PM Caustics - still having issues

[simmsimaging]

Yep - seems to have it solved:



I am happy with that - although it's brutally slow to render (53 minutes at 640X480!). Thanks again for your help Dmitry!

b

[anchovy]

WaterDrops with Caustics

Brute force at at 41mins on an old machine. Im thinking the sampling is hi so im sure it could be faster. Nice effects though. was getting a similar look but noisier with PPT. Ill try BDPT tomorrow if I get the chance.



-Michael

www.morphographic.com

[anchovy]

About half the samples and a less oblique light. Yeah the sampling could be a lot less than the first image and a bit more than the 32 brute force samples I had for this one.



-Michael

www.morphographic.com

[kosso_olli]

Michael, do you have a scene for this one? Looks great, would like to play around with it.

Regards

Oliver

About half the samples and a less oblique light. Yeah the sampling could be a lot less than the first image and a bit more than the 32 brute force samples I had for this one.

[ATTACH=CONFIG]14066[/ATTACH]

-Michael

www.morphographic.com

[Morbid Angel]

Well, I think the drops should be darker then the bg, and here is why, normally glass or water or any refractive object has absorption, so it tends to tint the environment color. If you simplify your scene to no exposure, set all lights to 1 and turn your blue plane to be not visible to refraction, then the droplets render full white, and only become black in the areas where your ramp is refracting black.

The dome light in it self is casting shadows also, which is contributing to the darkening effect. I think we are correct here, just the controls for the brute force caustics could be expanded a bit. Vlado perhaps adding a caustic multiplier for brute force caustics would be good to allow user to increase/decrease the effect.

Here is a reference image

http://fonris.ru/_ph/17/12362275.jpg

[anchovy]

Here is one I kicked off before calling it a night. Didn't consider the faceting of the spheres but it seems to render in good time.



-Michael

www.morphographic.com

[Morbid Angel]

Here is one I kicked off before calling it a night. Didn't consider the faceting of the spheres but it seems to render in good time.

[ATTACH=CONFIG]14072[/ATTACH]

-Michael

www.morphographic.com

That looks pretty damn real to me

[simmsimaging]

It might be correct but I have done and worked with a lot of photography of water drops over the years and I never had the problem of the drops being dark like that when dealing with actual lights. In any case, it's a lot closer now and the shadows off/on for the dome can simply be an art direction call now that I have some control over it.

Michael: your last test looks great too. BDPT?

[anchovy]

It might be correct but I have done and worked with a lot of photography of water drops over the years and I never had the problem of the drops being dark like that when dealing with actual lights. In any case, it's a lot closer now and the shadows off/on for the dome can simply be an art direction call now that I have some control over it.

Michael: your last test looks great too. BDPT?

Thanks guys, the last one is still just Brute Force, reasonable samples and the addition of a nice hdr in the dome light as well as moving the area light over to screen right and making it a bit larger.

Nothing fancy. I hesitate posting the scene file only because I may have screwed up the units when I first brought it over. My light values are unreasonably high but other than that is fairly typical settings I use all the time other than I don't often use solely brute force. But I certainly will in some nasty lighting situations. The biggest limitation that i'm seeing is just the drop geo. But that's easily remedied.

-Michael

www.morphographic.com

[simmsimaging]

re: HDRI etc - this highlights where I have always had problems with things like water drops. In order to get them looking good you have to use highly specific lighting, change up the environment etc. etc. Whereas in photography you can typically sweat-up a bottle in lighting that works for the bottle and the droplets look great. They do not need such special circumstances - which is why I have always struggled with this and had to do so much work in post or with multiple passes.

Anyway, the pure brute force method is looking good to me, but the samples do need to be pretty high if the caustics are even remotely sharp - but that is not a super common occurence in my work. I mainly need drops that don't look like dead spots

/b

[Raven]

Hey Brett

Recently did a beer commercial, what I ended up doing was IR with detail enhancement. Seamed to work quite well so it was only doing high sample Brute force on the droplets.
Wasn't using a dome light though, just hdrs on image planes.

My 2 cents worth.