CoolColJ's test pics thread

[CoolColJ]

render at 7300 samples, after many hours, and it's still adding more and more stuff...nowhere near finished yet. I'm not sure if these are caustics anymore...

with some bloom and glare it has a similar feel to the photo

[Kram1032]

looks like golden fairy dust or something like that, lol^^
I guess, that's a never-ending story...

[manitwo]

well CoolColJ, as for now this scene is impossible to render correctly. if a scene doesnt look right with that many spp i wont become any better - point.
and stressing ono to optimize it is senseless too - i'm sure he read your previous posts and is aware of that issue himself (and he's visiting ian's forum also).

[CoolColJ]

I estimate it might look right at around 21000 samples

I did a test render of this scene at 160x120 up to 12,000 samples and it only showed some big white dots on the water, which at that resolution is hard to tell
indirectly seen caustics are slow at the best of times, I probably should have just made a small bucket of water...

How they got underwater caustics to show so quickly and cleanly in the original MLT paper with the computers they had back then is beyond me

http://graphics.stanford.edu/papers/metro/

2.5 hours 800x500 100 mutations per pixel with 190mhz MIPS R1000 Processor 10 years ago!
There must be something missing in the MLT puzzle

[OnoSendai]

Well, in that image, the light source is obviously very large (occupying a large solid angle from the water), as you can see from the column shadows. Illumination from the sun is a very different and more difficult problem, due to the small solid angle subtended by the sun.

[CoolColJ]

maybe I should try it with a plane emitter then?

[OnoSendai]

Bigger emitter = faster caustics.
If you really want to track down the problem you try the pool scene with a large mesh/plane emitter, then make it smaller and smaller until its comparable with the sun.

[Kram1032]

yes, try so - wont be a fake-solution for sun, though...

what about Ian's approach? you "only" need to mark sun ray

[CoolColJ]

what about an EXR/HDRI ?

[OnoSendai]

btw can u jump on to chat on front page CCJ?

[CoolColJ]

Ono

what should the emitter values be, colour temperature and intensity wise, settings etc if I want to emulate the physical Sun?

How large?

should the mesh be a sphere?

[CoolColJ]

a large 6m light nearby sorta works, but doesn't look like the Sun though. Lights seem to have some falloff etc and well you don't get the sky
Big lights also don't generate the caustics all that well

Blackbody lights are bright as hell! I had to use only 2 watt and efficacy of 1!
Otherwise the RGB background colour gets blacked out and the textures get washed out. Background colour has no gain though in Cinidgo, should it?

And I get these shading errors on the cylinders with lights, even though they are not that low poly and have smoothing on. You can see the segments! Not a problem with Sun/Sky, so don't they shade properly with lights?

[CoolColJ]

Just mucking around with my C4D sample scene - rendered up to 1500 samples with exit portals, although it looked pretty good well before then

[Vanessa07]

Hi

I don't understand why caustics are outside the pool and not inside. I try everythng

[Big Fan]

do you have a 'glass' mesh in those openings or nothing?
the water looks like it has ripples...yes?