YafaRay rendering plug-in

[bdfd]

Hi,

Here are my own tests using cec's project.

1. standard rendering
2. enkonyito's Ver. 1.8 beta2
3. Yafaray's ver 2



No crash for me in this yafaray's second version !

[;)]
.

[hansmex]

Test on Demo home #7
A further test under Ubuntu shows that YafaRay uses only one (1) core during the preparation of the rendering (approx 35 seconds).
Once that is done, it uses all 6 cores of my computer for the full 100% during the actual rendering (13-14 minutes).

Test on Cecilia's test file
Testing Cecilia's file, I decided to do a Q1 test run to see if I had the same image as in the other tests. YafaRay behaves VERY STRANGE and shows me a more or less random part of the room in close-up, not the center of the room where the virtual visitor is looking at. Q2 gives the same result.



In Q3 and Q4 YafaRay is lightning fast and delivers an image within one minute.
Interestingly, the processor usage during the preparation phase can be very different when you do two consecutive renders of the same image.





H

[Mike53]

found ver. 2



[:(]

[enkonyito]

Now, SH3D crashes at every rendering whatever the scene. (Windows 10 64-bit)

I imported the source code of the plug-in into Eclipse (Projects from Folder or Archive) and added the necessary libraries except for the import [font=courier new]javax.jnlp[/font] .
I found that SH3D crashes when accessing the createMaterial, createBackground, createTexture, and createLight functions.

[Puybaret]

I found tonight how to make the plug-in work under Windows with its version 1.0 Beta 3 (running Sweet Home 3D either with the Jar executable, or with the free and paid installer versions).
Actually, I had to rebuild Yafaray with the latest version of MSYS2, and use more recent versions of the DLLs it depends on. I'll try the same rebuild under Linux in the coming days.

YafaRay is not using more than 32% of my CPU capacity, whereas Sunflow is using 98%.

It's strange, it's not what I see on my iMac, where YafaRay and SunFlow use as much CPU power as possible. I wonder where it could come from...

Just out of curiosity: Why go for another CPU-renderer and not a CPU/GPU-renderer?

There are not so much open source candidates that could be tested. From last year study, YafaRay seemed a better candidate even if OSPRay which uses the GPU too seemed interesting too.

[Ceciliabr]

There are not so much open source candidates that could be tested

Did you consider/test this one :
https://www.cycles-renderer.org/about/

Cec

[hansmex]

@Emmanuel
Congratulations!! Beta 3 now works under Windows 10, 64 bits (on my computer).
In Q1 and Q2, I get the expected image, contrary to what happened under Ubuntu with beta 1.
In Q3 and Q4, I also get the expected image, and about as fast as under Ubuntu.

Under Ubuntu nothing has changed as far as I can see.

H

[Puybaret]

Did you consider/test this one :
https://www.cycles-renderer.org/about/

Cycles was part of the last year study, but its performances didn't seem better than SunFlow.

[enkonyito]

Test version 1.0 Beta 3
Instability bug on Windows 10 64-bit fixed!

By varying the size of the tiles, the YafaRay rendering is quite stable.


For the same size of tile/bucket, YafaRay is a little faster than Sunflow.

[Mike53]

I have to say i think having a cpu based renderer as an option might serve users better. I don't usually comment on computer based topics, as i know little of what black magic and wizardry that goes on inside the black box attached via an umbilical to my screen, so if i have this wrong, well, i'm not surprised.
Not all pc's are game based, which have more powerful, better quality graphics cards, with better gpu's than standard pc's which are more work horse orientated and therefore are biased towards more grunt from the cpu. It may well be that different computers give different results/improvement by using the render engine better suited to individual systems.
I totally understand that Emmanuel/Enkonyito cannot be expected to test all types of pc configuration, but, that's what we are here for .
Mike