Foam point mode issue

[waaazoo]

No, will try to have a look at it next week...

Cheers!

ok, thanks a lot.

[waaazoo]

sorry to ask again. Any news on the subject ?

[Svetlin.Nikolov]

Nope, not yet...

[Svetlin.Nikolov]

Hey, just an update - I am looking at it, but still figuring out what's wrong.

Seems like the shader thinks these black spots are shadows, now the question is what are they doing in these places...

[Svetlin.Nikolov]

Finally! There will be a fix for this issue in tomorrow's nightlies.

Sorry for taking so long!

[waaazoo]

Finally! There will be a fix for this issue in tomorrow's nightlies.

Sorry for taking so long!

Hey, Sounds great ! So you are still working on phoenixFD

[Svetlin.Nikolov]

All day every day, but this was one of the nastiest issues I've seen for the past year, so took me a while to figure out

[waaazoo]

All day every day, but this was one of the nastiest issues I've seen for the past year, so took me a while to figure out

where did the issue come from ? Any explanations that you can share ?

[Svetlin.Nikolov]

Sure - each Point asked the lights how much it's illuminated at its center. However, instead, it should have asked from a point on its border that is nearest to the light. Otherwise, when points are very close together and overlap (for example when using a Point Radius more than 1), two points could turn out to be both shadowing each other, which if of course not correct. So the ingredients for the problem were particles that are very close together and larger Point Radius - and this could be also exaggerated by having higher Shadow Strength as well. If this happened, bunches of points were all mutually casting shadows on one another, making the whole bunch of points darker, while in fact all points that were on the surface of the bunch should have been fully lit by the lights.

[waaazoo]

Sure - each Point asked the lights how much it's illuminated at its center. However, instead, it should have asked from a point on its border that is nearest to the light. Otherwise, when points are very close together and overlap (for example when using a Point Radius more than 1), two points could turn out to be both shadowing each other, which if of course not correct. So the ingredients for the problem were particles that are very close together and larger Point Radius - and this could be also exaggerated by having higher Shadow Strength as well. If this happened, bunches of points were all mutually casting shadows on one another, making the whole bunch of points darker, while in fact all points that were on the surface of the bunch should have been fully lit by the lights.

ok, thanks a lot. It makes sense. Good find !

[Svetlin.Nikolov]

Thank YOU for helping catch this issue! It seems to have been happening much less often than I would imagine and your scene really made the problem show up clearly. In order to simplify the scene, I also added some custom code that allowed me to delete all particles from a cache outside a geometry and then save the result to a cache file - something like a render cutter for the cache converter - I hope this would go into the nightlies soon as well