Sheldon: Kitchen Scene

#1


Crits are most welcome.
Thanks in advance

#2

Interesting Scene!

I’d like to see it without GI or FG for comparison, so I can see what the GI is adding.

The yellow fill around the cabinet doors in interesting, I don’t know where it comes from. The cork casts a dark shadow without the yellow fill, so I guess it’s in some areas and not others? The reflection on the table seems as if there is a green bottle, maybe that’s just a highlight?

-jeremy

#3

Here is the same scene with no GI and FG

The one on top has gi and fg
The green reflection on the table, I thought would be cool to add so I placed a volume light in the vicinity that gave me that green specular line on the table.
What I dont understand is how does stuff look more occluded/shadowed when I use fg (the chair , the wall behind the trashcan.) I love the nice beam of light I get on the chair for example… I use fg and its only a sliver.

#4

See if you can simplify the lighting so the green light only comes from the green particles, and casts very soft shadows. It looks like you have an extra light casting a hard-edged shadow of the trash can and the sink faucet. It appears to be coming from a low angle, well below the green particles.

For colored light, the photon color should (roughly) match the light color.

-jeremy

#5

What I dont understand is how does stuff look more occluded/shadowed when I use fg (the chair , the wall behind the trashcan.) I love the nice beam of light I get on the chair for example… I use fg and its only a sliver.

That is weird. FG is entirely additive (diffuse + (surface_color * irradience)). Unless some of the lights/shadows around those areas are producing negative values, FG can’t darken any parts of your scene.
You sure you didn’t move the lights any between the two renders?

Nice creative touch with the scene. The only thing I’m wondering about is the cabinets, which I think Jeremy was pointing out. The sharp rim towards their base in the GI version seems a little strange. I know that it’s coming from the GI/FG, but it just seems a little weird, considering the position of the key light, that it’s just the edges that’re picking up the light there, while nothing else in that area is.

EDIT: Odd… The wireless connections here are blocking photobucket for some reason (or maybe it’s the other way around?). No wonder I couldn’t see any of your images last week… Why the hell would anyone need to block photobucket? :shrug:

#6

FG can make things darker when most of the scene is black, and the FG rays see the blackness. In scenes that have GI, if they aren’t surrounded by brightness, the FG rays will see the darkness and average it in, and the scene gets darker with FG turned on than without FG. I’d like to look at the scene in class before I say anything more specific, because I don’t know if the shadow on the chair difference is really from whether MR has to raytrace the primary rays or not, but sometimes you can get useful insight into what’s causing changes in your scene by going in stages (raytracing off, raytracing on but no GI or FG, raytracing and GI on but no FG, GI and FG) and comparing the results.

It’s also OK for a test to turn on FG without GI. Working this way is something we’re going to talk about in class and I want people using full GI for the assignment, but it’s something you can do in a test image if you want to see whats influencing an aspect of your output.

I hope setting the photon color to green for the green light will tone-down the yellow fill that’s appearing on the cabinets. Sometimes if the photons are too big you end up needing to use some occlusion on your GI pass to get these things working really well, but for this first version I’m hoping that people keep things simple and just show renderings with GI without doing compositing work.

-jeremy

#7

Well, seeing some of the source code, I have to disagree with that. The following is a section of code from your standard MR shading model, straight from mental image’s base shaders which, no doubt, Autodesk/Alias’ shaders are very similar to in terms of mr-specific features (like indirect illumination):


// Diffuse illumination
	/* Loop over all light sources */
	for (n=0; n < n_l; n++, light++) {
		sum.r = sum.g = sum.b = 0;
		samples = 0;
		while (mi_sample_light(&color, &dir, &dot_nl, state,
					*light, &samples)) {
			sum.r += dot_nl * diff->r * color.r;
			sum.g += dot_nl * diff->g * color.g;
			sum.b += dot_nl * diff->b * color.b;
		}
		if (samples) {
			result->r += sum.r / samples;
			result->g += sum.g / samples;
			result->b += sum.b / samples;
		}
	}
// Indirect illumination. 'diff' in this scope is not diffuse illumination; it is surface color.
// 'result' already holds the diffuse illumination, from the previous block
	/* add contribution from indirect illumination (caustics) */
 	mi_compute_irradiance(&color, state);
// Add the indirect illumination on top of 'result'(the diffuse illumination)
 	result->r += color.r * diff->r;
 	result->g += color.g * diff->g;
 	result->b += color.b * diff->b;
 	result->a  = 1;

Hence, additive. And if it’s additive, the only way that indirect illumination can ever darken the scene is if the indirect illumination at any point is negative, or if the surface color is negative. The latter isn’t very likely, and the only thing that could cause the former to be true is if something funny is being done with the lights (negative intensity, negative shadow color, etc.).
In the cases where final gather does see nothing but black, it should not darken the scene at all. All that that means is that there is no contribution from indirect illumination at that sample, and the color returned will be the regular diffuse illumination.

If you still disagree, try this: Create a plane, and give it Maya’s lambert shader. Place a sphere directly above it, with another lamber shader, only set the surface color of this shader to be pure black. Turn on final gathering. If things worked as you had suggested, you would expect to see a small dark patch on the plane at the point where the sphere is above the plane. But you won’t; the scene will appear exactly as it appeared before fg had been enabled, because indirect illumination is additive, and an rgb of 0,0,0 adds nothing. And if you think about it, it kind of has to be that way, or else you could never use FG without having an environment map: The environment map is factored into FG’s calculations, and when there is no environment map, the color FG rays see when they trace to the environment is pure black. FG would kill uncontained, environmentless scenes.

Anyway… sorry! Way off topic. Don’t want hijack your thread, Sheldon. :banghead:

#8

It’s true that it’s the GI that usually gets darker in turning on FG, and yes that’s still adding brightness to the diffuse shading compared to if there were no GI or FG, but you can see it get much darker when you turn on FG in a scene with GI in a limbo environment. I guess this’ll be a demo in class on Monday.

-jeremy

#9

Okay, okay. If you’re just referring to GI+FG vs. GI w/o FG, then yeah, I’ll concede that point. :stuck_out_tongue:

#10

Final gathering seems to be more of an averaging of colors. Come to think of it occlusion works the same way, only difference it looks at distance. In fact if one does some reading, it is basically a lookup in a kdtree of stored points based on a radius you provide it and it averages the colors given out from the tree…
Which would give interesting effects if say the room model has an open door where some photons would escape wouldnt that area be predominatly darker as opposed to the other areas of the room after final gathering? Wouldnt it be more efficient if the 4 walls of a set be shut tight so that all the photons get bounced and the fg calculated be accurate?

#11

You guys would like MasterZap’s blog if you don’t read it already – scroll down to the section on MR 3.5 for what’s new in Final Gather in the current version:

http://mentalraytips.blogspot.com/

-jeremy

#12


Decided to take a bright pass to this. What do you think?
I think the pink on the table is too loud, but I like that opposing color, not sure though…
Thanks for the c&c’s in advance

#13

I like it generally being a bit brighter, but I think it’s too uniform a brightness. The illumination doesn’t feel motivated anymore by the bottle explosion. Such as the corner above the top cabinet is almost the same intensity as the edges of the table - and the seat cushion should be way more in shadow if the only light source is the bottle directly above it.

Hope it helps…

#14

Please post all the versions as GI/noGI pairs of images.

You’ve actually got pretty good quality shading, in terms of getting rid of glitches in the GI. But I don’t see all the light looking as if it comes from the glowing green particles, and getting that green light working should be priority #1 for your scene.

-jeremy

#15

So given the critiques I got in class here is my update…
no Gi

with Gi&Fg:

c&cs are most welcome.

#16

Could you post an image with just the key light, no fill or gi or anything else, just the green glow?

I think the key shadow looks very hard-edged for a light source that’s a few feet tall, and I think you have lots of other stuff going on there that could probably be reduced or simplified.

-jeremy

#17

This is my key light with no gi or fg

#18

Work on the shadows.

-jeremy

#19

This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.