign weights to various features of the mapping, and even change those weights between different areas.

Speed is of course an issue, too. The more cool features one builds into the relaxation procedure, the slower it will run. That might be another argument for doing it as a post-processing step: let the regular UV software do the basic mapping while working quickly enough for interactive use, then optimize it with a separate tool which might take a bit longer to run.

M just snap the UVs in UVLayout :)

I was refering to anchors A key, S key for pining. Those two give excellent control for matching the seams. I will not go into explanations here, there is loads of great tuts for beginners that Phil has made on site :)

UVLayout have everything you need and then some more, thing is that his workflow is through hot keys. So many beginners don`t even scratch the surface of the app because many commands is not straigh forward :)

Red areas (streched) or blue areas (commpresed) can be tweaked with B, R or 6 key.

Just as in BP you can work with model and UVs, if UVs are messy and tricky to cut the edge for example. Its possible to select the edge on 3D model in 3D view (3 key).

Tihomir

You are absolutely right, tantarus. I haven't really understood all the features of UVLayout yet. Many of the interesting things are only available in the pro version, and not explained at all or only very cryptically in the user manual.

Being able to tweak things that the relaxation algorithm messes up is not good enough for me, though. I want a better algorithm to begin with, so I have to tweak less. Or, you know, at least have a post-processing tool that does most of the tweaking for me.

F is basic flattening, press SHIFT+F and UV chart will be streched. Leave it for few sec. or more depends on mesh complexity, than press SPACE to release it and leave it to flatten the mesh.

You can see allot of twirling around, but it will flatten in the end. If mesh is really dense and complex I tend to chop it into smaller parts and map each part separately, then weld and tweak the transitions. Finally match the seams and straighten the edges so its more PS texturing friendly :)

Pro version is a bargain, it worth every penny :)

Tihomir

Looks good, I would make it different. But thats personal preference nothing more :)

Tihomir

As I said earlier, I have no experience in texturing, so I tried to mimimize distortion while placing seams where they would be the least obvious. Next time, I might do it a bit differently based on better knowledge of the workflow.

Anyway, this may illustrate some points. The default flattening process in UVLayout produces localized distortion on the nipples and areolas - a bit reduced here due to the extra seams I added, but still pretty obvious - and overlaps where the fingers and toes meet. I agree that UVLayout has some great tools to tweak such things, but tweaking by hand is always a bit messy, tedious and error-prone. I think it should be possible to automate some of that kind of tweaking by using an alternative flattening algorithm. Specifically:

1) For the breasts, change the cost function so that distortion will be spread over a larger area rather than being localized.

2) For the hands and feet, add collision detection to the flattening procedure, so that the algorithm finds the best overlap-free solution.

That's the kind of thing I'm planning to implement. So I'm wondering whether experienced users of existing software - particularly UVLayout - would rather keep tweaking by hand or consider using my hypothetical standalone tools for post-processing.

An additional problem to consider is that both UVLayout and Blender have some trouble importing .obj files with existing UV coordinates. That basically means that as soon as one starts tweaking UVs in a separate program, one may lose the ability of doing any further editing in ones primary application. There might be ways to fix those issues, though, if I look into them a bit more.

A program that could recognize repeating topography and allow me to stack templates automatically, so I could do one texture paint and have all the repeating stuff get covered by it. Of course there are times when that should be optional. But yeah, the first idea would be a nice quick hack to have sometimes.

Ability to annotate the template layout with text during and after mapping. That way you can label around things so you know what those mystery bits are. Or if something is top or bottom or whatever. Not always easy to know if you're painting over the guides in a 2D app.

Also something that might be a little better at discerning how much anti-aliasing occurs along seams. I know if you crop too tight with overpaint on a template, the anti-aliasing will bleed the background color of the map onto the texture where you don't want it. The problem with this is that behavior may differ amongst rendering software.

Also I'd like to have something that is free and easy to use (possibly even open source), because I'm a cheapskate. (And the economy/job situation isn't helping that any.) Something where I could manipulate a model in a predictable way in 3D space after mapping, and then get my tablet out and paint on it. Apparently free solutions as of now aren't all that easy to use or are broken in some other way (like requiring tris or no n-gons or a certain image file format or whatever). But I guess I can't have everything.

Quote - A program that could recognize repeating topography and allow me to stack templates automatically, so I could do one texture paint and have all the repeating stuff get covered by it. Of course there are times when that should be optional. But yeah, the first idea would be a nice quick hack to have sometimes.

I don't understand what you mean by this. Could you give an example?

Quote - Ability to annotate the template layout with text during and after mapping. That way you can label around things so you know what those mystery bits are. Or if something is top or bottom or whatever. Not always easy to know if you're painting over the guides in a 2D app.

Ah, glad you're mentioning template generation. I'm not quite sure what existing apps can do in that respect. I ended up making my templates in Wings3D. Annotation would definitely be good. Maybe automatic generation of seam guides?

Quote - Also something that might be a little better at discerning how much anti-aliasing occurs along seams. I know if you crop too tight with overpaint on a template, the anti-aliasing will bleed the background color of the map onto the texture where you don't want it. The problem with this is that behavior may differ amongst rendering software.

I'm not sure I understand what the problem is. What would keep you from generously overpainting everywhere in order to avoid that antialiasing issue?

Quote - Also I'd like to have something that is free and easy to use (possibly even open source), because I'm a cheapskate. (And the economy/job situation isn't helping that any.) Something where I could manipulate a model in a predictable way in 3D space after mapping, and then get my tablet out and paint on it. Apparently free solutions as of now aren't all that easy to use or are broken in some other way (like requiring tris or no n-gons or a certain image file format or whatever). But I guess I can't have everything.

If I make those hypothetical tools of mine available, I'll probably start with some simple open-source ones. If things take off and I start charging money, I might offer a free license for hobbyists and a separate one for commercial use. But really, my secret evil plan is to get people talking, then be hired by Headus. :biggrin:

Definitely not planning on making a texturing program, though. 3D-Coat is supposedly very good and only 150$ for a professional license.

odf:

Just the two points you mentioned would be a great improvement, for myself those two are really the most pressing. The tip of the nose has always been my biggest problem, but then I have not UV mapped a female model yet and I expect the breasts will definitely be trouble.

Collision detection sound like very intriguing idea, if you can pull that off that would be awesome. Flattening algos. depends from mesh to mesh how they will behave, there is always room for improving. UVL is fantastic, but its good to be ambitious and try to get it better.

Paul are you talking about copying UVs, UVL have that too and you can choose to stack them in one place or distribute them on U or V :)

Tihomir

I've done the stacking before in Wings...  So say you have a repeating topography feature, lets say you have a cube. And for whatever you're doing, you want the same design on all six faces. Instead of manually taking each face and stacking them on top of each other, have a software method for doing that. Therefore when you paint the one square, all six faces are done.
(I know a cube is a quick and overly simple example, but there are more complex things where if there is repeating geometry I want to be able to stack the templates automatically. Of course this may be trickier on things like circular shapes, the software has to know which way is "up" on the mesh and things like that.)

As for the annotation, I do that sometimes, but I have to rely on my memory and a 2D app to apply the labels. It would be nice if text was available in the UV-ing phase, so I'd know exactly what some bits are without having to worry about my own mix ups.

As for the overpaint thing, I guess it's an un-necessary neatness thing. But sometimes it's easier to "read" a texture map if the shapes aren't overly blobbed over on the sides. It's probably just a matter of some software that knows where the templates lie and then says to overpaint by n-number of pixels outside of them. (I usually find that 3-6 px works ok, and then you don't get the seam-AA bleed-through.) But it would be neat to have a "paint on the model in 3D" type app that outputs 2D textures and uses this idea for neat output.

Quote - I've done the stacking before in Wings...  So say you have a repeating topography feature, lets say you have a cube. And for whatever you're doing, you want the same design on all six faces. Instead of manually taking each face and stacking them on top of each other, have a software method for doing that. Therefore when you paint the one square, all six faces are done.
(I know a cube is a quick and overly simple example, but there are more complex things where if there is repeating geometry I want to be able to stack the templates automatically. Of course this may be trickier on things like circular shapes, the software has to know which way is "up" on the mesh and things like that.)

Ah, I understand. Automatically recognizing things you might want to stack sounds tricky. I've been working on my mind-reading software for years, but I'm afraid it's not quite ready for production use yet. :biggrin: But if you selected a bunch of charts and told the program to stack those in the nicest possible way, that should be doable. Of course, first results would very much depend on the degree of similarity between those charts.

I've seen that UVLayout can do stacking, but I haven't tried it yet. My priority for features to implement would obviously be things that UVL can't do and/or that are easy.

Keep the ideas coming. Even if I won't be able to do everything that people suggest - at least not in the short term - I find it an interesting topic. I've started to think about how to do a flattening algorithm in slightly more concrete terms now. I also have some code in place to read and write .obj files, do certain manipulations on the meshes and - as of yesterday - display meshes in 3d with OpenGL. As soon as I add a 2d display for the UV coordinates to that, I'm pretty much set up for launching my first wild map optimizing experiments.

Umm, the binary won't be small, though. A good 10MB at the least, plus one would need to have Java 5 or later installed. Does anyone find either of these prohibitive at this day and age? (On the plus side, it will be a "one size fits all" - one single binary that runs on Windows, MacOS and LInux.)

Paul, yes UVL can copy the most complex mesh UVs as long as topology is the same. All you need to do is select one edge (not loop) from master and then same edge from all other, press SPACE and bang all is stacked 100% perfect. Neat thing is that only original must be mapped, others will spread according to him once copied :)

Looking forward to see what you will make odf :)

Tihomir

Quote - Paul, yes UVL can copy the most complex mesh UVs as long as topology is the same. All you need to do is select one edge (not loop) from master and then same edge from all other, press SPACE and bang all is stacked 100% perfect. Neat thing is that only original must be mapped, others will spread according to him once copied :)

The topology must be the same and the user even specifies the corresponding edges? That's easy, then. :biggrin: And it is in fact quite useful, I agree.

Maybe try to find solutions for hard surface models. They are for sure more tedious than organic :)

Tihomir

Quote - Maybe try to find solutions for hard surface models. They are for sure more tedious than organic :)

Good idea! I'll keep that in mind.

Some quick ideas: auto-seam along hard edges, recognize certain types of surface (for example tubes) that can be mapped automatically, make automatic layout suggestions with similar parts grouped together with an option to quickly mark them for stacking. But I'm just brain-storming here, of course. Getting something good working would probably take a lot of effort, but might be worth it, especially if it hasn't been done before.

Now you talking, if you can pull of all that that would be awesome. For organics you have UVL, Unfold 3D, and few more that do their job just right. But for hard surface you still have to spend time because its not hard but tedious as hell. I use BP for hard surface, his Interactive mapping is big help but still not perfect.

Tihomir

This is a very interesting discussion. I wonder how much of the problems we face are due to the need to make the maps friendly for other people to make textures with? For example, the classic Poser UV map has the body pretty much all on one map and the parts attached to minimize seams, with the idea it makes it easier to paint on. However, this often requires detailed maps to be way to large for the average use of the app (still images for screen with characters in mid range of camera).  A better method - one that would lend itself to more detail using smaller maps would be to break the body into more parts, allow each part to take up a much bigger space on the map than they could if attached to each other. This approach is how game characters are done and why they look rather good on screen despite using maps no larger than 2048 (512 is the common, with 1024 and 2048 saved for hero characters).

Not to derail the thread, which is what made me think of this in the first place but what do you think? With more and more 3d apps providing tools to texture across seams inside them , how important is that classic UV setup these days?

Teyon:

Good point. Personally I like the Zbrush  method using adaptive UV tiles. Each poly is mapped acording to it's surface area and then packed into very tight packages, very little wasted space.  But you could never work on them in a 2D app like photoshop. But then as you say many apps now have the ability to directly apply textures to the model. And some of them like 3D Coat are relatively inexpensive. I really think that this is the future even for Poser because soon it will be common in most aplications and it is much easier to do.

From a standpoint of excellence it is much better as well because these methods of UV mapping have very little or no distortion at all. And even the new Photoshop (CS4) now is able to apply textures to the 3D model directly.

Many people still hold on to the old antiquated methods of UV mapping I believe because its familiar, and also because programs which were able to paint in 3D in the past were very expensive and out of reach for most hobby artists. This is changing.

Hi again,

Another thought. The perceptions of the user base. The vast majority of users have no concept of any of the processes used to create the figures for Poser or D/S. Most users believe that larger textures are better, many also believe that the more poly's the more detailed the model. So 5 or 6 very large texture maps are a big selling point for many in the user base, and merchants are going to support what sells. So how to go about changing the user base perceptions?

Also, what does the use of these newer types of UV mapping do to shaders? I don't know much about shaders, but it seems to me that trying to use shaders to create things like stockings or nylons might not work with Adaptive UV Tiles.

Quote -
Many people still hold on to the old antiquated methods of UV mapping I believe because its familiar, and also because programs which were able to paint in 3D in the past were very expensive and out of reach for most hobby artists. This is changing.

Personally I don't think painting in 2D is antiquated. If you have the chance to unwrap a surface onto a plane and work on that, you just do it. It's so much easier than constantly having to adjust your viewing angle. Besides, 2D drawing programs have been around for a long time and are fairly mainstream. I find it rather unlikely that 3D applications will be able to match their workflows and tool sets any time soon. That said, if programs like Photoshop start integrating 3D functionality, that might indeed change things.

While I think I could get used to hand-drawing textures directly in model-space, I'm not so sure about working from photos (which still seems to be what everyone does in the Poser world). I've seen video demos for both ZBrush and 3D Coat, so I know it can be done. But I have to confess when I gave it a try in 3D Coat, I became frustrated very quickly and went back to Gimp. I found the process of aligning the model with the reference image a major p.i.t.a. In a 2D application, I can open a bunch of references in separate windows and simply cut and paste or clone.

For me, the major advantage of having affordable 3D painting software is that seams should become much less of a problem. So basically when doing a UV layout, one could go for logical patches that fold out flatly and are easy to texture in a classical 2D programs rather than bothering with huge pelt-like charts just to avoid a few seams. Visible seams resulting from the 2D texturing phase could then be dealt with fairly easily by cloning and suchlike in the 3D application.

But I have to say I don't have a lot of practical experience in texturing yet, so I might be wrong.

Quote - Also, what does the use of these newer types of UV mapping do to shaders? I don't know much about shaders, but it seems to me that trying to use shaders to create things like stockings or nylons might not work with Adaptive UV Tiles.

Procedural shaders don't have to rely on UV coordinates directly. One could use image maps to feed them the necessary positional information. For stockings, one could encode the height along a person's leg as a grayscale value - unless of course the legs were modeled to go straight down and the shader had access to the original point coordinates in object space, in which case things would be even easier.

A trickier case would be the creation of clothing textures on a nude model from image patterns rather than purely procedurally. I've been toying with the idea of using alternate UV maps for the affected regions of the body for such applications and then basically rendering the texture created with the alternate map onto the original map. I imagine with enough resolution and appropriate filtering, that should give pretty decent results. But I haven't tried it in practice yet.

"And even the new Photoshop (CS4) now is able to apply textures to the 3D model directly."

Really?  How?

About seams, yes, anything that makes seams better would be such a boon (can't afford UVLayout).  Tried painting directly in Bodypaint, it was okay but I found the resolution a bit naff.  But I'm looking more into that this Easter.

I prefer painting my own in PhotoShop.

Stacking and annotation are both very helpful ideas, I agree with those.

I sometimes incorporate photographs into one of the layers when I'm painting a map - I think BodyPaint won't let you do that???  But maybe I just haven't gone into that enough yet.

(I use Wings3D UV mapping)

Quote - "And even the new Photoshop (CS4) now is able to apply textures to the 3D model directly."

Really?  How?

Actually, I'm pretty sure that's in CS4's more muscley cousin, CS4 Extended, which is about half again as expensive.

Yes, only CS3 and 4 Extended support 3D formats, and allow painting on model. In PS that is just basic tool, BodyPaint is specialized application for UV mapping and texture painting. You can paint on several channels at once, for example Color, Reflection, Displacement, projection painting allow you to paint seamlessly. What you see in working window is just preview of texture, render to see final result. Anyway preview resolution can be increased up to 4K but that slow down the work.

Has anyone used both BodyPaint and 3D Coat for painting textures? I'd be interested in how they compare?

By the way, I am currently working on a mesh and uv mapping viewer and manipulator interface. Normally, I spend most of my time on developing features and have sloppily implemented interfaces that just do enough to allow one to steer the application. This time I thought I'd be more thorough with the interface structure. We'll see how it goes.

I don't know how you can work with hotkeys, I find them an absolute nightmare.

Menus, menus, menus, for me everytime.  My mind just can't cope with having to remember all those letters and numbers - and sometimes my joints just won't let me stretch to all the keys 'cos I need to use an ergonomic keyboard to reduce RSI, so stretching across the gap can be quite a strain.

I'm just falling apart here!

Lol.

P.S. Oh by the way, if anyone can explain about this PhotoShop method of painting directly on a model, I'd be very grateful - perhaps in another thread? (I do have "Extended")

Quote - I don't know how you can work with hotkeys, I find them an absolute nightmare.

Menus, menus, menus, for me everytime.  My mind just can't cope with having to remember all those letters and numbers - and sometimes my joints just won't let me stretch to all the keys 'cos I need to use an ergonomic keyboard to reduce RSI, so stretching across the gap can be quite a strain.

I'm not completely disagreeing. Programs that can only be operated via hotkeys are a nightmare. I like to have well-structured menus so I can find everything easily. But hotkeys can save a lot of time for things I do often. My favourite programs are the ones which let me define my own keyboard bindings. Maybe I should look into that.

Anyway, I'll have to do some more work on the basics of selecting and manipulating (dragging) stuff and lay some groundwork for the interface before I can introduce any interesting functionality. I just thought I'd let you know that there is some, albeit slow, progress.

As a side note, I'm using this project to familiarize myself with a new programming language, Scala.

La - Scala?  Do you have to sing?

ROFL.

Sorry, couldn't resist.

About keyboard shortcuts.... okay I'll admit to using 5.

Two are the old CrtlC and CtrlV for copy and paste.

One is A to zero the view in Wings and one is the space bar to deselect everything in Wings.

Oh-kay, one more:- CtrlAlt Z for undo.

I ain't admittin' to anythin' else.

But all of those are reachable with little finger and thumb or little finger and index finger or just one finger - that way I can keep one hand on the mouse to keep control of the view etc.  If I have to keep removing my hand from the mouse I find that my speed at modeling is drastically reduced, and you get distracted.  (and I've never been very fast at the best of times.)

I do feel inclined to sing arias in praise of the Scala programming language, but I decided to spare myself the humiliation. Instead, allow me to refer you to www.scala-lang.org

As for hotkeys, I frequently use X, Y and Z in Wings, as well as the obvious V, E, F and B. I also use L for edge loops and G for edge groups, and quite a few self-defined ones. I feel that having to use menus on a regular basis would slow me down quite a bit, so it's a good thing that there are convenient hotkeys and that I can define my own for things I do a lot. For example, I use T to get in and out of tweak mode and the slashes ("/" and "") for rotating edges. But of course it's a good thing that the menus are there and I can easily find all those functions that I haven't yet bothered to memorizer the key bindings for.

On the programing language, I'll take your word for it.

I'm sure I can hear singing... faintly in the distance.

As for the matter of hotkeys, I can see the value of being able to assign your own hotkeys, lately even I have found myself doing the same thing over and over - unfortunately it's usually deleting unused textures in Wings3D, laboriously... one by one.... over and over again... there's no hotkey available for that.

I really wish there was.

Anyway, so I'd vote for any software that allows you to assign hotkeys if you want to, and allows you to use menus if you don't.

I like it when default hotkeys are somehow hinted in menu commands. I consider that to be part of good UI design. So when you pull up a menu item for selecting edges you see "select Edge". But even then, you should still be able to re-assign other than default if it can help your workflow.

As for Fran's deleting of unused textures in Wings, did you post something pertaining to that in the Wings3D wants thread? If you haven't done so already, you should. Sounds like having commands to delete unused textures or mats might not be a bad idea.

I have start to use 3D Coat, its really impressive software. Anyway description that pop up when you put cursor over any tool or command is making learning curve minimal. It actually say what tool or command do not just name of it which is useless in other apps.

Tihomir

Paul,

I had meant to, but I couldn't find a message by me in that feature request thread (it took a fair bit of searching to find the "Features Request").

So I have now.

I dunno why I didn't do it before, deadlines probably.