Dark refractions with material X

   1364   20   4
User Avatar
Member
838 posts
Joined: April 2008
Offline
I just got in to a issue with refraction when I have it interacting with other surfaces.
In the first image I have the droplets on the fruit getting a little dark inside where with caustics on I would expect to see only the refraction and reflections and no self shadow.
In this case my apple have a SSS value of 0.5 if I change to 1.0 then it seems the my refractions start to work properly but I loose the caustics on the apple in the 2nd image.

I thought this behavior weird so I decided to post here.
All materials are Material X and rendered with XPU
I am using Houdini 19.5 build 332
Edited by mzigaib - Aug. 8, 2022 21:12:46

Attachments:
original_close_up.fruits_H19.5_t01_v14.0001.jpg (668.8 KB)
droplets_issue_sss_1.0.jpg (697.4 KB)

https://vimeo.com/user2163076 [vimeo.com]
User Avatar
Member
349 posts
Joined: April 2018
Offline
I've experienced similar issues and despite tons of experimenting I can't make it look good. It's way too hard to get simple caustics working in Karma when in other renderers it's usually just a matter of ticking a checkbox. I think I'll email SideFX to see if their Karma gurus can recommend a solution.
Watch Perceptarium on Amazon! https://www.amazon.com/dp/B0846GB3WG/ [www.amazon.com]

http://youtube.com/brianhanke [youtube.com]
http://twitter.com/brianhanke [twitter.com]
https://www.instagram.com/brianrhanke [www.instagram.com]
http://facebook.com/brianrhanke [facebook.com]
User Avatar
Member
6075 posts
Joined: Sept. 2011
Offline
Does the droplet geo go inside the apple skin geo? If so, then it could be shadowing the SSS scattering on the inside, as I don't think SSS paths take caustics into account.
User Avatar
Member
838 posts
Joined: April 2008
Offline
BrianHanke
I've experienced similar issues and despite tons of experimenting I can't make it look good. It's way too hard to get simple caustics working in Karma when in other renderers it's usually just a matter of ticking a checkbox. I think I'll email SideFX to see if their Karma gurus can recommend a solution.

Thanks for the feedback! I am not tech savy enough to figure what is going on under the hood, for me it seemed like a self shadow issue where in Mantra a simple faux caustics would fix it. Having a "easy" option to exclude self shadow would also be useful.
https://vimeo.com/user2163076 [vimeo.com]
User Avatar
Member
838 posts
Joined: April 2008
Offline
jsmack
Does the droplet geo go inside the apple skin geo? If so, then it could be shadowing the SSS scattering on the inside, as I don't think SSS paths take caustics into account.

It does I used to do it all the time in Mantra, I also did a test without intersecting and the issue is the same.
https://vimeo.com/user2163076 [vimeo.com]
User Avatar
Member
349 posts
Joined: April 2018
Offline
mzigaib
BrianHanke
I've experienced similar issues and despite tons of experimenting I can't make it look good. It's way too hard to get simple caustics working in Karma when in other renderers it's usually just a matter of ticking a checkbox. I think I'll email SideFX to see if their Karma gurus can recommend a solution.

Thanks for the feedback! I am not tech savy enough to figure what is going on under the hood, for me it seemed like a self shadow issue where in Mantra a simple faux caustics would fix it. Having a "easy" option to exclude self shadow would also be useful.

Yep exactly! Faux caustics should help, but it's hard to control as I posted about a few days ago. Excluding shadows would also be great. Arnold in Maya has a "receive shadows" checkbox which seems like a pretty basic feature that isn't present in Karma.
Watch Perceptarium on Amazon! https://www.amazon.com/dp/B0846GB3WG/ [www.amazon.com]

http://youtube.com/brianhanke [youtube.com]
http://twitter.com/brianhanke [twitter.com]
https://www.instagram.com/brianrhanke [www.instagram.com]
http://facebook.com/brianrhanke [facebook.com]
User Avatar
Member
6693 posts
Joined: July 2007
Offline
jsmack
Does the droplet geo go inside the apple skin geo? If so, then it could be shadowing the SSS scattering on the inside, as I don't think SSS paths take caustics into account.
they should with SSS limit > 0
Tomas Slancik
FX Supervisor
Method Studios, NY
User Avatar
Member
6693 posts
Joined: July 2007
Offline
BrianHanke
mzigaib
BrianHanke
I've experienced similar issues and despite tons of experimenting I can't make it look good. It's way too hard to get simple caustics working in Karma when in other renderers it's usually just a matter of ticking a checkbox. I think I'll email SideFX to see if their Karma gurus can recommend a solution.

Thanks for the feedback! I am not tech savy enough to figure what is going on under the hood, for me it seemed like a self shadow issue where in Mantra a simple faux caustics would fix it. Having a "easy" option to exclude self shadow would also be useful.

Yep exactly! Faux caustics should help, but it's hard to control as I posted about a few days ago. Excluding shadows would also be great. Arnold in Maya has a "receive shadows" checkbox which seems like a pretty basic feature that isn't present in Karma.

- if you don't want your droplets to cast shadows, use Render Geometry Settings, Enable Render Visibility and set it to: -shadow

- also some shadowing is natural for dielectric surfaces, simply the reflected portion is light that will not reach the surface therefore will cause shadow, so turning shadows completely off will most likely be unnatural, but its an option

- faux casutics are automatic when real Caustics are off even though in that case there is no additional control on top of IOR
Tomas Slancik
FX Supervisor
Method Studios, NY
brians
Staff
187 posts
Joined: May 2019
Offline
jsmack
Does the droplet geo go inside the apple skin geo? If so, then it could be shadowing the SSS scattering on the inside, as I don't think SSS paths take caustics into account.

In Karma (both CPU + XPU) SSS "random walk" will only calculate scattering with regards to the local mesh, and will ignore surrounding meshes. So this means the SSS calculation on the apple will only take into account the apple geometry, and ignore that the droplet is there.

This may change in the future (named trace-sets) but is the behavior currently

tamte
faux casutics are automatic when real Caustics are off even though in that case there is no additional control on top of IOR
BrianHanke
Faux caustics should help, but it's hard to control as I posted about a few days ago

So as Tomas says, faux caustics are on automatically in H19.5
Are you saying you want more fine-grained control?
User Avatar
Member
349 posts
Joined: April 2018
Offline
brians
Are you saying you want more fine-grained control?

Here's the thread where I asked my question: https://www.sidefx.com/forum/topic/85850/ [www.sidefx.com] The Fake Caustics node gives me a great iris, but it also turns the entire eyeball into the surface of the sun so there are no longer shadows on the floor.
Watch Perceptarium on Amazon! https://www.amazon.com/dp/B0846GB3WG/ [www.amazon.com]

http://youtube.com/brianhanke [youtube.com]
http://twitter.com/brianhanke [twitter.com]
https://www.instagram.com/brianrhanke [www.instagram.com]
http://facebook.com/brianrhanke [facebook.com]
User Avatar
Member
838 posts
Joined: April 2008
Offline
tamte
- if you don't want your droplets to cast shadows, use Render Geometry Settings, Enable Render Visibility and set it to: -shadow

If I do that than the droplets won't cast shadows on the apple which would be unnatural, I am referring to objects not casting shadows on itself but the rest. I hope that makes sense.

tamte
- also some shadowing is natural for dielectric surfaces, simply the reflected portion is light that will not reach the surface therefore will cause shadow, so turning shadows completely off will most likely be unnatural, but its an option

- faux casutics are automatic when real Caustics are off even though in that case there is no additional control on top of IOR

If you see here on this Entagma tutorial illustrate very well I am talking about, if you see the droplets you just see the refraction and the reflections of the droplets although you also see the shadow of the droplets on the objects that it is interacting but not on itself, the way I see that is natural and it is exactly what I get with mantra with fake caustics I don't have this control on Karma.

Your text to link here... [entagma.com]
Edited by mzigaib - Aug. 9, 2022 00:52:30
https://vimeo.com/user2163076 [vimeo.com]
User Avatar
Member
6693 posts
Joined: July 2007
Offline
mzigaib
I am referring to objects not casting shadows on itself but the rest
that's what I initially thought you meant by self shadowing, but since it's not possible for dielectric object to cast shadows on itself (since it doesn't have any diffuse, sss or volume component) so I assumed you potentially mean just shadow casting
Tomas Slancik
FX Supervisor
Method Studios, NY
User Avatar
Member
838 posts
Joined: April 2008
Offline
tamte
that's what I initially thought you meant by self shadowing, but since it's not possible for dielectric object to cast shadows on itself (since it doesn't have any diffuse, sss or volume component) so I assumed you potentially mean just shadow casting

So if it doesn't, cast shadows on itself it should not get those darker internal refractions with caustics and dielectrics correct? And just in case I also tested with CPU and the result is the same.
Edited by mzigaib - Aug. 9, 2022 01:10:26
https://vimeo.com/user2163076 [vimeo.com]
User Avatar
Member
6693 posts
Joined: July 2007
Offline
mzigaib
So if it doesn't, cast shadows on itself it should not get those darker internal refractions with caustics and dielectrics correct?
it is still casting shadows on the apple, and you can see those shadows through refractions
those shadows can be brightened by caustics, but obviously you don't want them to be completely gone in case like droplets, otherwise it wouldn't be realistic
of course it also depends on your lighting
Tomas Slancik
FX Supervisor
Method Studios, NY
User Avatar
Member
838 posts
Joined: April 2008
Offline
tamte
mzigaib
So if it doesn't, cast shadows on itself it should not get those darker internal refractions with caustics and dielectrics correct?
it is still casting shadows on the apple, and you can see those shadows through refractions
those shadows can be brightened by caustics, but obviously you don't want them to be completely gone in case like droplets, otherwise it wouldn't be realistic
of course it also depends on your lighting

But it is just darkening the refraction and not refracting the apple with shadows and the darkening is diffused internally darkening the whole thing and it is not looking natural that is my point.
Edited by mzigaib - Aug. 9, 2022 01:42:26
https://vimeo.com/user2163076 [vimeo.com]
User Avatar
Member
838 posts
Joined: April 2008
Offline
One more example:
Just one light and the way I see it the drops close to the light or even the far ones should not be dark like this, it looks like it is self shadowing internally, makes sense?

Attachments:
apple_light.jpg (527.3 KB)

https://vimeo.com/user2163076 [vimeo.com]
User Avatar
Member
6075 posts
Joined: Sept. 2011
Offline
mzigaib
One more example:
Just one light and the way I see it the drops close to the light or even the far ones should not be dark like this, it looks like it is self shadowing internally, makes sense?

Is this true caustics? because it's most likely energy loss and not shadowing if so. Change the color limit to a million instead of the default of ten if your lights are small enough to need large values. Small light caustics generally don't resolve well under unidirectional path tracers anyways.
User Avatar
Member
6693 posts
Joined: July 2007
Offline
it's hard to debug based on screenshots, maybe post your file or simplified example
did you try raise SSS limit above 0 since you are using SSS?

I tried on a simple scene with SSS, it seems to behave pretty realistic without any special treatment
Edited by tamte - Aug. 9, 2022 04:05:02

Attachments:
water_droplets_caustics.jpg (139.1 KB)
water_droplets_caustics_v0001.hipnc (793.9 KB)

Tomas Slancik
FX Supervisor
Method Studios, NY
User Avatar
Member
664 posts
Joined: July 2005
Online
Color Limit is probably the culprit like jsmack mentioned. Anytime I've messed with caustics I've needed to increase it. The renders below are with the default value of 10, and then with the Color Limit set to 1000 (both direct and indirect).


Edited by Siavash Tehrani - Aug. 9, 2022 10:56:42

Attachments:
CL_1000.jpg (359.2 KB)
CL_10.jpg (321.8 KB)

User Avatar
Member
838 posts
Joined: April 2008
Offline
Thanks for the feedback guys!
I will try the tips given here and let you know what are the results since I have them.

Cheers!
https://vimeo.com/user2163076 [vimeo.com]
  • Quick Links