I’m pretty new to 3d modelling and Blender (I’m still in that “it can do that?!” newbie whirlwind). I’ve been mucking around with a dragon, based loosely on the artwork I remember from my AD&D days (many years ago) and it’s got to the point where:
a) I’m not totally embarrassed by it and
b) I need some feedback on what to do next
The textures are quick and dirty procedural textures, with the exception of the eyes which are based vaguely on the Pixar eye tutorial and actually feature an original bitmap™ :).
Any observations on things to fix are welcome. And if anyone has any views on how to approach texture-mapping something like this (OrCo style like Enrico Valenza or LSCM-based UV mapping?) that would be appreciated.
I think you’ve done a good job on this, particularly the texture and the head. One problem I see is that the style of the head isn’t really reproduced in the rest of the body. The body is just too simplified and un-stylised compared to the head.
Also, the sharp edge on the its head looks too sharp, maybe you could smooth this out a bit.
Actually, at one point the body was a lot more complex. In fact, far too complex - I had made the mistake of going nuts with the knife tool and ending up with a massively complex collection of triangles that really didn’t work particularly well. A bit of reading and a lot of fiddling later merging vertices cleaned the mesh up a lot.
Is it better to put a lot more detail into the mesh? Or should I think about having a detailed normal map to indicate texture. Reading through modelling tutorials, there seem to be two schools of thought.
Make a massively detailed mesh and let that describe the surface (ala Doom3 style modelling, where the original models are 250,000+ polygons, the detail is baked into normal maps and the poly count is then reduced for the game). That’s not to say I intend to actually make a game with this (but I was toying with the idea of messing around with GameBlender and doing an “attack the helpless villages with the dragon” game although that is realistically months away at the earliest).
Make a basic mesh and then use a normal map to “up” the detail level and indicate muscles/creases/structures that way.
I have plenty of books on human anatomy so I have good source material for adding detail to the shoulders, arms and legs. The hands and head are already fairly well detailed (the palms of the hands, for example, have distinctly human-style pads) but the rest is currently fairly straightforward.
And maybe I will take some sandpaper to the head ridge. It is a little severe
You are basically right in pointing out the normal map/detailed mesh issue. I’m usually prone to the second because normal maps are anyhow difficult to UVmap and cannot give you all the level of detil/variation you need.
Furthermore normal maps need vertices
Probably a fairly detailed mesh, with high SubSurf AND a bump (not normal) map is still the best thing.