Looking for some help form 'realtime' guys

[gordonrobb]

Hi, I'm looking to model the Power Loader form Aliens. I have made the decision that I'm going to predominantly use DDO to texture it, and therefore, have decided on a perhaps convoluted modellnig route. I am doing the modelling in ZBRush, just using dynamesh and booleans etc to get a high poly, high detail part. I then want to create a low poly (i am most comfortable doing this in Topogun). But I'm kind of stuck when trying out the normal map creation.

I want to be clear, I'm not trying to create a low poly, game res type mesh. Just lo enough to animate, and perhaps render in Unity (mainly to play). The main reason is to have a low-ish and a normal, to get the best out of DDO.

My questions are....

1. If I have rounded edge, should my low poly have a bevel or perhaps be on the edge.

2. When I create my low poly and export it for use in xNormal, should my low poly have smoothing on?

3. When I render, should my low poly have smoothing on?

4. When I have a feature that is too big to just be on the normal, should I do topology that is connected to, or separate from the main body of the piece?

Basically any advice any experts in this field can offer, will be really helpful.

One example of a part is this...

[gordonrobb]

Got it sussed....

OK, after hours of messing with different versions, I think I've settled on a process....

1. Model in Dynamesh (creativity unleashed)
2. Basic mesh in Topogun
3. Tweak mesh in Modeller to make it SubD-able
4. UV
5. Create Normal and colour ED
6. Fire it through DDO
7. Render in Octane..

This is that, textures took about 30 seconds to tweak.

[prodviz]

Looks cool.

Interesting workflow.

[gordonrobb]

It's very convoluted I admit. The main thing is how much easier I find to make complex mechanical shapes if I do them in Dynamesh. You're just left with completely useless Topology.

Shapes like this are a breeze....

[prehabitat]

My questions are....

1. If I have rounded edge, should my low poly have a bevel or perhaps be on the edge.

2. When I create my low poly and export it for use in xNormal, should my low poly have smoothing on?

3. When I render, should my low poly have smoothing on?

4. When I have a feature that is too big to just be on the normal, should I do topology that is connected to, or separate from the main body of the piece?

Hi Gordonrobb,

I'm trying to get my head around the whole thing too... as primarily a Revit user who uses software to sell and subsequently build physical things I have close to ZERO knowledge of UVs, so this whole thing has been rather steep..
I've got an index of my paraphrased defenitions and instructions for myself (my process) and wondered if you answered your questions above... just trying to fill in all the blanks in my knowledge of this area...

I assume the answers are:
1. low poly hard edge, with an 'tangent space normal' bevel?
2. smoothing will increase the tris right? or it is just a tag that is picked up by the host software and changes the way the mesh is displayed in the viewport?
3. would depend on the form right? if it has curvature smoothing helps, whereas if it was largely a form you could create on a conventional CNC machine (all flat faces and single-curve radiuses) the smoothing would achieve very little?
4. This would be based on materiality only right? in some examples they separate the geometry to prevent edge bleed/geometry issues during the baking... does this apply here?

Once again, so far out of my main area of expertise its not funny, actually came across this post while looking for info on high/low poly meshes and normals

[gordonrobb]

I've learned a ton since posting this thread, and a lot of my questions came from not really understanding the process of normal map creation But my answers the questions now are...

1. Yes/no/maybe You can have a sharp edge in your lower poly (when the high poly is rounded), but the edge really has to extend beyond the rounded edge. Also, you are better having the UV split at the edge. Also, baking is tricky as it can end up with a line on the edge. So each model is almost a separate case, and it depends on how extreme the bevel/curve is.

2. Absolutely yes. When I say smoothing here, I mean material smoothing, not something like what other apps call smoothing groups (actually smoothing out the geometry). What is important is setting the smoothing to the right smoothing angle. 89 degrees often causes artifacts.

3. I'm still not clear on this, but I think yes is the answer.

4. This really doesn't matter. Yes or no is fine. What I would do though if I made it separate is make the geometry go through each other so there are no potential gaps.

Hope any of this helps.

[ristoraven]

I've been scratching my head also on this lately. I haven't been in low poly biz before. I have thus far just modeled an object and simply used that as it is..

Anyways, this is what I have figured.

1. I create high poly model
2. high poly model on one layer, I create a low poly model. Then I heard of a program called Meshlab and it does that job with one click.
3. I copy the low poly model in order to create a "cage" to be used in xnormal. It should be scaled so that it is slightly bigger than the real low poly model
4. I create UVs
5. I create color id map
6. in xnormal I import high poly and low poly and at the low poly section of xnormal I set up the cage thingy
7. I export tangent space normal map and curvature map
8. Quixel DDO and Ndo
9. export maps to UE4 or Octane

Not the easiest thing to do game assets..

[grimm]

I'm not one of the "real-time guys" but I have been working with a similar workflow, although I use 3D-Coat for most of it.

1) Create base mesh in Blender or 3D-Coat's voxel mode.
2) Sculpt mesh and add in details with 3D-Coat's sculpt (poly) mode.
3) Auto retopologize the mesh using 3D-Coat's Autopo function. This works really well for the more simple models, although you might have to do a manual retopo on more complex models.
4) Add UV seams and unwrap the low-poly retopologized model. 3D-Coat has a bunch of automatic functions for this and makes it really easy. I usually break up larger islands so they fit better on the texture image.
5) Paint the model with 3D-Coat's PBR smart materials.
6) Export back to Blender so I can render with the Octane plugin.

3D-Coat also has a app-link function that makes it easy to move the models and textures from it to other packages and back, like Blender. 3D-Coat also lets you bake the normal map and/or displacement map from the high poly model, and the curvature/AO maps as well. You can then use the AO and curvature maps to control the smart materials.

Jason

[gordonrobb]

Ristoraven: it kind of depends how you've made the model. A lot of mine are sculpted in ZBrush using dynamesh, so I have to create the LP using retopologising, in some programme. Also, if you are going to model something in SubD, using edge loops etc, you can do that. Then copy the mesh removing all the edge loops and that can be your low poly . One thing I learned that was really useful, is you can create a cage from the low poly inside XNormal itself. Also, Quixel suite is excellent at creating curvature maps. It can use Normal only or geometry only (if you inky have a high poly) or a combination of geo and normal.

Grimm. Yeh, that's more or less my process normally. I been trying to learn to go from mynHight Poly ZBrush monsters, to something a but more manageable for animation etc.