I'm thought I'd try to recreate Jill from Resident Evil, as an exercise to teach myself how to really model (high-detail, advanced techniques). So far this is a huge headache, but I think I figured out most of the tricks. Going for high detail will hopefully simplify the texture making process, no photoshop nonsense.





The hair is a mess, I know. WIP. I didn't do much on the face yet, those are HARD.

More butt!

Anyhow, looking gneat

Is she camera shy?

Looks pretty good so far. What program are you using?

Ever tried Zbrush? It will allow you to import a model, and 'paint' detail onto it.

Yea, what prog are you using? Also, some modeling notes, the shirt seems to be made of impossibly thin fabric so try to add some depth to it, her elbow seems like it has arthritis or something when looking at the back, and her fingers seem to be too thin, even for a girl without knuckles. The clothing is the easiest to fix, the rest are harder points to get right, I can't even get it right on my human model. Also, symmetric folds in cloth is never a good thing, but I assume in the final model this can be adjusted. One more thing I noticed is the cloth under the harness by her breast (1st shot) is not likely to occur as well as it looks like the wrinkles over her breasts has something strange going on (this might just be a rendering artifact tho, can't tell w/o a headon shot). Overall I like the model, keep working on it, can't wait to see the finished product.

PS - I would like to think I know somethings about doing faces and modeling in general, but I'm probably deluted so don't take me too seriously.

Yeah, the breast area does look weird. It's that divot in the middle. If the fabric is stretched tight, it shouldn't sink in there. There might be a fold, however. Remember that folds radiate in arcs away from spots where the fabric is tight. You've gotten this right on the legs (Though they still need some work, they are ridges right now. They should be more like cliffs; gradual slope up to a sharp fold.) but not on the breasts or shoulders.

A lot of the fun of modelling high detail meshes is to keep the polys low. A lot of the clothing folds you have modelled out would do much better as a normal map on a lower poly model.

Texture detail is an essential part of modelling "high detail" meshes and therefore is very questionable to ignore. It's generally a lot easier to have more texture detail than mesh detail. Obviously as technology progresses the poly limit increases however it's no excuse to start bashing out details in polys rather than pixels.

My recommendations:

- put more definition in the edges of her clothing, and less in the folds. Make a bump or normal map to represent the folds, this way when you distort the model in animation sequences, the folds will look more realistically lit and less "built into the clothing" as they will when animated now.

- Wear some similar clothes yourself and observe where folds occur when restrained by straps etc. Some of your folds are very well placed while others seem less necessary.

- Anatomy: The upper arm (near armpit) and wrists could be slightly thicker. Needs a more prominent butt, looks really flat right now, mainly because the upper leg and butt look in line. I recommend curving the thigh more (when looking from the front/back).



I know I've gone on a rant and a half, but don't let my opinion put you down. The model is of good standard and has great potential. Just a couple of things to iron out and it will be looking great.



Any chance we can have a screenie of the mesh?

That's exactly wrong. Most modern models now are done at incredibly high poly counts with all the surface detail modeled on. The low-resolution mesh and normal maps are generated from that model.

Most modern environmental textures are made that way too.

I'm using JED beta .951 :downs:.


Just kidding, I'm using Blender :). I know about Zbrush, it looks amazing, but rather expensive and of course I'm hesitant to get it.. at school I'm not studying CGI, or any kind of computer-aided art design, or any art at all :(. Blender has a similar sculpting feature to Zbrush, and a good artist can potentially make models just as well, if only the workflow is slightly more complex, which I'm a little used to (JK EDITING ANYONE? LOL! :suicide:). I know I'm missing out, but you can't have it all, at least not yet :P.

I'm aware of how the pros make models at ridiculously high-detail, and then easily derive low-poly models whenever they feel like it, that's what I'm trying to learn. 3d art programs, including freeware Blender, have all the features to work this way. My game plan for Jill is to make the high-detail mesh, UV unwrap it, bake various details to 2d using those UV coords, make a low-poly mesh around the high-poly, UV unwrap low-poly mesh to the same-ish pattern, apply the detailed maps, party.

Here's the mess errr mesh.


It's made poly-by-poly then subsurfed for the first pics. I'm gonna work on removing badly flowing edges to permanently subdivide the polygons, and start using sculpting mode instead.

Go Blender sculpt! Gonna work out the symmetric wrinkles..

Electronik supersonik mats not permanent

Yeah apologies I forgot that people didn't use textures at all anymore. Colour is SO overrated.

JM's right, bump maps are generated from super high poly maps, then the models are reduced in poly count and the maps applied (:eng101: see Doom 3 for example). In CGI animation however this effect is minimalized since the animation is prerendered. Hand done textures are used to add color variences and extremely fine details such as fabric texture (Last time I checked at least, it might be possible nowadays to model that with polys with a displacement map).

Now onto the model. I'm going to rant more and act like a modeler again, so please ignore my less than amature opinions. From looking at teh mesh, it seems like it won't do too well in animation, poly lines should follow the muscles as much as possible because when we move, thats what moves too. It looks like it around the upperbody and knees, and even tho its clothing, it wont easily fold correctly. Also there are prob too few poly's in the ankles and arms to make a smoothe animation (she has muscles right? XP). Last thing, her belt around the waist has too few polys, it seems impossibly thin (might be the lighting/angle) and I can see very sharp angles and on anything high poly, you should never see this unless it is suppose to be there.

Its looking good tho, and feel free to bash my models up and down too since it feels like I'm just ranting here. XP Also, OMG YOU DID THAT IN JED, \/\/T|=, |_| 4|=\3 /\/\y |\|3\/\/ /\/\4st3r *bows*

The arms look a bit funny, particularly from the shoulder down the the elbow

I know this is getting very offtopic but I just want to say that making a normal map is not a substitute for proper texturing. I'm dropping it though, there's no point arguing over it because in the end it makes little difference.

My advice to you __Ace_1_ is to watch the distortion areas for animation, basically where all the joints are, as in their current form you're going to get some funky issues. You're cutting across faces in an unusual manor to get the detail you're wanting.

I've attached an image of the basic theory of cutting into meshes to make folds. The advantage of using this method is that you can keep your mesh in a 4-sided poly state which is ideal for rigging/animating at a later stage. Just a cleaner way to model. For more advanced cuts (such as your diagonal ones, you just follow the same idea, cutting only perpendicular and simply creating more corners. It's very likely you'll have to add more loops in to get the same effect however it will distort much better).

Also, would be worthwhile adding more loops into the belts/straps at some point as they're very polygony, but I bet you're aware of that one and just haven't got round to it yet.

Otherwise it's looking nicer with every update.

No, it is an integral part of modern texturing.

That's going to be a texture resolution issue. Most games will use just a single texture for the entire character, so they can render the model in a single draw call. Texture size is limited, so..



Modern 'texturing' does not use texture maps. It uses Color maps, Specular maps, Normal maps, Bump maps, Displacement maps; texture maps are obsolete. A 'texture' today is some combination of these elements. Most of these items can be generated directly from a high poly model. Color maps aren't terribly detailed usually, they aren't like textures used to be - they are flat; most modern games use no baked lighting in color maps at all. All the depth information is in the normal, bump, or displacement map.

Blender should have the ability to create this low-poly version automatically, while keeping the UV mapping consistent between the resolutions. It should also have tools for generating a normal map given the high resolution model and the low poly version. It's probably all contained in one tool. And if not 100% chance there's a plugin to do it.

Diffuse maps, gloss maps, normal maps, height maps, emission maps, variously encoded into different channels on different textures.

Right JM, I forget you're a prodigy of the games industry/entire freaking world.

I can't be bothered arguing any further. Ignoring colour texture detail in favour of normal maps is retarded, simple as that.

That's not what I said.

Dumbass.

You're all wrong

Seriously though, the "color map" or diffuse map, as is the preferred term, is still very comparable to what we've been calling a texture. The shading has largely been moved to the normal map, sure (with the exception of whatever ambient occlusion you decide is necessary) but it's still a texture. It still takes a lot of work.

Oxy - You're right that it's a mistake to think you can get away with not doing a texture, but I'm not sure where you're coming from pitting "texture detail" against "poly detail"--poly limits are irrelevant because the high detail mesh we're talking about is getting baked into a normal map so it won't be polygons anymore.

Ace_1 - If you're trying to do a current-gen model there's no way you're going to be able to avoid doing a diffuse map for it. Normal maps are cool but if you rely on them completely you're going to end up with something that looks like a plastic doll. Also, if you're using the high-poly mesh to bake a normal map, you shouldn't have to UV map it at all. This is a good tutorial: http://www.poopinmymouth.com/tutorial/normal_workflow.htm

JM - You're on point with a lot of stuff here but using a plugin to auto-generate a low-poly mesh is a TERRIBLE idea. I don't care how good it is, it'll still throw stuff like edgeflow and economical polygon density out the window.

I never said I wasn't going to use color/diffuse maps. What I really meant was that I'm going to avoid spending a year hand painting a load of crap ontop of a wireframe in PS as much as possible.

I added another level of resolution for sculpting, then Blender crashed. Blender doesn't have any apparent way of splitting a multires'ed mesh into more layers, so I'm adding the details like the shirt seams the old way, it's way sharper at least. It looks like I'm going to have to work on the complexity of the wrinkles last. I also cleaned up a lot of bad edges, and started adding loops at the joints.

And I'm saying you're going to have to anyway, because that's as much a part of character modelling as it ever was--more so, even since the lack of shading on the texture maps is more than offset by the resolution & detail that's expected.

As a general rule, going higher-tech isn't going to decrease your workload.

I'm going to avoid the death spiraling of cost/software/time/work required to do anything with anything ever. This is still an excercise/experiment and hopefully i'll learn to make things see the light of day rather than dwelling on how hard it is to paint her arm hairs. Since I'm not terribly good at painting in 2d, I imagine how it can be accomplished in the 3d editor.

Anyway here's a small update with some details added, removed the sculpting

Crazybump.

He'll have to go back and tweak it, but rebuilding the whole mesh in lower poly from scratch is a waste of time.

Unless you use NURBS there's no way to reliably generate a low poly hull that retains enough silhouette detail.

Unless you start from a low poly mesh and create your high detail one from that.

You may think it's a waste of time, but it's how professionals and experienced character modellers in general do it. Auto-optimized meshes can in no way compare to a person who knows what he's doing.



The usual procedure is to create a basic low-poly shape mesh with very evenly-spaced polygons, and then throw that into a sculpting app and slowly subdivide and sculpt out your high-poly mesh. Once that's done, you do the actual ingame mesh.

The reason using an ingame mesh as a base for sculpting doesn't work is the polygon density is uneven in order to maximize efficiency and ensure proper deformation. The problem is when you subdivide something like that, the distorted polygon shapes carry over and it's much harder to sculpt.