ZBrushCentral

Help! 3d mesh and Hi res textures?

I have read on more than one instance that if you are going to use high res textures such as 4k or higher, that somehow the base mesh comes into play.

Is this true?

What I have not read, anywhere, is “why?”

I’m not even certain I use the term “textures” correctly. Here’s what I’m going for:

Facegen + real life photos of people to create a base mesh.
NON ANIMATION!

Bring that obj into zbrush, sub divide a few times to get a really high res mesh, and then start the process of making a skin for the mesh.

I have read that Facegen doesn’t create a good enough model for the use of very high res “textures, etc…etc…” and I’m just wondering if this is true and why?

If I’m not going to be animating, I’m not sure I understand where the problem/s come into play.

Yes, I realize that Facegen may not create an ideal mesh with great topology and everything, but it does create a “close enough for me” mesh of a character I’m trying to replicate in 3d.

Does the problem lie in the UV unwrap? I’m just confused. Some guidance would be great. I want the final skin to have bump maps, SSS, etc…
and I don’t understand the importance of the 3d mesh in this process.

Thank you.

Hello concretefire.

Yes you do need a higher res mesh in order to create a high quality texture from photos. The reason is because when you paint in 3D, Zbrush uses a method called poly painting. What this does is paints on the surface of the objects polygons instead of directly onto a texture map.

Because a 4k map would cover alot of pixel area, you will need enougth polygons to capture this detail. I have found that for the best results I have to use Photoshop for texturing with photo refferences as a whole character will need a very high poly count. That said, these days I split my characters into several Uv zones with a smaller map applied, a 2k one for each area. If your just working on a head then sub deviding it enought shouldnt be a problem providing your compuer has enought ram and processing power.

When animating your use the lower poly version of the model, we only need high res meshes to make the texture maps, and displacment and normal maps.

This is really only an issue if you’re using poly painting. If you’re using standard UV based image projection mapping, the only thing the underlying mesh affects is the quality of the silhouette (whether it looks blocky or artifacted at all), as usual, and has no bearing on the texture itself.

I don’t know anything about FaceGen, but the FG to ZB to3dsMax demo vid I just watched didn’t have poly painting involved at all, so I don’t know what the issue would be.

If you’re using ZB’s UV-independent poly painting ability, yes, youre going to want a very high resolution mesh for high rez textures. Outside of Zbrush, this is not an issue.

ARRRRGGGGGGG.
This 3d stuff is really confusing me. Lol. What’s the line in that movie? “I find it disgusting, yet oddly fascinating”

Well I have learned some things along the way.

  1. I have no desire to “sculpt” or even model using polys or any other method.

So that’s good. I suppose. At least I know I don’t want to learn modeling, and sculpting. I only have ONE interest in 3d at all. Humans. I have a program that will create a 3d face model for me. Facegen.

But here is where the confusion and frustration starts for me. Let’s use someone famous as an example. Pick anyone, Bill Clinton, Obama, President Bush, Clint Eastwood. Anyone ok?

If the 3d “model” is not 100% right on the money, I don’t care about that.
What I DO CARE about is the render. I want to be able to zoom in on their cheeks and look extremely closely, and see the pores in the skin. See maybe even a vein or two. Perhaps even small hairs. It doesn’t just have to be the cheeks. It could be anything you want. Eyes, nose, lips, whatever.

But zooming in (within the confines of a scene) on one particular aspect, nose, eyes, ears, cheeks (pick one) and I want the “skin” to look as incredibly realistic as possible. That’s what I want. Plain and simple.

Now, zoom back out, and look at the overall model. If it doesn’t look exactly like Bill Clinton, Obama, Eastwood, or Bush, I can live with that.
I can live with that because I’m not into the modeling aspect. I don’t have to have every curve of the cheekbone just right. Etc, etc.

I just want the skin to look as ultra realistic as possible.

The confusion and frustration I’m having is that I’m getting two different stories. On the one hand I hear that my base mesh/model must be perfect, all quads, etc. etc. for this to be accomplished.

Then I read somewhere else where It doesn’t have to be perfect, only extremely high resolution/many sub divisions.

So , which is it?
Doing an exact character model of Bill Clinton would indeed be cool, however, I am much less concerned about “nailing” a replica of him than I am about how the overall skin looks once rendered.

If someone said “Man, that’s the worst Bill Clinton Character I have ever seen, but the skin is incredible” I’d be the happiest man on the planet.

I hope I’m making sense. Please advise further. Thank you all so much. You really are helping. It’s starting to make some sense to me.

Ok, I have re-read both responses from my original post and I think I’m seeing the picture better maybe. Let me see if I’m on the right track:

Facegen, all by itself, does not create a model that has enough polys to support ultra high res “Skin”. < This much I do understand.

However, that is where zbrush comes in very handy. Pop in the Facegen .obj and subdivide until my hearts content, or rather, divide it more than enough to acheive what I’m going for > ultra realistic skin renders , even if the overall model doesn’t look exactly like the person I’m trying to recreate.

Am I on the right track here guys?

I understand what you want. You want ultra realsitic renders and realistic, high definition skin tones. You dont want to learn anything else except what will help you achieve that.

The problem is, this is an art form, that requires you to develop a lot of ancillary knowledge and expertise with various tools, to get any single thing done to that degree. Tools like face gen are great, but there is still no "Magic Button". You are going to have to put some effort into learning a number of things, and there is always more than one way to do any one thing. I can't (and wouldn't if I could) spoonfeed you all you need to know here, but I will give you some basic factoids, that might help you steer your learning. 1) The textures (mapped images wrapped around a mesh) in most 3d applications are image based, and the quality of that texture (how crisp it looks, how large of an image you can render with quality, and how far you can zoom in without getting blurry) will depend the *image resolution*...essentially how large that image is in terms of pixels. The realism can also be enhanced with any number of render level shaders, like bump mapping, and enhance details in the image textures themselves, to look more realistic. Even with high resolution textures, it still takes quite a bit of experience to know how to manipulate the render settings of any given software package to produce realistic renders. In this case, the primary effect the underlying geometry has is to block out form, catch the light , and cast shadows. The more realistic and the more polygons the mesh contains, the better the *form* will look, the more realistic shadows it will cast, etc. A particularly low poly mesh (like that of a game model) might have a "blocky" looking silhouette. But even so, you shouldnt have to increase polycounts to an extraordinary degree to come up with good results using this method, and regardless, it has no bearing on the quality of the image map itself. 2) Zbrush, and sculpting programs like it allow people to increase the polygons in the mesh to levels so high, you can actually "sculpt" high frequency details into the skin with real geometry, like pores and blemishes, that have traditionally been done with render shaders. Now, Zbrush can work with image textures in the traditional manner, but what it can *also* do, is embed color information directly into the polygons of a mesh. It is called Poly Painting. The big benefit to this, is it lets you create or exchange multiple textures, and to paint directly on the model with the ZB's painting abilities , *Independent* of any UV unwrapping. When you make use of this feature, *mesh resolution* (how many polys it has) does become an issue, because essentially, each polygon becomes like a pixel, and in order to hold more detail, the mesh must have a lot of polygons, usually into the millions, for the best results. Beyond this though, I dont know that you'd even need to make use of poly painting. It doesn't *seem* necessary for your workflow. It seems as if Facegen creates a model with a face texture already UV mapped to it, ready for rendering in another application. The results you'd get rendering that that will vary greatly with how skilled you were setting up renders with that program, and how skilled you were with tools like Zbrush that might help you to sculpt more realistic detail into your meshes. If you are rendering in an application outside of Zbrush and the face texture looks blurry when zoomed in, you need a larger, higher quality *image* texture. Anything beyond that, I would need to know more about your workflow, and what you are trying to do. I suspect, it is simply a matter of you getting some experience with the things you are trying to do though.

Thank you for your response. Believe me, I know this is an artform. I promise you I’m not looking for a magic button. I have put forth a lot of effort over the past months and that has opened my eyes to many things in this venture. And precisely like you said, there are many ways to do essentially the same things, which leads someone like me into mass confusion. Lol.
I am more than willing and eager to learn. The problem is, I don’t know where to start.

Maybe this could help you, help me. >

Forget about human faces. It’s got curves, etc.etc.

Think Brick. one brick. I can model a brick easily. Now I have a virtual brick.
I want to make the brick look like a brick. The pivots, dimples, etc. The whole nine yards. And this is where I get totally lost.

I’ve got my form. (the brick) Now what do I need to learn in order to “skin” it?

Do I need to learn about UVmaping? Polypainting? Both?
And I do understand that even if I learn all about both, I could still get horrible results simply based on my rendering knowledge, and settings.
(An art all by itself)

So believe me when I say I’m not looking for a magic button. I just need to know which road to go down in order to skin my brick, or form.
And I’m being totally serious. I am going to forget about human faces myself for a while and concentrate on one single brick form.

I want to learn how to skin my brick, and make it look like a brick.
Just point me in the right direction, and I’ll jump into it. Once I have learned all about that, I should be able to apply most, if not all, of that knowledge to a human face.

Again, thank you for your time. I don’t want to be spoon fed. Lol. I just want to know “what” to learn? Uvmapping? Poly painting? Something I’ve never even head of? If you could point in me in the right direction I feel like I’d be taking a huge step forward. Thank you.

Even this inquiry is very broad, I’m sorry. What you really need to do is read all the documentation on texturing, lighting, and rendering in whatever application you are going to be using to render, do any available tutorials, and practice. These things will become clear, but you are going to have to learn them for yourself.

When you have specific, focused questions about Zbrush though, come back, and Ill try to help you out.

I think everyone has prety much said it all, but let me try to sum it up as simple as I can based on a single brick as you mentioned.

You had the basic form, its looks like nothink at this stage, so you need to add several eliments to get it looking real good for close up shots.

First thing is you do have to have it UV unwraped. There are two methods that one would use. The first is unwrap manualy. You choose where you wish the semas are to split the brick into a 2d, flat packed map. This method is best for if texturing in a 2D application like Photoshop. This kind of Uv unwraping looks clean in the way its laid out, and enables you a visual aid to what part of the map your painting on in 2D. Most use this method regardless, as it opends up the model to be textured by anyone with any painting application.

The next is a Auto unwraping method. Zbrush has these, often AUV, and GUV tiles are used. This makes better use of the full UV space, but can not be painted on in any 2D application, as the Uvs apear somwhat scattered, but usable by Zbrush via polypainting. The other advanrtage of using this method is that a single 4k map will hold much better detail as the Uvs are distributed better than most peoples manual uv mapping.

Once you have this done, your next need to add a colour map. Reason why your do this next is because you will want the bumps and details to match the colour texture details. For you, using a real picture of a high res brick would be the fastest way. You can use the brick picture for both the colour infomation, and for the displacment infomation.

First your either use a 2d program like photoshop or use Zbrush Zproject. For pure Photoshop you have make a UV template, a wire version of your Uvs. Using UV mapper classic can do this with ease. To keep it more simple you can use a plugin called Zapplink which allows Zbrush to work with Photoshop.

To keep things even more simple, your just use Zproject in Zbrush. When producing the zproject cast of colour infomation, you will need the mesh to be high resoloution to hold the high detail texture info, just as mentioned above.

You may want to retail the initial shape, so your store a morph target before doing any sub deviding of your mesh. Also you may want to keep the mesh from smoothing duing sub deviding, so you turn smooth off.

Now you can start to use Zproject to get that high res colour onto your model. Next, your need to get some high detail into the brick. Again there are many methods. First way would be to convert your texture map into a alpha, then displace with that alpha, all done in Zbrush.

When done, your have to go back to sub d level 1, switch the morph to get back the original brick shape, as all this sculpting on higher levels will alter the first level. When thats done your generate a 16 bit displacment map.

Once that is done, export your brick as a OBJ, if you havent already done so, and save out that displacment map which has to first be sellected from the alpha pallet, and fliped on the vertical via the alpha menu.

Delete and store your morph target again, should you wish to do another such map later, then go right back up to the highest level again. Make shure the RGB at the top of the interfacc is set to 100%, make a blank texture 4096 x 4096 from the texture menu, then in the texture menu to the right of the UI click colour to texture button. I dont have access to Zbrush for the exact name, its abreviated. Then export your texture after flipping it on the vertical.

Now you have a brick obj, colour, and displacment map. Take these 3 into your rendering application, set the lighting up and scene, add a shader to that brick so you can setup things like how the light effects the surface like diffuse amount. Set up things like the specularity level, for a brick quite simple.

Apply the colour and displacment map, render your scene. This is rather simplified.

Now do the same for the human head, but with these steps.

1: Store morph target.
2: Sub devide.
3: sculpt the head with basic deformations
4: make a new layer and add high end sculpting details, either by hand or wait and use the colour texture map as a alpha.
5: hand paint via polypainting, or combo of zproject.
6: complete step 4 if havent done so.
7: go down two sub d levels from the highest and generate a displacment map. This will be used as a bump map. Or make a normal map which il cover in a min. Normal map not done at this stage.
8: turn off the high res detail layer sculpt, go back to sub d level 1, switch morph target, and make a displacment map. save your project as a Ztl, then delete all lower levels so you only have a few higher ones left.

At highest sub d level turn on your high res layer. Go to sub d 1 and use Zmapper to make a normal map, a whole chapter can be made on this alone.
If your mesh is said to be too high res by Zmapper, use reconstruct mesh in the geometry tab to the right of the UI, this will make another lower level.

Now make shure both the normal and displacment maps are flipped on the vertical, there is a option in zmapper to flip the normal map when generated.

Now export all map, you havent done for so far.

In your render applciation the fun starts. You must make a skin shader with the right settings for how the light penitrates the skin, use Subsurface scattering for this. Your have to get this just right. Normaly a map gets made to control this. This would be a grey scale map. Darker shades alow less SSS, while light shades alow more. So your paint a map in Zbrush based of this. Lips, ears will have more lighter values, while a darker shade fo grey for the rest. Also maps can be used to control the specular, reflection, difuse.

Now you gota spend the time getting the perfect ballance between the correct lighting, SSS, the whole shader with every component doing its job at the right ballance, this is a art in itself.

Then if you have the skill, your have a realistic looking skin.

This is a quick run through, its not a quick process by any shot. The more time you invest, the better it will look. Lighting and bump detail is essencial to realistic skin, without these right, no matter of SSS and skin shaders will look great.

:+1: :+1: :+1: :grimacing: