There are more then a few research groups tackling the problem of turning objects in photos into 3D models. One group at Stanford has been very successful at doing it. However, photos to 3D is still in the research world as of now, no consumer products have a feature capable of it.
There are also 3d scanning solutions available but they tend to be very very expensive. There are also some cheap alternatives to this, (google “Milkscanner”), but frankly, you would spend more time trying to get that working for quality results then you would spend just learning to model. (Or paying a modeler to do the work for you would be a cheaper route.)
No need to reinvent the wheels and who knows how long it is going to take me to learn Blender …
wondering if there are available sets of human objects in Blender resources, where we could just change the face, skin tone etc ??? Change clothes, change movement perhaps ?? Like a preset we could modify ?? That could be the best alternative solution ???
it is comming soon (next year probably) http://code.google.com/p/libmv/ currently there are a few programs photomodeler, image modeler, and others. even microsofts photosynth is being adapted to create 3d models. i know most are thinking microsoft and blender? but watch this video starting at about the 25 minuet mark. http://video.google.com/videoplay?docid=-5778605234686979545 it does look impressive. it would take more processsing time to go from sparce point to dense point (libmv will be dense point) but it can be done by the same methods. it’s still just structure from motion. that video is about a microsoft program phototourism, improved and renamed photosynth. but the modeling software for phototourism has been relesed as opensource http://phototour.cs.washington.edu/bundler/ . and i admit i was suprised, linux is the recomended os for a microsoft partnered project. i’v tried running it on vista via cigwin but it gives and error about a missing cyg62.dll (or something like that) and closes. i did download the dll and put it in the bundler folder but evidently thats not the right place.
from there it should be easy if you can get that far to draw edges between the points and aquire the face textures directly from the photos the faces came from.
all this was in response to the original post, not about building humans and clothing them. this would be all about creating the enviroments (buildings etc) around the humans.
I came across a app a few years back that makes accurate models of photos. It’s from a German software house. You make multiple photos of your model on a turntable. I could look it up again …
If you’re not interested in learning to make things yourself, DAZ products are fine indeed.
If you want realistic renders of high quality models, you’d better prepare yourself to spend at least a few years to learn everything about the wonderful world of 3d.
You don’t just jump into a swimming pool without learning to swim first, do you?
D Sculptor exports models as:
Lightwave .obj (also for Maya & Bryce)
3D Studio .3ds
VRML .wrl
DXF .dxf
Shockwave 3D .w3d
Direct X (.x)
It’s quite expensive. (500bp, 700+au) It might be a must though… It would seriously help.
Although, I saw something in development a while back… I think it was a phone app which did the same sort of thing. I’m still looking, it may have been on the Unity Community forum.
Photogrametry has been around for years. You can see early versions in the first Matrix movie. There is a recent online app that converts photo sequences into uv wrapped mesh too. Don’t know if this is the sort of thing that the Tomato branch could aspire to.
well you can sort of do it if you have a top/side/or front view…
UV map a plane to the image
subdivide the plane like 1000 times
add a displacement modifier, with the image as the texture
this will result in sort of an effect like the profiles on coins, flattened but contoured topology. You could then perhaps scale it and retopologize it
Not sure how well either eventually manage. I tried the first, a year ago, but failed to follow through due to time constraints. Maybe you’ll have better luck:)