Page 9 of 17
Re: A baking test scene
Posted: Mon Dec 09, 2019 11:44 am
by Racleborg
Bump should be (and will be) baked in a separate map file. Does it happen without bump too ?
Without the bump node connected to the Velvet shader, then the baked texture image looks correct (in example given, nice bluish velvety colours etc)
With the bump node connected to the velvet shader, then the baked texture image looks almost black.
I don’t have this issue with bump connected to other shaders, such as the Disney shader.
Are you saying that bump shouldn’t be connected to shaders for baking?
If bump is added after the colour is baked, how can it be set up to effect a baked texture connected to an emission shader (in Cycles)?
(or have I miss-understood?

)
Thank you
edit; here's the file in case it can be of help:
https://1drv.ms/u/s!AhdwgLaw01tLpwJcl4e ... y?e=3docEG
Re: A baking test scene
Posted: Thu Dec 12, 2019 9:31 am
by Dade
I tried to get Intel Oidn or BCD denoiser to work with baked maps but both have problems and are, apparently, not suited for the task.
Re: A baking test scene
Posted: Thu Dec 12, 2019 10:02 am
by epilectrolytics
Maybe OIDN needs to be trained for those maps?
I think Attila Áfra has an
account here and could be consulted about the issue.
Also they have just introduced a lightmap denoise functionality which sounds similar to me.
Re: A baking test scene
Posted: Thu Dec 12, 2019 11:51 am
by zuljin3d
Maybe Optix will work?
Re: A baking test scene
Posted: Thu Dec 12, 2019 12:15 pm
by Dade
zuljin3d wrote: Thu Dec 12, 2019 11:51 am
Maybe Optix will work?
It should suffer of the same Oidn problems (i.e. not trained for baking maps).
Re: A baking test scene
Posted: Thu Dec 12, 2019 1:17 pm
by marcatore
Re: A baking test scene
Posted: Thu Dec 12, 2019 9:50 pm
by Racleborg
It’s only a simple setup so far, but it was only 4.15 seconds on a GTX 2070 – that is so fast!
The bin isn’t baked - I think it will look better that way when the camera moves about?
The glass isn’t baked – it’s the cycles glass node.
Thanks for this new tool. It's awesome!!
Re: A baking test scene
Posted: Fri Dec 13, 2019 9:42 am
by Dade
Racleborg wrote: Thu Dec 12, 2019 9:50 pm
It’s only a simple setup so far, but it was only 4.15 seconds on a GTX 2070 – that is so fast!
My end goal is to run it in real time with OpenGL/Vulkan/etc. I'm looking into glTF format (
https://www.khronos.org/gltf) and the idea is to have an exporter in that format at the end of all the baking steps. A glTF scene can be packed in a single binary .glb and seen in real time even inside a browser.
It is a very cool solution, I'm only a bit hesitant to spend a lot of time working on it because I consider this topic a bit out of the LuxCoreRender project focus.
However having a real-time side product of usual high quality LuxCore renders can be quite useful in ArchViz, industrial design, etc.
P.S. Racleborg, you may be able already to do it, by hand, by exporting your test scene in glTF from Blender.
Re: A baking test scene
Posted: Fri Dec 13, 2019 11:55 am
by Dade
For future reference, and example of glTF online viewer:
https://gltf-viewer.donmccurdy.com
Re: A baking test scene
Posted: Fri Dec 13, 2019 12:01 pm
by Racleborg
However having a real-time side product of usual high quality LuxCore renders can be quite useful in ArchViz, industrial design, etc.
Absolutely – it will be fantastic!
P.S. Racleborg, you may be able already to do it, by hand, by exporting your test scene in glTF from Blender.
I’ll certainly try that. At the moment I’m trying to improve my UV unwrapping skills, which are a bit shaky
With materials like metal that look different depending on the angle that it’s looked at (does that make sense - sorry for the layman’s understanding) is it best to bake or not? I’m guessing not as I think you have explained this to me before. (specular).
updated:
I'll try it later today, hopefully
Many thanks