Texture Projection in Zbrush and other possible apps - how to make?

Hello! I have few questions about creating custom textures for characters and texture projection in particular. Recently I played with facegen to make custom characters with faces from photorefs and the result was poor in general. So I decided that the best way to make lookalikes of celebs or friends is projecting face textures on the morphed face and attempted to use Zbrush 4 for this purpose, because it has a bunch of tools for projection now. The first tool is ZappLink, but when I use this tool with Photoshop bridge, it can load back in Zbrush only low-res textures of a face, when I enlarge the document size in Zbrush and make high-res phototexture in Photoshop for the 3000-3000 texture map, ZappLink crashes or slows down too long. The other way is to use Spotlight, but it produces poor blurry textures even when projecting hi-res image. I know it's a common problem of Zbrush, and one needs to subdivide mesh to high levels. But I don't know what levels of subdivision must I reach for Genesis or Genesis 2 mesh to transfer precisely sharp distinct phototextures? What number of polys must it have? To avoid overloading of RAM I use only the head or the face mesh without eyes, inner mouth etc. All the same I want to mix and blend a custom face with base textures for Gen1\Gen2. But how to make Zbrush transfer the photo texture precisely? What number of polys must it have? I've never used projection textures in Zbrush before, so I feel confused with all this pro tricks and need a bit of explanation from people who use it for texturing for long.
And what other software can you advice for such purposes? I feel there're better and simpler tools than Zbrush, so it's created for modeling originally, not for texturing. I've read about Bodypaint, but they use a complex workflow with freezing views and PSD layers (Zapplink is simpler) and I haven't found rather clear tutorials about this way of texturing. What else can I use?
Comments
It may be worth looking at 3D COat (has direct painting to maps instead of per-polygon painting) or mari Indie (it wouldn't, as I recall, quite cope with a full Daz figure but should be fine with just the head). As for how many polygons in ZBrush, square your desired texture resolution (remembering there's not much point in going above the photo resolutuion) as a guide.
Thank, Richard, I feel 3DCoat and Mari is excellent apps for professionals, but I feel Zbush for me is more preferrable as it has much more tutorials and available support than any other modeling software. And I like its Zapplink for ability to alter textures in Photoshop. But I cannot figure how to get nice resolution of texture - If I want to get the texture resolution = 96 dpi as in the base textures of G2F, so as I understand I must use photo not lower than 150 dpi, true? Or the resolution of photo must be equal 96 dpi too? The size of texture maps I decided to leave base too - 4000*4000 px. Squaring this number gives me 16 000 000 polys, if I understood it right.
I cannot grasp the difference between direct painting to maps and Zbrush painting - what is more precise and easier to perform? Zapplink in Zbrush is an ideal way of texturing for me - I can morph a head in DAZ Studio, refine it a bit in Zbrush, and then export it as different views to Photoshop and fit the face on a photo to the morphed head. So the texture will cover the face precisely where I want. Has 3D Coat any similar feature?
PPI is not a relevant factor - it's just a number at the front of the fiel saying how big the pixels should be when printed. All that matters here is the number of pixels.
ZBrush doesn't paint textures - when you paint in ZBrush you are changing the colours of the polygons, which is why the number of polygons matters; once done you then bake the polygon colours to a map (Bake Polypaint to texture) using the UV mapping of the model. Mari and 3D Coat can paint diectly to maps, regardless of the polygon count (3D Coat can also do per-polygon painting like ZBrush, I don;t think Mari can - at least in the indie version). I've not used the 3D painting in PS (I'm on CS6, where it is fairly basic) - is that what you are using, or are you doing 2D paints on different projections? In any event, I suspect some of your blurring may come from converting backand forth between pixels and polygons as you use ZAppLink - in 3D Coat you are painting in 3D space directly to a map so there's no blurring from the back-and-forth.
Yes, when working via Zapplink I can transfer 2D photo image to different projections - front, back, right etc.. It's the best way of texturing in Zbrush for me now. I don't know if 3D coat has such a feature, but when in Photoshop I need to fit a face on a photo to the projection view of a morphed head, I can scale the photo, stretch it, use Liquify etc, and when both images coincide nearly perfectly I send it back to Zbrush and get pretty accurate texture, . I found this way is the best, though you may be right about blurring problem. I wonder, can 3D Coat connect with Photoshop so easily or has it any tools to alter the photoreference within itself?
I'm not sure, it's not an area I've used - it's certainly possible to get a Spotlight-like feature in 3D Coat.
3D Coat can connect to Photoshop. I have not used this feature so I can't speak to the results.
https://vimeo.com/5983289
3D Coat can certainly scale, stretch, flip, rotate, etc. reference images that are used in per pixel painting to make them directly line up with the view of the model.
https://www.youtube.com/watch?v=Da8sheHAesg
They also have a 30 day trial, and they usually have a sale around Thanksgiving and Christmas.
http://3d-coat.com/download/
If you are used to working in Zbrush, 3D Coat might be very different, so I would recommend the demo to see if it works for you.
The one thing I really wish Daz had was parametric or procedural rendering, where the surface hugs (to an extent) the curves of the surface.
i subdivide G3F to 14.5 million polys for painting with Spotlight
Oh, yess, it's just that counterpart of Zapplink, which I've read of on some sites but could not find exactly what it is in 3D coat. Than you so much for that Vimeo link, Destiny's Garden! As I see, it's even simpler than in Zbrush, and if I understand it right, 3D Coat can apply textures straight on the UVmap of any pre-mapped model like head of Genesis etc, true? So if instead of painting I use edited photos as layer 1 above, I guess I can get the projection on the model like in Zapplink. If so, it's a great feature of 3Dcoat and maybe I'll give it a try if upgrading Zbrush to x64 won't help me with crashing. Does 3D Coat require high amount of RAM or 6GB on my laptop will be quite enough for simple tasks like texturing and morphing? I wouldn't like abandon Zbrush, of course, but sometimes its sudden crashes drive me crasy and I need more calm workflow. Even Hexagon 2-5 and Sculptris are more comfortable for working for me so far. But I've read 3dCoat has some problems with making morphs of Gen1\Gen2 figures - something about symmetry and vertex order. And there is definetely no bridge to DS.
3anson, thank you for the tip, I think 14.5 million polys will go normal for my system - is it for the whole G2F of for the head only?
There is a "room" or a set of tools in 3D Coat called "Tweak." Those are the ones you want so you don't disturb the amount of vertices. I have not noticed any problems with symetrical morphs, but I don't do a lot of morphing.
3DC can defiantely suck the RAM, especially if you are using texture maps over 2048. 6GB should be OK, but again I recommend the trial to be sure.
"3D Coat can apply textures straight on the UVmap of any pre-mapped model like head of Genesis etc, true?" True. It is very easy.
Well, as for Zbrush, at last I upgraded it to R7 x64 and now can say the performance became really better, new Zbrush can use all my available RAM. So now i can subdivide heads to 10-12 mln polys and Spotlight works much better too. But I have problems with Zapplink, because I don't understand its workflow fully, mainly about corresponding document size in Zbrush and Photoshop. Here is one typical example of a problem - when I want to use high-res and large photos, I resize document in Zbrush to about 5000 px at height and export the view to Photoshop. And I zoom the head so that I can apply photoreference onto the face without scaling it down too much. In Zbrush it looks like this -
But in Photoshop I get the document of corresponding size but mainly empty with the small head in the center, so I cannot avoid scaling down the photo.
How to transfer the large frontal view of the head to Photoshop to keep its size mainly for the head? And the second problem - my Zapplink works mainly when I use views (front-back etc), but without them it cannot load the texture from Photoshop on the 3D head. And I've noticed that it works only when I turn on Polypaint from texture and Mrgb mode. But I've seen a lot of videos where people use Zapplink without Polypainting and make it to apply high-res textures on the mesh straightforward. What should I do exactly to get such result, which steps? So far I do such steps
1) I load the mesh, turn on edit mode
2) Subdivide it to level 5 about 650 thousands polys for Zapplink
3) Create the new texture from UV map about 4000 or 4096 px wide and fill it with white color
4) Enlarge the document size to 5000 px height, then rotate and zoom the head to the full frontal position
5) Turn on Polypaint from texture and mrgb (it requires mrgb checked at first). Or else Zapplink often fails to load the texture onto the mesh.
6) Check front view and export to Photoshop
7) Load the photoreference above the Front layer front, fit it to the mesh projection and merge the photo down with the Front layer, applying the mask. Then I save the temp PSD file and port it back to Zbrush
8) Most of the problems I have at this step - often Zapplink thinks a lot, saves quicksave files, but fails to texture the mesh. It happens even with polypaint turned on.So I have to redo all the job. How to avoid it? I feel I do something wrong at some step. Some times I got the textured meshes but textures were too blurry. So I have to fit the photos in Photoshop to the head and apply it later via Spotlight. The result are not exactly what I want, but with higher levels of subdivision they are suitable. But I want to texture in the simpler way via Zapplink, as many people do. Which steps do to make exactly in the case of texturing via Zapplink to get high-res textures applied on the mid-res mesh?
Ok, so I took the Genesis 3 Male at base resolution (17K polys) into Zbrush and painted a red line oh his check. You can see how it spreads because polypaint is dependent on the mesh resolution. If you use ZappLink now and bring it back to ZBrush, when you pick up the canvas, it will also bleed because it is still utilizing PolyPaint.
The key to using higher resolution textures that maintain the detail is to set your UV size to, say, 4096, and create a new texture on the model before you ZappLink. The second image is the same base resolution with the new white texture, picked up through ZappLink. It opens in Photoshop, and I used a similar sized brush and painted a line on the other cheek and saved. When I re-entered ZBrush, I picked up from canvas and you can see the line is much cleaner.
Oh, I just checked my doc size. I didn't change it from what it was set to (1063x809). http://prntscr.com/94hh0g So for this example, I didn't enlarge the document or anything.
When you intend to use a larger size document, make sure you use the zoom button (if you're using a default ZB layout, it should to on the right hand side of the document) and zoom to make sure you actually see the boundaries of the document. I am guessing that when you created the new size, you didn't check that and the dcoument extended past what you could see.
If I understand your workflow, you are trying to create the texture after you have done the ZappLink and returned to ZBrush, so that is adding to the problem as I explained above.
You can't paint on the model with a texture on it; you can only do that with polypaint, but that brings back the problem of dependence on the vertex count. What I might recommend is doing your basic painting in Polypaint (for texturing, you can get decent painting with about a million polygons) and then ZappLink for doing all your detail work. There may be other workarounds; I'm still constantly learning.
Here's a further example:
Here’s a small detailed symbol on the zapplink image in Photoshop: http://prntscr.com/94hrh1 Notice how small it is compared to the rest of the face.
Here’s how it comes in and ready for pick up: http://prntscr.com/94hrnx
The result: http://prntscr.com/94hrsk Now you can see the edges are a little jagged, but a level of subd or so and the graphic would be much better and I wouldn't be working with a terribly high number of polys with just a couple of added subd levels. Also keep in mind that the line I painted on the polypaint pretty much bled out, and this symbol takes up a much smaller space than the line and stayed pretty intact.
If you're working with Spotlight (which is a great tool), you would need a high amount of geometry to support the graphics you are using to project. So again, use polypaint and/or Spotlight to do most of the work and then any real fine, clear details could be done with ZappLink.
i always work on the full figure, because i am usually doing a full skin texture. importing just the head and painting on it would result in a bad seam that would need fixing after. by working on the whole figure you reduce seam artifacts tremendously.
Yeah, it's much better to do as one subtool, or object. In another thread, someone was trying to subd and hit a stopping point at 6 levels on his G3 figure. I suggested removing the inner mouth and eye parts including the lashes, when doing textures, as it allows you to do another level if needed.