New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Discuss: Remove all Bump Mapping and Parallax Occlusion code #10479
Comments
I don't use either of them. Looks ugly IMO |
TL;DR: RBA shaders is half-assed work and must be redone from scratch (example: tonemapping is a postprocessing thing normally) |
@kilbith I agree with the state of the shaders. I'm trying to undo at least the parts that do not work, so we have less to redo. Re: RBA (RIP)... I will say that he just didn't get to finish them. Someone (forgot who) showed me some video of his uncommitted work a while ago and that looked really impressive. Related: Wasn't there some postprocessing PR? What happened to that one. |
👍 Bumpmapping looks so bad i think it is better to have none until it is done well than leave it in months or years waiting to be fixed and giving users a bad impression. Parallax occlusion seemed to work and looked more acceptable to me. But i do not know the code as well as you do. The code may be closely linked to the bumpmapping code. It might help to make a fresh start with that too. So, i would support removing that too if you judge that to be best. |
Occlusion not exclusion. :) Ok... I'll file a PR to remove all the code, including the UI and the various options passed to the shaders. |
@HybridDog Any opinion? (Since I missed #9241) |
I'm against the removal of Parallax Occlusion. In my opinion it is an interesting graphics feature and almost works correctly.
I've recently added this documentation: #10096 |
I was under the impression that parallax occlusion does not work without bump mapping. Can parallax occlusion work without correct normals? Edit: I got confused about normals. Parallax Occlusion can work without bump mapping. I am still in favor removing both and starting again |
Me too. I did some testing of parallax occlusion. When more than 1 node away the node top face seems to appear ok, but note that the other node faces are inconsistent and much flatter, and partial pixels from the opposite edge are present: Looking at the node top face, as your view angle decreases the height of the raised pixels decreases relative to their widths: So the appearence is messy and has weird behaviour. Up close the raised pixels 'lean' in odd ways, although this is much more subtle at typical distances, so not much of an issue. It is also not surprising since this is 'faked' 3D structure: |
I think this is supposed to happen unless the mod explicitly disables it for the tiles with
This may be intentional, too. When the height decreases with the angle, the pixels from the opposite edges are less visible. |
Yes. The wrapping pixels are needed for an area of nodes, but look wrong for one node. The 'tileable' flags force one behaviour or another, regardless of whether a node is alone or in an area of nodes, so they are not a complete solution.
I doubt it, sloppiness is far more likely. |
#10487 is merged. |
@HybridDog Happy to work together to add graphics features back in. |
As I look through the code - prompted by the inclusion of the bump-mapping test nodes into devtest - I'm coming more and more to the conclusion that the code should just be removed until we find someone who knows (and has time) to implement this for real.
Issues I've seen:
Then there are other issues such as hardcoded lighting of the faces in the engine, which is at odds with lighting in the shader, since now you'd have to adjust for that.
I played around with this for a few hours, but it looks like there are other problems. Normals missing (or wrong), etc.
In the end, as much as it pains me (I'm a fan of better graphics), I think we might want to simply remove all bump mapping and related code, keep it simple. That way we also simplify the shader code and maybe provide a cleaner slate for trying again.
Thoughts?
The text was updated successfully, but these errors were encountered: