This may be a silly idea, but I was just thinking that if you had a budget of say 10000 polys for a next gen main character, you could model a hell of a lot of detail in it and not bother with a normal map( in some instances)
Sure you would n't get those subtle things that normal maps bring out,but my point is that you could still make a pretty decent looking model without it.
just a thought though . you may have guessed that I am not a big fan of normal mapping right now
Replies
What I'd like to see is a more split use of normal mapping. Fine details could be added to armor, mechanical elements, while high poly can be used for skin and clothing areas.
I see it like pyromania. Normalmaps for clothing and armour, in some cases for hands. I personally wouldn't use normalmaps for faces that are quite highpoly again, rather use it like a bumpmap to provide detail information. That avoids problems with morphtargets, too, especially around the mouth area. Of course, if you have a real wrinkly, detailed or "monsterlike" head, you might need even more polies in order to get a relativly smooth looking mesh and normalmaps would be the way to go.
[/ QUOTE ]
depends on whos modelling the head. Pior's work takes normal mapping to a whole new level compared to the workd some people do. His low rez meshes really work.
However I was more referring to "plain" faces like you find in Silent Hill or Final-Fantasy.
'When you try and redefine a character whole topography with normal maps they look cheesy.'
I totally agree with that.
For example the head I am working on has about 1400 polys. I reckon it would only take another 1000 to make it look convincing as a high res head
BTW i liked the style pior used for that head in P and P.
So anyone think that 1000 polys is a good guesstimate for next gen characters?
Though if I made a 10k poly model, I think I'd normal map it, just to smooth things out and add in some minute details. You also don't have to worry about bad silhouettes with a 10k model. Besides now games are starting to have anti-aliasing, so little edges won't be noticed anyway.
I don't know, I just think as long as you can normal map it, you should. Though I guess if you can't do good high poly modelling then maybe you shouldn't. Or if the game calls for a more stylised look, and more cartoony textures would do. Although Psychonauts was very cartoony and it still used some bump mapping in it.
NOrmalmapped chars in next gen games are already 10k with normalmaps often.
A problem I see in some games is overaly uniform use of normal maps - lumpy clothes look good, but you need to tone that down on faces.
At the simplest level, just a meshsmooth with no extra detail will round out the lighting.
perhaps thats more to do with poor implemenatation than the limitations of the tech.
I have found what is working for me is to design the character fully, get the diffuse/specular spot on and then think about the normal map after as an additional tweak.
I look forward to then time whane we don't need em personally
Yeah, but one more thing to consider (as Poopybabycutey stated it in his tut) is the silhouette 'match'. Ben recommends building a rough 'lowpoly' model that meets the project/engine requirements beforehand, in order to see how far you can get in terms of detail and silhouette. Then he recommends thinking about the highpoly, so that it fits well (because some details can't be faked with the nmap only, they need a shape to lay upon)
I'm really not clear, beh. Better have a look at his tute... I think it's very clever and effective for complex models.
Kindof like low level/mid level/high level details in classic sculpture...
(perhaps I am just pissed at all the extra work I am having to do)