How are bump maps created in the industry?

I am very confused about bump maps, I am trying to texture paint an object and I can somewhat understand drawing decent stuff on the Diffuse and Roughness maps, but trying to paint a bump map is terrible. For one thing the bumps generated look almost shiny, so the color information beneath it is washed out, another problem is if I try to draw a line where I want a scratch, the middle is pushed way down, in a way that doesn’t look good, no matter what brush falloff I use.

How are bump maps made in the CG industry?

Bump maps are just another way of manipulating relief info on the surface normals, similar to a normal map but using only 2 dimensional height information rather than 3 dimensional data. Bump maps are calculated with grayscale or black and white data just like displacement maps. White areas would indicate the highest points of a bump map, while black indicates flat areas. Shades of gray indicate values in between. Usually, bump maps are used only as a compliment to normal maps or displacement maps. High frequency details like skin wrinkles or pores where displacement mapping is not efficient would be a good use of a bump map.

Some displacement maps could be substituted as bump maps. You could convert normal maps to height maps online and use the height map as a bump map.

Some things to note here:

  1. Bump maps are mostly “baked” out of real geometry (objects or sculpts) where the result is very accurate, or perhaps sometimes there could be a “bump map generator software” (such as CrazyBump) that would be able to approximate the result as good as possible can, though not totally accurate but quite good at some times. However it is impossible or extremely difficult to paint the bumps in a custom way and get perfect results. In order to get an idea about how bump maps are created see this video as an example where you can get any model you want in 3D and bake it to 2D which is almost the same as “drawing” it but you draw in 3D instead. Perhaps you can edit the result after to add some fine details but at least you would have the 99% of the surfaces pre calculated for you so you won’t have to think yourself how a sphere or how a cone would translate from 2D to 3D.

  2. Also is quite good idea to separate your bumps into two categories. Those bumps with fat shapes would control the displacement better so you can decide how far your depths would be. Such as for example how deep you want creases with mortar in the brick wall to go. The other bump maps would be very high detailed. Such as the brick wall having small details like cracks, scratches, and chips missing.

  3. Another idea is that the bump map should control the shininess also so you might have to use a bump as well for this (either one or the other or a combination between these two). It is common that when you have scratches etc that light can’t get through or reflect anything. However the shaders produce the bump effect but also make everything look like is rainy wet. So you truly need to control the shininess that way with a bump.