UV maps on Poser 7 Sydney G2



  • I think the biggest problem with the maps is the fact that the whole body is one large map. Maybe if it had been broken up the UV's would have been lined up better and smaller islands are easier to work with.



  • @anomalaus These last two images are showing the grouping for rigging. Her body is one single map. There are only individual islands for her finger and toe nails, which are separate geometry parented to the body mesh. There are no separate maps for the arms, legs, hands... etc. Other figures have things broken up. It's easier to create a body suit with the G2 characters because of the single map, but if the UV is messed up, then you get the issues I have. Like you stated, the back is out of alignment. It might not look like it's that much, but when I have all the striping on that costume, it's all out of whack.

    Even around her bicep area, the map twists around her arm and the ends don't come together (you can see that in the second pic I posted quite clearly).

    I am going to use her original geometry .obj file in the runtime and see if I can do anything with the mapping. It might just be more trouble than it's worth, but I won't know until I mess around with it.



  • @seachnasaigh that's a python script currently under development to produce UV templates and automatically generate seam guides directly within Poser. There are two incarnations of it at the moment. The first, and originally intended end product, simply creates UV template images with seam guides and optional facet numbering. The second creates actual planar geometry objects representing the unwrapped UV maps with a morph to wrap the flattened UV template with seam guides into the original object's shape, so animating the wrap morph allows one to follow exactly where seam facets go as they transform from 2D to 3D.

    I will release them when I have a useful GUI to simplify application and when I can decide whether they are sufficiently merit-worthy to submit myself to vendor martyrdom.



  • @rokketman I think I've found what you're referring to. At first glance, the torso and limb maps appear to be bilaterally symmetric, but having pulled that template into an image app, and horizontally flipped and realigned a 50% opacity duplicate of the template, the seam edges do not line up with the opposite side of the body. It's especially bad on the leg seams (shown in the second image below), so painting textures will be a nightmare, because you can't just paint one side and flip the texture to the other side and have it automatically aligned.

    0_1507115117052_Screen Shot 2017-10-04 at 10.00.50 pm.png
    0_1507115226203_Screen Shot 2017-10-04 at 10.06.18 pm.png



  • @anomalaus Yes, exactly. It's that lack of symmetry that made it impossible to weight map her. Someone tried and failed. While Sydney still has a soft spot in my heart for her, I won't be doing anything else like this with her. I am going to use Vilter's trick on a couple of the other figures and may go ahead and make the second skin spider gal for one of them. Thanks for taking the time to look her over and confirm my suspicions. I would like that script when you release it. I'd offer to beta test it for you, but I am constantly out to sea and I can't download anything on the ship's computers.



  • @rokketman

    I have good news and bad news.

    During my last days at SM, I had been working on revamping Sydney, and fixed a LOT of the issues she had. First order of business was to reduce the polygon count in the mouth parts (which alone removed about 40K polys from the model, I got her down to a more respectable 62K polys). Next order of business was to fix the asymmetry and the welding and extra verts at the group seams. Next order of business was to improve the UV mapping, which is made up of head, torso, arms, legs, mouth, and eye maps.

    I softened her face a lot, so she is more generic (the original Sydney had a very distinct face that was very recognizeable, whereas this one is a lot more neutral.

    Now here's the bad news. I can't distribute it because the OBJ is totally different than the original (unless maybe I am able to get special permission from SM to do so). She isn't weight mapped yet (weight mapping figures is not my idea of a good time), and I also kept the rigging the same so she could work with all the existing content. Except the new OBJ doesn't work with the face room, seeing as the poly count in the head is WAY different and it would have to be done over again.

    0_1507124248695_sydneyg3.png



  • Oops I mean existing clothing will work. Morphs and textures for old Sydney will not, due to all the changes.



  • Another render of the reworked Sydney

    0_1507130729753_sydneyg3 2.png



  • @Deecey nice :-)



  • @anomalaus

    Thanks!

    I'll putz around with the weight mapping a bit, if I'm successful maybe I can touch base with SM to see if there might be a way to distribute it. No promises though! LOL



  • @Deecey Considering she's the one popular figure SM has, you'd think they'd jump at the chance to put her out again.



  • @Deecey while pondering the beauty of those renders, I wondered how other texture remapping ventures go about their business. If one has the ability to generate a mapping from the original UV texture vertices into the remapped UV texture vertices, then the placement of all pixels in between can be interpolated. Sufficient sub-pixel sampling can account for jaggies or other artifacts, I'm sure.

    There may be a market for in-Poser conversion of textures, if the mappings exist, of course...

    I suppose the face room works on the same principle, with the user providing the image and then assigning the avatar's specific mapping points to the image for re-wrapping onto the figure.



  • @eclark1849
    Exactly, which was why she was the first figure I took a serious look at for updating. Spent a LOT of time on fixing the mesh, creating new morphs and textures, etc. The potential is there, I just have to try to finish it.

    I'm not the best at weight mapping figures (clothing I can handle, figures are a different beast entirely), but I will give it the old college try.



  • @anomalaus
    Question is, how would one handle it if the polygon count is entirely different? I can see if the vertex count and vertex order was the same, but as soon as you start meddling with changing polygons the old UVs are toast.

    The original Sydney had somewhere around 100K polygons, now she's down to 62K.



  • @Deecey and yet the salient features are common to all human figures. How do cranial and orthopaedic surgeons go about accurately positioning their instruments for scanning and surgery? They find reliably identifiable features and make marks.

    My scripts are having to deal with non-one-to-one mappings between vertices and texture vertices (I'm in the middle of debugging why both of V3's eyes morph into the right eye socket, which doesn't happen on other figures which have separate UV vertices for each eye. The figure obviously loads correctly into Poser, so it's just a logic error in my script. The same thing was happening with Sydney's eyelashes, though I didn't comment on it earlier, because as soon as I work out why, the problem will go away.

    Similarly, if the logic exists for a human to determine (tedious, though that would be to do manually, as I once found out trying to make Blackhearted's GND4 perfectly bilaterally symmetric in Hexagon) which vertices should map to where, it may be possible to algorithmically encode that logic into a script, even if it requires manual generation of a table of known mapping points by a human to start with.



  • @anomalaus
    The body of the reworked Sydney is identical to the old one, in that case you might be able to use projection techniques to project the old texture onto the new one.

    However, the face, eyes, and inner mouth parts (if I remember correctly, it's been over a year since I looked at this) were reworked quite a bit as far as shape and polygon count. In that case, the results for projection won't be as good.

    UNLESS ... UNLESS I create a morph for the old Sydney that will fit her new face, eyes, and mouth to the old shape. Then you could dial the morph into the old Sydney, save that OBJ, and project the texture onto the new one with external software that does that type of stuff.

    I'll have to think about this a bit.

    The reworked Sydney will, as far as textures and morphs go, have to be treated as a new figure, even though she'd be able to fit in all the old clothing.



  • Old face and texture left; new face and texture right

    0_1507134673836_heads.png


  • Poser Ambassadors

    @Deecey - what about using RTencoderPY (free at Renderosity)? It should allow you to take the original OBJ, compare it against your new version & code the differences into a distributable file.



  • @caisson
    Hmmm. That's a possibility.

    Maybe I can convert it "as is" that way. Gawd I hate weight mapping! LOL


  • Poser Ambassadors

    Many people have redone Sydney, and it went no where.

    I redid her as well, and well, it didn't get real far either....

    The figure is basically done, uses parts of the original Sydney, and I can't distribute it either.....
    This version came about when I tried to weight map Sydney G2, and found that it is next to impossible to do so due to all the errors in the Sydney G2 mesh. Poser weight mapping needs a 100 percent symmetrical mesh to work properly, and Sydney G2 is so far from symmetrical, I gave up on it and made G3.

    It took me about 6 months to do it, and then things changed. So now it rots in my Runtime....

    0_1507138263358_Sydney G3 ssf.jpg

    0_1507138274285_Sydney G3 ss.jpg

    0_1507138848373_sydney g3 run.jpg

    As you can see, I have more than a few working versions of it.... Both G2 weight mapped, and G3...