Normals map newbs question
Images are sometimes not shown due to bandwidth/network limitations. Refreshing the page usually helps.
You are currently reading a thread in /3/ - 3DCG
You are currently reading a thread in /3/ - 3DCG
Hello everyone, I'd like to ask you something about normal maps, how do you get right of hard edges sticking out of smoothed normals?
Im not talking about the 90° uv split or smoothing groups, I already know that, but still, if I have for example a box like pic related, and the high poly has smooth edges, those sharp ones in low poly will always stick out, fucking up my normals, how do I avoid that? It happens to me with every smoothed edges normals, any tips?
The mesh is a 90 degree corner. Normal maps only affect lighting. The only thing you can do is bevel the edge.
>>512784
I should also set a unique smoothing group for the bevel right? In order to have it smoother enough
>>512783
I don't get fucked up normals? Perhaps your low poly edges stick to much, and the mapping from low to high poly misses.
Maybe you aren't added some space between the shapes in the UV map. I mean, if I zoom in a lot, i can see some little edge there, and if this is a concern, then just accept the poly cost of the bevel.
>>512794
Could you please show me the beveled low poly?
>>512794
I didn't bevel it. It's just a box (6 quads).
I only ever so slightly increased the size of the high-poly version, so that the edges didn't protude, but otherwise, it's as simple as it looks. If i zoom in a lot, there are some wonky areas in the corner.
You must make sure that you have extra space whereever there is a seam.
>>512822
So incresing the hi poly size when baking doesn't fuck it up? And can fix this? I usually leave those as they are
I always heard that hi poly and low poly have to be the same size
>>512823
I don't increase it a lot, i increase it so that it's closer to the average size of the
For every point on the low res model, it needs to matching point on the high res model. So the low poly surface should, as closely as possible, match the high poly surface.
90 degree (or higher) are particularly problematic, so it's a bit of a special case i guess.
Anyway, here is sort of how i had the models in the examples i have shown.
>>512834
Just to be clear on this; it's not perfect. You will be able to see some slight normal error here, but I doubt you'll see with diffuse texture on and other noisy stuff.
>>512834
and you could also just leave smooth normals on the low poly model and bake using that. The smooth normals will map to the high-poly more accurately. The final normal map is relative to whatever normal the object would have, so it looks the same if you're consistent.
This lets you avoid UV seam artefacts.
>>512838
WIlling to provide an example? For the sake of learning :)
>>512834
Did you bake with a cage object?
Always, always bake with a cage. WIthout one, you'll get artifacts wherever there's a UV seam.
>>512905
I didn't. I roughly knew about the technique, but I didn't know it existed in blender (had to switch to the cycles render for that). I tried it out now. Turns out very nice thanks.
