I’ve posted a similar question at the “works in progress” section but I’m not getting much help…
I am modeling a realistic human form and will need to start on the details soon - veins, skin, wrinkles etc.
I would like to know how to go about with this. Do I need to subdivide the model untill my system chokes and then start sculpting the normals and map them - or is there another way? Normal mapping is great but you always need to use a high res version of the model which is kinda a catch 22.
You could use texture paint on a B/W high-res texture, then use that as a displacement map. The drawback of this method is that you can’t see the bumps in the viewport, but only when rendering. So when painting, you’d have to set the texture up as diffuse to see it. I’d start with middle-gray and then paint both darker and lighter on the texture.
Once the render looks OK, you can bake that displacement map to a tangent-space normal map.
Alternatively, if you have a UV map set up, cut your model up into the parts defined by the UV islands before subdividing into high-res. I’d do it in a separate file for every part, just delete everything else from the mesh, duplicate + subdivide as necessary, and keep the original lowpoly mesh around for baking to.
That way you can get more performance + higher poly count where it matters. If you bake the parts (selected to active) and then combine the image in an image editor, you might get reasonable results.
well… there’s another way. -you know how you can get a greyscale specular from something you’ve painted -like a diffuse color map…?
Well… there’s this thing in gimp (-makes tangent space normals from greyscale-have to find and download) that makes tangent normal maps from greyscale images… -in the end, you save on processing department to generate the maps, but that means you have more awesomer ideas to try quicker than if you painted on a huge trillion poly mesh… -sides… those painted on the model directly will always look kinda sloppy… where as in 2d you have MORE control…
However… I shit you not… -it gets harder past that point, (since you’ll realize that a lot of fine tuning is involved -somewhat) but it’s better than killing your machine over iffy normals you won’t feel like correcting… (the most processing your machine will do using this route is during rendering… and that doesn’t even take more time than if nothing was mapped. -cool, huh?)
(did this on a slow laptop… and the maps I made myself…)
all it is is a subdivided cube, smoothed and one 3 maps… specular, color, and normals…, and alpha. -no baking required XD -speculars HAVE to be derived from tangent/normal/bump maps (basically same image) or you’ll get some really messed up results…) -all but diffuse/color are the same image, but inverted greyscale. -um… the only map that’s allowed to be off is the color map… -the diffuse. -just remember that if you forget everything else.
dunno what to call this world… -(no multires was used in the making of this model )
Interesting proposal. Dunno if I’ll go through all that trouble though.
That is brilliant! So in fact you paint the normal maps instead of modeling it. But I’ll have to try it first to exactly understand what it does. I don’t use gimp but from what I gather it has some pretty amazing plug-ins. The only thing keeping me away from the prog is the non-existent pantone range. I could propably type in the reference numbers myself from my pantone library but thats … I’m drifting off the subject.
Aaanyway. I haven’t started on her yet. Still had to fix up some areas - she’s got a little more muscle definition now in the front and the back. She’s not the ideal sexy type but thats just it. She’s a normal girl with maybe a bit too much fat on the wrong places and too little tone.
I should’nt be asking this here but if you find something annoying or awkward on the model let me know. I’ve posted her (with close-ups and toplogy ) in the works in progress section for some feedback.