Hi!
Currently, the Rasterize Geometry output suffers from aliasing. Does anyone know of a way to reduce it? I've tried looking for an OpenFX plug-in for this, with no luck. The closest I've found to a workaround is Natron's openfx-misc plug-in [github.com] collection, which features a Shadertoy implementation and would allow me to run existing FXAA implementations for it. However, the build of it that is shipped with Natron is a bit old and does not work with Houdini. It is possible that a more recent one might work, but there aren't binaries available.
Another solution would be to write an OpenCL FXAA (or any other anti-aliasing algorithm) implementation... has anyone found a solution to this issue?
Thanks!
Rasterize Geometry antialiasing (or the lack of it)
1000 7 1- matteoscopel
- Member
- 3 posts
- Joined: Sept. 2021
- Offline
- PHENOMDESIGN
- Member
- 172 posts
- Joined: May 2021
- Offline
COPS has the tools you need to implement an advanced AA filter. (ie. ID to SDF, Eikeonal, Derivative, Dot Product, etc.) and maybe even the Denoise AI would work for a naive approach. I am on Apple Silicone and does not support Intel Denoiser that that is based on so I cannot speak to that.
You can also go the old-school route and super-sample in SOPS after:
https://medium.com/@lennnart/vex-ray-tracer-3-supersampling-anti-aliasing-62ea7e31d6aa
Quick question, is this a static rasterization you are optimizing or are you rasterizing the geo across time with dynamic data?
You can also go the old-school route and super-sample in SOPS after:
https://medium.com/@lennnart/vex-ray-tracer-3-supersampling-anti-aliasing-62ea7e31d6aa
Quick question, is this a static rasterization you are optimizing or are you rasterizing the geo across time with dynamic data?
Edited by PHENOMDESIGN - July 16, 2024 12:38:27
PHENOM(enological) DESIGN;
Experimental phenomenology (study of experience) is a category of philosophy evidencing intentional variations of subjective human experiencing where both the independent and dependent variable are phenomenological. Lundh 2020
Experimental phenomenology (study of experience) is a category of philosophy evidencing intentional variations of subjective human experiencing where both the independent and dependent variable are phenomenological. Lundh 2020
- matteoscopel
- Member
- 3 posts
- Joined: Sept. 2021
- Offline
PHENOMDESIGN
COPS has the tools you need to implement an advanced AA filter. (ie. ID to SDF, Eikeonal, Derivative, Dot Product, etc.) and maybe even the Denoise AI would work for a naive approach. I am on Apple Silicone and does not support Intel Denoiser that that is based on so I cannot speak to that.
You can also go the old-school route and super-sample in SOPS after:
https://medium.com/@lennnart/vex-ray-tracer-3-supersampling-anti-aliasing-62ea7e31d6aa
Quick question, is this a static rasterization you are optimizing or are you rasterizing the geo across time with dynamic data?
Thank you, I will proceed that way and put together a solution.
Answering your question, it is a just a static rasterization for texturing purposes.
- tamte
- Member
- 8786 posts
- Joined: July 2007
- Offline
- matteoscopel
- Member
- 3 posts
- Joined: Sept. 2021
- Offline
- ronald_a
- Member
- 92 posts
- Joined: Aug. 2017
- Offline
- Gagui
- Member
- 55 posts
- Joined: May 2017
- Offline
- terry_williams
- Member
- 53 posts
- Joined: Sept. 2018
- Offline
-
- Quick Links