Xavier will prepare patch for image compression, which will also be landed right after DPC++.
Other
Nikita brought up the topic of having an option for the software image sampling
Previously in the related patch it seemed we can simplify our lives by not worrying about this permutation at all. The main goal of that was to simplify workload as much as possible (and not so much the removal of lines of code).
Since there is a request for this, and if Nikita is fine maintaining the code path we can have it as ifdef and/or CMake option. It will be easier and practical than maintaining a separate patch/branch.
OIDN GPU doesn’t work on Blackwell
There is a hardcoded check in OIDN for the SM level of 99 which is below the Blackwell
We can prepare a patch against upstream and tackle it as library update
Patrick Will check whether it is as easy as a change in that specific condition
Nikita brought up the topic of libraries updates and dependency between updates and adding new features.
If library update happens too late there is no time to utilize it for the release cycle
There is nothing actionable to mention on this topic. We are looking into this, and will bring some structure and amek it more predictable for everyone.
Brecht, Sergey, and Weizhen had a meeting to map topics Cycles team would like to focus on for the next months. Unfortunately Lukas could not attend the meeting due to schedule conflict, so it is not the final version. The topics so far are:
Finishing volume octree + null scattering
Finishing Camera shaders
Adaptive subdivision fixes: bringing adaptive subdivision out of Experimental
This is a weekly video chat meeting for planning and discussion of Blender rendering development. Any contributor (developer, UI/UX designer, writer, …) working on rendering in Blender is welcome to join and add proposed items to the agenda.
For users and other interested parties, we ask to read the meeting notes instead so that the meeting can remain focused.
Currently on hold. Weizhen gave a good respone in Blender chat the other day on why it’s on hold.
It was more complex than anticipated
Some of the structures neccessary for a full ReSTIR implementation do not exist in Cycles. So along with adding ReSTIR, other structures had to be added.
The global illumination part of ReSTIR doesn’t appear to be ready for production renderers like Cycles.
New papers that bring noticable improvments to various aspects of ReSTIR keep releasing, so they wanted to wait a bit more so ReSTIR can be implemented into Cycles with many of these advancements.
Unfortunately, I don’t think so. Some of the work is already “in-flight” (there are WIP patches, etc), and other topics are much more complex than what is typically possible to tackle as part of GSoC.
Does It mean that Cycles Will only load in Memory the needed resolution for a specific texture based on angle/distance?
Does Tiled images mean UDIMs textures?
Does It work on big non Tiles textures as well?
Does It mean that Cycles Will only load in Memory the needed resolution for a specific texture based on angle/distance?
Technically it is based on the ray differential, so some rough bounce could make it so a lower resolution texture will be samples afterwards.
Does Tiled images mean UDIMs textures?
Does It work on big non Tiles textures as well?
No, tiled as in the pixels are stored as tiles (and not as scanlines). Some formats have native way of storing pixels like that. There would need to be some preprocessing for images to convert to such formats, similarly to maketx from OIIO.
Keep in mind, the above is more like ideas and not a concrete plan or design. We still need to work on that.
Thanks for the Clarification Sergey!
I Hope there Will be more news about It. In my opinion this Is a crucial feature for Cycles to be even more competitive with other engines!
Me too! =)
Would be nice to have a clear roadmap of the further steps of the Principled V2 BSDF project listed here
(thin sheet mode, nested dielectric…)
I can not give more details other than we plan to finish those. And it is not necessarily Blender 5.0 topic (as in, such changes do not really need to be aligned with the major release - major release is more about allowing to break some forward compat in order to remove some maintenance burden).
P.S. I am not sure it is the greatest format to be inquiring about all sort of features in the meeting notes thread.