Hello everyone! So I have a cliff setting, which is basically a mesh with a multi resolution modifier. My computer is being painfully slow by just even selecting the cliff, so I thought I should create a normal map instead. The only problem is I don’t know how to since the mesh isn’t a simple cube. Please let me know what you all think or if there may be an even better solution to this. Thanks!
I would use the base mesh from your first image and shrinkwrap it onto the displaced mesh on the bottom.
Then unwrap the base mesh and bake the normal map information from the displaced mesh.
To get a better result you might want to use a cage.
Actually it should suffice to bake the normals for the hires object onto itself. Afterwards disable multires modifier.
Yeah, just watch a few youtube tutorials on how to bake, in Blender, you have every-thing you need, you can always take the simple model, go into ’ Edit mode ’ and, snap the vertices to the detailed model, using snap to face, so you get a really nice low-detail mesh, that follows the detailed one closely - - then, I think you just select the detailed one, the shift click low-detail one and, hit bake, it’s that simple, just watch some tutorials, be aware that a detailed mesh like that will probably require a high-res normal map, at least 2048x2048 - - also, if you unwrap the low-detail manually or, at least add good ’ seams ’ to it ( in edge menu, in UV unwrapping mode ), then Smart UV Unwrap will give good results, just add good seams first, so it unwraps those, as well - -
you’re almost there, Keep goin - -
Yes, it shoudl be sufficient for the small details, but he wont get the large deformations .
@strapazie
@James_Unknown.64
@omgold
Thank you for your responses everyone! To tell the truth, I’m still a bit overwhelmed with this problem, even after looking at a few more tutorials. I guess part of it feels strange because it always seems that you can tell if an object is “flat”, even when it has normals.
It’s also strange because even with a high multi resolution, you can see the low poly-ness of the mesh from a closer distance. This confuses me when I watch tutorials like this one where they use fewer subdivisions and yet it looks cleaner…
Expectation 
Reality 
In general, I just suck at making rocks, but thank you for all of your support everyone! I will try to improve! 
Normal maps Don’t add details from very sharp angles, you need to make sure the over-all geometry is there and, use the Normal to add details, to think that don’t need to have a silhouette . . if your basic mesh doesn’t have enough faces, it will look flat - - Still, you can go from 100.000 to, may-be 5000 faces, that way - - Your base mesh needs more detail, That’s all - –
To get a result like that a normal map is not enough, you need real displacement, that you can get with a B/W height map linked to a Displacement node, or to a Displace modifier.
There is also the Vector Displacement node available which make use of a RGB map, however a normal map doesn’t work here, and a proper map must be generated on other programs.
The artifacts along the fine creases look indeed odd. Can I see the normal map and a render without the normal map? They must come from somewhere. There might have something gone wrong during baking.
The thing to think about is a detail map like a normal/bump/displacement will not affect geometry of the mesh and are intended for more of a distant shot. The closer you get to the object or change the rotation of the object you start to see that it is indeed flat and just smoke and mirrors. So the main thing to consider is how close is the object going to be in the render to the camera and will it be animated (moving closer and further away to the camera) Then make your decision on if you will need that geometry or if you can just get away with a normal/disp/bump map. This thread someone was asking about rocks and I show an example of the difference between a normal map render and one with geometry. What is the best way for modeling? Blender Guru had a tutorial as well of a way to do a cliff face and he goes over the sculpting side of it,
look, when you bake the map, you must input a distance, called ’ ray extension ’ or, something it determines how far the ’ base mesh ’ looks for detail, I think it might be that, I agree, it should definitely look better and, if some-one can do better but, with less detailed base mesh hmm . . if the scale of your object is very big, like 30 x 30 m, in Blender units, the Bake distance needs to be corrected, since it’s in meters - -?
I’m sorry! I forgot about this thread! I actually didn’t create any normal maps, I just used some sculpting brushes.
So at the fine creases, this means that the mesh itself has the ripples? Hmm, in that case finer subdivision or retopology would likely be needed. I’m not into sculpting, so cannot say whether dynotopo is a good idea.
Finer subdivisions has helped alot. I think I’ve finally gotten it!
