Light fall off problem ?

[Thomas An._cg]

Light fall off problem ?

Wouter,

If you redo the test it would be interesting to know what was the render time for your image .

Thanks

[flipside]

Light fall off problem ?

Here it is, pretty high quality settings for ir map and LC. I used now a ratio of 10/1 for strength of sun/sky, which was too much because now my overall image is brighter than maxwells. I wonder what kind of lights they used in the maxwell test. Does maxwell have a linear falloff light type? Or was this done with sun/sky system?

[Thomas An._cg]

Light fall off problem ?

Thank you for this test Wouter !

I wonder what kind of lights they used in the maxwell test. Does maxwell have a linear falloff light type? Or was this done with sun/sky system?

Well, in nature there one way that light travels (in the "air" medium; given normal turbidity and ozone levels), so Maxwell light is handled automatically internally... so to answer your question, no there is no need for falloff types (it is fully automated).

Also, no this was not done with physical sky because the shadows are wrong (non parallel). The guy just used that little sphere above the three windows.

[flipside]

Light fall off problem ?

Ah ok, if someone can figure out the ratio bewteen sky strength and sun strength, please tell me. Also the diffuse color of the grey walls, which is very crucial in this scene...

[corey]

Light fall off problem ?

@Thomas,

Is the scene rendered in Maxwell from Max or Rhino? Are you rendering that scene in Maxwell with Physical Sky and with Sunlight enabled?

Regards,

Corey

[Edit]Reading helps..."Also, no this was not done with physical sky because the shadows are wrong (non parallel). The guy just used that little sphere above the three windows."[Edit]

[Thomas An._cg]

Light fall off problem ?

@Thomas,

Is the scene rendered in Maxwell from Max or Rhino? Are you rendering that scene in Maxwell with Physical Sky and with Sunlight enabled?

Regards,

Corey

Hi Cory,

I have not tried this scene in maxwell yet, so I don't really know.

[Thomas An._cg]

Light fall off problem ?

Well, here is my try on this scene.
It was redone with sky and sunlight (the sunlight casts different shadows).

I think the Vray output is quite alright (although the back wall is still a little darker. Of course the gamma affects the materials and so the teapot color is a little more washed out even though the RGB is identical).

[flipside]

Light fall off problem ?

Hi Thomas,

Are you sure you used gamma correction on the vray image? The whole maxwell image looks brighter to me, at least in the dark parts which indicates different gamma (shift of the mid grey point). How did you match the lighting and materials in vray to the maxwell scene?

[Thomas An._cg]

Light fall off problem ?

Yeah, I was fiddling with the color mapper (gamma corrsction mode) and the bright/dark multipliers for a while... most of the time I was getting very flat images. (The back room would be lighter but the whole image would have no contrast) .... So, I gave up on the back room and just chose values for good contrast :P

For the colors, I used a photo and eyedropped the colors for the sky and sun. Used the same color and multiplier for GI and backround colors and used a directional light for the sun. For the teapots and the rest of the structure, the RGBs were chosen to have the same values in both renders.

One thing to notice is that in the Maxwell image the right-hand wall above the sliding doors seem slightly darker then the Vray image, but the backroom is lighter. Also, in the Maxwell image the light distribution in the recessed area of the front step (where the near teapot is sitting), is overall brighter and more even. So there are differences in illumination... the best would be to have a photo of this room and make sure (although, I have performed tests with photos and Maxwell in the past and it has performed very well )
-

[flipside]

Light fall off problem ?

Well that is a problem, the colors should be converted to a gamma 2.2 as well. In max you can use a color correct map to do so. If you look in my file, the grey of the walls looks very dark, but in fact it is mid grey if you look at the greyscale in gamma 2.2 space. So if you don't adapt the colors, you get very flat images if you use color mapping, because you are almost using white for the walls (all the colors will be washed out). Kinda hard to explain but like it is now, you cannot use gamma color mapping in rhino 100% correct because you have no display gamma settings which affects material creation and color swatches.

I'm not sure how maxwell handles the material colors when you use gamma correction there and why the colors don't get washed out.

Can you give me some details about your measured RGB values and light multipliers?

[vlado]

Light fall off problem ?

To be honest, I wouldn't bet my life on which renderer gives more accurate illumination... or that any of them is accurate at all, for that matter. Without any proof, it would be just as valid to assume that Maxwell is wrong, and V-Ray is correct... and vice versa. Proofs in this regard are difficult though. While I have made a few tests to confirm that V-Ray is correct in many specific situations as far as the theory is concerned, for more complex scenes it is very difficult to compute the right answer and compare renderings to it - so there is still work to be in that regard.

Best regards,
Vlado

[flipside]

Light fall off problem ?

If you consider how different the two renderers are in material approach etc, I find it already amazing that the results look so similar. This is proof enough for me that both generate a 'correct' result. At least it looks correct, that's more than enough for me!

[Thomas An._cg]

Light fall off problem ?

Can you give me some details about your measured RGB values and light multipliers?

Sky RGB: 34 66 102
(Use as both GI color and Background)

Directional light (Sun) RGB: 249 222 151

The sun/sky multipliers are 1:1 ratio. In this case I used 15 for GI and 15 for the light (but it doesn't have to be ... we could use 10 and 10, or other)

The RGB for the red teapot is: 171 49 49
The RGB for the blue teapots are: 60 70 140
The RGB for the rest of the structure (walls etc): RGB 128 128 128

[Thomas An._cg]

Light fall off problem ?

If you consider how different the two renderers are in material approach etc, I find it already amazing that the results look so similar. This is proof enough for me that both generate a 'correct' result. At least it looks correct, that's more than enough for me!

Well, the thing is that I have tried to "make" them look similar, by playing with the gammas and also with the curves thingy later of the framebuffer.
Like Vlado mentioned, maybe we don't know which one is more accurate unless there was a photo (this is one reason I do not really like this direct engine to engine comparisons, because there is no way to tell which one is correct)

One of these days I will do a cornell-type experiement (but with indirect lighting) and take some photos.

[flipside]

Light fall off problem ?

With a photo you will also 'make' it look like the photo. How will you ever know what multiplier the skylight or sunlight for example in photo is?