Question for the 3D modelers out there.

[TunaBreeze]

A little bit about me - I'm an aspiring game developer, I started work on my engine about a year ago, hoping to have it done in 1.5-2 years.

I'm thinking ahead and trying to anticipate the best way to handle the 3D art for my games, specifically the creation of the 3D meshes. I've done tons of research in to equipment that can digitize objects in to 3D meshes and it looks very promising. Short creation time being one of the main factors, and ease of use for myself, so that I can easily convey all the concept art ideas I have floating around in my head. Some objects could simply be purchased and scanned, and others could be modeled out of clay.

I have plenty of knowledge about the structure of 3D files, and how to implement them in to my engine but I know very little about the 3D model creation process itself. My question is for the 3D modelers out there. Given a high resolution mesh as a guide would it be relatively easy for you to create a new mesh with the desired amount of polygons, and add the necessary bone structure for animation, and do the animations? Would it drastically reduce the time it takes to produce a model?

My aim is to be able to go from clay model to 3D file in the shortest time frame possible to speed up development time and save money in the long run. The initial cost of the equipment is pretty expensive but I think it would pay for itself in the amount of time saved modeling if I'm able to set up a system as described above. Thoughts?

[jlv]

I think most modelers can decimate vertices automatically. At least Blender does -

http://wiki.blender.org/index.php/Manual/Decimate

[XIX]

I don't think it helps, unless you can use as is, exactly what you get out of the scanner.

If it's going to take an artists to clean it up into something usable then just give the artist an object/photos of object and ask them to model it. The 3d scan seems to be a pointless step.

[brianhay]

Low-polygon modelling is an art form in itself. In the case of characters, you need to put the detail at the joints so that when the model is rigged and animated, the muscles flex and skin folds naturally. Texturing is also tricky.

If you're modelling for cinematic (prerendered) output, yes it's practical to scan a detailed clay model but for games, even with mesh /reduction decimation tools and normal maps to approximate detail on low poly models, I think you're better off modelling low poly to start with.

In AAA games, where normal mapping is common, the extra surface detail available in a scanned clay model is useful but for casual games it's not worth the hassle IMHO.

[electronicStar]

I modern AAA games, they use tools such as zBrush to model directly in the computer and to generate complex normal maps (things that would be difficult to create even with clay or by drawing, see Gears of war for example).
They have specialized software to create low-poly+normal map versions from a high poly version.

[Pkonst]

Imho, there is no need in such complex technique. It's really easier at this moment to sculpt your model in Zbrush/Mudbox and convert high-poly mesh to Normal map for your lowpoly mesh. It's really not that hard process.

[Sysiphus]

I was in that thinking...maybe a decade ago, I don't remember...At the end, is too much time consuming. There's a way of scanning, with sort of a pointer which , previously having done a "painted wire" with the mesh wires as low pol as desired, and knowing how it must be for good later on joint bending and animation, you're actually scanning only some points instead of a whole cloud of points of some millions of vertices, which anyway, tend to kill most machines, specially hard is to edit that, though possible (but many standard modelers which are good for general modelling, simply die with so hi res meshes, unless you have a terrific hardware)

That way of scanning, instead of a points cloud, maybe would save som time of that too long proccess. Other way would be scanning in its actual millions of verts, and then use something like Silo's topology brush, or blender's newest retopo Paint, so to get a low pol mesh painted "over it".

And still...you would need to uvmap it correctly, this means not only good, non distorted UVs, but also eficient ones, so to use well the texture pixels...this itself is quite of a task.

Then of course, the texturing, and it's needed quite of experience to do it well... though surely the first 3d casuals wont (or don't) need much texture definition...

Finally, one of the most complex parts, rigging and animation. Is a profession by itself, and I have known animators not wanting to do the rig , but let it to the modeler (I prefer to do all proccess in my own) ...others are all the opposite.

What saves you the scanning, is the fastest step: modelling. Provided these days there are very fast modelers like Silo, Wings3d, etc, is not a real advantage, but probably you're adding extra steps and slowliness in that area, specially if it's for casuals 3d real time.
Is not as simple anyway, the procedure of generating the normal maps (in case normal maps are clever to use with the target hardware for casuals) , doing the low pol, and a collection of technical matters, needs technical touches everywhere, I am afraid.

And anyway, specially low pol is fast to do, with software.

there was an expensive 3d scanner which by a video cam recording system and 3d recognition, could even create textured models, t owhat level of accuracy, i dunno, but I don't trust much in that...Not for low pol games, at least...

My intention so many years ago, was to avoid to learn all what...I later on had to learn ;) As I would have rather let it be 10% software, 90% traditional sculpting and painting, way more enjoyable ;)