HBAO pass is not the "real" HBAO, but N8AO

[Rabbid76]

Description

HBAO pass is not a "real" HBAO, but N8AO

The HBAOPass and HABOShader implement an algorithm presented here: N8AO
This algorithm is a kind of mixture of SSAO and HBAO, but not the original HBAO algorithm.
Actually, this is not a bug or judges the quality of the algorithm, it is just a naming issue.

Solution

Since there are various ambient occlusion algorithms around, it would be nice to have a pass where the AO algorithm can be simply selected, this pass could completely replace the HBAO pass.
All the AO algorithms operate on the same source data and produce a result in the same format. The algorithms differ only in some details of the internal formulas and the sampling of the hemisphere. It should be possible to activate the different algorithms simply with preprocessor definitions in the shader.
The use of the names AOPass and AOShader would also not indicate a particular algorithm like the name HBAOPass.

Alternatives

Rename HBAO to N8AO or something else.

Additional context

Since I implemented the HBAO pass, I feel responsible for this. The problem arose from a lack of knowledge. I was familiar with the math behind various AO implementations, but I paid less attention to the names than to the algorithms themselves. I found my mistake by reading some AO papers in the last few days and I feel guilty. To compensate for this, I would like to provide a solution.

[Mugen87]

Sorry, but like described in #27296 (comment) I vote to remove HBAOPass.

The repository you have linked does not only represent an AO algorithm, it's a full pass implementation for EffectComposer. What #27114 did was to replicate this implementation into something redundant. We are now effectively maintaining the same thing two times in the community which is bad.

If you want to contribute improvements to N8AO, it's best to do this in @N8python's repository.

[mrdoob]

Trying to understand what happened here...

1. @Rabbid76 submits Examples: webgl_postprocessing_hbao #27114
2. r159 gets released
3. Someone asks about the differences between HBAO and N8AO in the forum.
4. 20 minutes later @Rabbid76 opens HBAO pass is not the "real" HBAO, but N8AO #27295
5. At the same time @Rabbid76 submits Examples: webgl_postprocessing_ao (SSAO, SAO, N8AO, HBAO, GTAO) #27296

Is that the right order of events?

[Mugen87]

Yes, I believe this is correct.

[Rabbid76]

@mrdoob I just wanted to clarify that I didn't know about 3 (Someone asks ...), when I did 4 and 5. My intention was to implement a better AO pass, and as written in point 4 ("... this pass could completely replace the HBAO pass. ...") the AO pass was intended to replace HBAO, I just didn't want to delete something without feedback.