Skin textures and photo sourcing

I want to take a photo of myself and some friends and resize/distort/whatever it takes them to fit teh SL face mesh templates so I can put us into SL as accurately as possible. Can anyone recomend a good (and preferably free) program that allows me to do this effectively?

The big three are Photoshop, Paintshop Pro, and GIMP. Off the top of my head, here are some quick pluses and minuses for each:

Photoshop


Pantshop Pro


GIMP


As for the "effectively" part of your question, all three of the above programs are more than capable of doing what you're talking about. However, each one is only as good as the person using it. These programs are just tools, nothing more. Turning a photograph into an SL skin is a highly skill-intensive process. Having good tools makes a difference, sure, but the onus is on you as the artist to make it work. What program you use is really just a footnote to the artistic process, not the process itself. There's no software in the world that is smart enough to make a photograph into a skin on its own. In the end a pixel is a pixel is a pixel, and that's all the program knows. It's up to you to make a collection of pixels into a good looking image, and that takes a lot of practice, time, and dedication. There's no way around that.

Anyway, you can download trial versions of Photoshop and PSP (from www.adobe.com and www.corel.com, respectively) for free, and you can get GIMP from www.gimp.org.

The question I have is that with all of your talent and expertise, have you ever been able to import a realistic photosourced face for an avatar?

If you have not been able, then I would most likely not stand a chance.

Playing around a bit, what I want is something similar to PSP's mesh warp tool, but with teh ability to designate my own warp-points, instead of having to use the fixed grid points that that tool gives me.

I've seen photo-realistic faces in-world. So it is possible, just very hard; and the lack of bump mapping/proper shaders doesn't help. And no, I can't do it. I've tried!

Theres that 'Smart Objects' function in CS2 that would help make it easier. Its not that hard to match it to the template.

However, the main difficulty with photosourcing faces is adjusting the colouration and getting rid of seams when placed on a skin. That'll take more effort than the warping itself.

Sorry I forgot to mention. When taking photos of your face, try and avoid getting shadows and highlights on your face. Preferably, set up a few diffuse lamps around to make your face look as 'flat' as possible.

You can add in some shadowing later in the image editor if its necessary.

What is this "CS2" program, and who makes it?

CS2=Adobe Photoshop CS2

Hi Warda,

Yes, it can be done, and no, it isn't easy.

The first step, of course, is taking the photos. Take a bunch, under lighting that is as neutral as you can manage. Not just from the front, but from the side (make sure your ears are showing,) from below the chin, from a 3/4 view, and from the back, and 3/4 back, if you want your hair as part of the "head.

Take photos with your eyes open, and with them closed, with your mouth open just a bit, and with it closed. If you want to be really accurate, take some of your teeth, tongue, and eyes, too. (For those, you'll want to hold the flesh away from the teeth and eyes with your hands. The more you can see either the iris and gums, the less retouching you'll have to do.)

The second step is done in Photoshop, or whatever graphics program you prefer. Use the templates, and put them on the top layer of your image. I'd recommend mine or Chips, because they've been subdivided. (If you're using mine, the "wires" are on a transparent layer, which makes it easier. I don't remember if Chip's are like that.)

Reduce the opacity of the UVs to something that allows you to see both the photographs and the UVs clearly.

Open the straight-on shot of the face, and copy one element, for instance, the nose. Leave plenty of room around it, for blending.

Put it on a layer under the template, and Resize it so that it matches the nose shown in the UVs as closely as possible. Put the nostrils where you see the UVs kind of pinch, and the edges of the nose where the edges are.

Now, if you're using a version of Photoshop that has it, use the Liquify filter to fine-tune the photo to fit the UV wires.

Repeat for each of the other features, placing each on its own layer.

Blend between them, using the Mask tool, or the Clone Stamp tool, to fade the edges out completely.

Do this all over the head, using the other shots to get the bits that belong in those areas. For instance, the under chin shot to do the skin under the chin, the side view for the cheeks and ears, and so on.

Remember, as you work, that the image is going to wrap around the head, and don't be afraid to use the shot that's "head on" to the area you are currently working on.

Unless you are going to be using another painted texture for the body, "fade" the texture into transparency at the neck, so there won't be a visible seam. (The edge of the neck should be black in the Alpha channel.) If you're using another painted body texture, then fade your image into that one, matching the colors as closely as possible.

During this whole procedure, you'll want to make test .tgas fairly often. Don't actually upload them; just check them on the upload Preview head, to make sure the features fall where they should.

When you're finished, and there are no seams in your image, you'll want to use an adjustment layer to make the color a little darker, and a little more saturated than the photo. Otherwise, you're likely to be much paler than you expected in SL. (Of course, if you're a Goth, and this is what you want, skip this step. )

Upload, match the body color to the new head color, and you're all set.

There are other ways to do it, of course; but I've found that placing the features individually generally gives me the best result.

There are programs that allow you to do this more easily; for instance, MAXON's BodyPaint allows you to project an image onto the model, so you can do it with several layers, using projections from various angles, and blending between them. But, since the .obj models we can download aren't exactly the same as the ones in-world, I've found that method to be less than optimal. (And for what you'd pay for BodyPaint, you could commission an in-world skinner to do the work for several of you.)

Hope this helps!

Damn. That was fucking long. At least it was informational.