CGI tech news box

General computer graphic news and papers.
User avatar
Sharlybg
Donor
Donor
Posts: 3101
Joined: Mon Dec 04, 2017 10:11 pm
Location: Ivory Coast

Re: CGI tech news box

Post by Sharlybg »

Dade wrote: Wed Mar 21, 2018 4:36 pm Importance Sampling of Many Lights With Adaptive Tree Splitting: https://sites.google.com/site/ckulla/home/many-lights

Interesting but it may be redundant once we have a light cache.
This light cache things are really for me the Next point to go.The impact on quality and efficiency is so huge. Just put a direct link to the PDF :

http://cgg.mff.cuni.cz/~jaroslav/papers ... slides.pdf
Support LuxCoreRender project with salts and bounties

Portfolio : https://www.behance.net/DRAVIA
User avatar
B.Y.O.B.
Developer
Developer
Posts: 4146
Joined: Mon Dec 04, 2017 10:08 pm
Location: Germany
Contact:

Re: CGI tech news box

Post by B.Y.O.B. »

Sharlybg wrote: Wed Mar 21, 2018 5:16 pm This light cache things are really for me the Next point to go.The impact on quality and efficiency is so huge. Just put a direct link to the PDF :
http://cgg.mff.cuni.cz/~jaroslav/papers ... slides.pdf
(from the paper/slides)
(from the paper/slides)
Do they really mean they rendered this image for 4 weeks?
A bit overkill if you ask me... or maybe their reference renderer is the "truest of the true" unbiased renderers?
User avatar
Sharlybg
Donor
Donor
Posts: 3101
Joined: Mon Dec 04, 2017 10:11 pm
Location: Ivory Coast

Re: CGI tech news box

Post by Sharlybg »

Do they really mean they rendered this image for 4 weeks? 
A bit overkill if you ask me... or maybe their reference renderer is the "truest of the true" unbiased renderers?
I think this to make obvious the kind of improvement they method make possible.

So they compare a 1hour render to a 1 month reference. And you can see it is relatively not so far. Also the reference image normally must be rendered with normal renderer (a less efficient method compared to what they show).

I also think 4 weeks is an estimation. Like they render the ref on a supercomputer for a certain amount of time and then return an estimation based on normal computer rendering performance (extrapolation).
Support LuxCoreRender project with salts and bounties

Portfolio : https://www.behance.net/DRAVIA
User avatar
B.Y.O.B.
Developer
Developer
Posts: 4146
Joined: Mon Dec 04, 2017 10:08 pm
Location: Germany
Contact:

Re: CGI tech news box

Post by B.Y.O.B. »

Sharlybg wrote: Wed Mar 21, 2018 10:15 pm I also think 4 weeks is an estimation. Like they render the ref on a supercomputer for a certain amount of time and then return an estimation based on normal computer rendering performance (extrapolation).
Good idea, that could be the case.
User avatar
Sharlybg
Donor
Donor
Posts: 3101
Joined: Mon Dec 04, 2017 10:11 pm
Location: Ivory Coast

Re: CGI tech news box

Post by Sharlybg »

Nice video rendered on GPU but with Precomputed light cache on CPU : :)

https://www.youtube.com/watch?v=U-t2d4M98Bc
Support LuxCoreRender project with salts and bounties

Portfolio : https://www.behance.net/DRAVIA
User avatar
Dade
Developer
Developer
Posts: 5672
Joined: Mon Dec 04, 2017 8:36 pm
Location: Italy

Re: CGI tech news box

Post by Dade »

Sharlybg wrote: Thu Mar 29, 2018 8:26 pm Nice video rendered on GPU but with Precomputed light cache on CPU : :)

https://www.youtube.com/watch?v=U-t2d4M98Bc
Isn't Vladimir Koylazov one of the founder of Chaos Group and owner of V-Ray and Corona ?
Support LuxCoreRender project with salts and bounties
User avatar
B.Y.O.B.
Developer
Developer
Posts: 4146
Joined: Mon Dec 04, 2017 10:08 pm
Location: Germany
Contact:

Re: CGI tech news box

Post by B.Y.O.B. »

Dade wrote: Thu Mar 29, 2018 9:23 pm Isn't Vladimir Koylazov on of the founder of Chaos Group and owner of V-Ray
Yes (btw. in the top right corner it says V-Ray RT).
Not sure about Corona though, is he involved there, too?
edit: Apparently ChaosGroup aquired Render Legion (the developers of Corona).
User avatar
Sharlybg
Donor
Donor
Posts: 3101
Joined: Mon Dec 04, 2017 10:11 pm
Location: Ivory Coast

Re: CGI tech news box

Post by Sharlybg »

Dade i think that the day you start to work on LC and denoiser stuff are good timing to launch a Special Crowfunding campaign. Just like blender foundation do with blender 2.8 project.
This money gathering model is really powerfull and can help the project get the old missing healthy level. People like to be involve in great project and their like to be part of change. Maybe this can lead to bigger sponsoring partnership.

It will be a good experiment to also test the community reactivity arround the project. Of course it is for days after public release.
Support LuxCoreRender project with salts and bounties

Portfolio : https://www.behance.net/DRAVIA
kintuX
Posts: 809
Joined: Wed Jan 10, 2018 2:37 am

Re: CGI tech news box

Post by kintuX »

Sequences with Low-Discrepancy Blue-Noise 2-D Projections (source)

Image
Left: our staged per-tile optimized scrambling, applied to a Sobol sequence of sampling points, produces a power spectrum close to Blue Noise. Right: Rendering of a challenging scene featuring depth of field and high specularity (jewels). The sampling was done with the Sobol sequence (top) and our sampler (bottom); both use 256 samples per pixel and 3 light bounces. Note the improvement in aliasing when using our method in comparison to the original Sobol sequence.
Video @ Youtube
User avatar
Sharlybg
Donor
Donor
Posts: 3101
Joined: Mon Dec 04, 2017 10:11 pm
Location: Ivory Coast

Re: CGI tech news box

Post by Sharlybg »

OCTANE 4.0 bring AI at all rendering level :

https://render.otoy.com/forum/viewtopic.php?f=7&t=66014
New Features of V4

AI Light

With the introduction of AI Light in Octane, you will see a great improvement in light sampling, especially in scenes that have many lights with localised distributions. As a learning system, AI Light improves as you render more samples. When used with Adaptive Sampling, AI Light gets even better, since it will learn that other lights become more important, as some pixels are no longer sampled.

AI Scene

AI Scene in Octane brings you much faster scene loading times, and a dramatic improvement in viewport interactivity. You are able to position and modify geometry in real time, even in scenes with several millions of triangles. Vertex animation with constant topology is updated in real time in complex scenes, as you move the time slider. Using the gizmos in the viewport on massive scenes is also in real time.

AI Denoiser

The much anticipated Spectral AI Denoiser allows you to render noise-free in a fraction of the time. You will find the settings to enable it in the Camera Imager settings.

Out of core geometry

Octane now supports out of core geometry, allowing you to use a significant portion of your system memory for geometry data. Multiple GPUs are also supported with out of core geometry.

In your preferences, you will see four sliders under Out-of-core. System RAM usage limit allows you to limit how much system memory Octane will use for out of core geometry and textures. Geometry data GPU RAM usage limit allows you to specify a limit on how much GPU VRAM you want to use for geometry, and underneath, you can set the same for textures. This allows you to balance the level to which textures relative to geometry are evicted from VRAM.
Support LuxCoreRender project with salts and bounties

Portfolio : https://www.behance.net/DRAVIA
Post Reply