How does the renderer handle it when an inward bump causes an otherwise occluded object to become visible? Like, imagine there was a red ball behind the train. When rendering a pixel near the edge of the train at an indented location, won't the renderer's ray hit the train before the ball? Does the renderer then check the bump map, see that the train is indented in that location, and then render the ball behind it instead?
If so, I wonder if many bump-mapped objects next to each other in sequence could cause greatly increased rendering lag.
buggy213
For bump mapping, there is no actual deformation to the geometry occurring, so this situation isn't really possible. Bump mapping is just fudging the normals at each point to make it look like there is extra geometric detail, and the effect doesn't work as well at grazing angles for this reason. True displacement mapping on the other hand would avoid this issue entirely since new geometry with the indent would be created, meaning that the renderer would not even register an intersection with the displaced geometry
koizura
I'm a little curious about the differences in use cases for displace mapping, bump mapping, and normal mapping. It wouldn't make sense to use all 3 at once.
After a little bit of googling, i see: 1) Bump maps a more outdated version of normal maps and simulate height without adding geometry (and hence impacting normals) 2) Normal maps simulate different light interactions without adding geometry 3) displacement mapping moves geometry, which means it requires extra geometry to work in the first place.
How does the renderer handle it when an inward bump causes an otherwise occluded object to become visible? Like, imagine there was a red ball behind the train. When rendering a pixel near the edge of the train at an indented location, won't the renderer's ray hit the train before the ball? Does the renderer then check the bump map, see that the train is indented in that location, and then render the ball behind it instead? If so, I wonder if many bump-mapped objects next to each other in sequence could cause greatly increased rendering lag.
For bump mapping, there is no actual deformation to the geometry occurring, so this situation isn't really possible. Bump mapping is just fudging the normals at each point to make it look like there is extra geometric detail, and the effect doesn't work as well at grazing angles for this reason. True displacement mapping on the other hand would avoid this issue entirely since new geometry with the indent would be created, meaning that the renderer would not even register an intersection with the displaced geometry
I'm a little curious about the differences in use cases for displace mapping, bump mapping, and normal mapping. It wouldn't make sense to use all 3 at once. After a little bit of googling, i see: 1) Bump maps a more outdated version of normal maps and simulate height without adding geometry (and hence impacting normals) 2) Normal maps simulate different light interactions without adding geometry 3) displacement mapping moves geometry, which means it requires extra geometry to work in the first place.