sharper GI shadows?

[KonstantinosD]

I've just made a quick test on the same exact interior scene with just a sun and light coming through the window on the right. I used three different (unbiased) renderers, first is Octane, Second is Thea and third is Corona. I don't want to start a comparison between renders or get too deep into settings etc but as you will see although octane is the faster one, is also the most diffused-washed out one. Shadows of the table and chair are not so defined. This is rendered with PT and I've tried with PMC as well, same result. Can someone explain how to get sharper GI shadows? suggestions are welcome.

[RobSteady]

Interesting. Can you share the scene?

[aggiechase37]

I'm just guessing, but have you tried to fiddle with path term power?

[RobSteady]

I'm just guessing, but have you tried to fiddle with path term power?

Has nothing to do with this. Check the manual.

[KonstantinosD]

I can't share that one but I will make a version with free models and share that one ..

[bepeg4d]

Interesting comparison... just a guess, but I think that Gamma and response curve have a great impact here for comparison
What are your camera imager settings?
Also I wonder if you have tried to reduce the GI Clamp value, in theory you should have more contrast on GI speeding up the render
ciao beppe

[brasco]

To me, it looks like the environment is totally different, are you sure the sky in Octane isn't flat white?

Would be good to normalize the comparison using Linear responses for each and the same # of bounces etc.

cheers
brasc

[KonstantinosD]

I'm sure, I used the sun on the same location on all, no flat sky and also I used the linear response to normalize the outputs but still the same result. I'm uploading a different scene using free models, so you can check it yourself. The more people testing it the better, maybe we'll find a solution or even it's something simple that I totally overlooked, which sometimes is..

[KonstantinosD]

Here are my results with the new scene: (also desaturated the images to even the different sky models colors)

Feel free to download the scene, try for yourself and share the results here!

Octane_test.rar

(11.01 MiB) Downloaded 430 times

[brasco]

That is really interesting, thanks for the test scene.

It's definitely the distance the GI rays are travelling that washes out the detail out. If you clamp it more and drop the steps to 4, you'll see you get the shadow definition back, but the scene will darken and you'd have to compensate with really high exposure:



This is a that with a really low GI clamp of 0.0075 and 20 Exposure(!). I also got rid of the vignetting and also increased the highlight compression a touch. Coherence is off in this one too just in case (I highly doubt that impacts the GI shadows)

Still not a match, but a bit closer in the shadows. Would like to know exactly what's going on behind the scenes on each engine to cause this difference.


cheers
brasc