I'm still doing something wrong with Glass

[scatman]

Hello all, and thanks for all of your help so far (http://www.indigorenderer.com/forum/vie ... =8&t=10734). I've been distracted as of late with blender 2.57+.
Well, I have everything set the way I want it except for the dna/rna strand with the rods. The strand with the beads is fine, but the one with the rods is still too dark when viewed from the side. In the render below you can see the different colors of the beads even when looking through the clear glass of the spiral, but when looking through the clear glass of the spiral with the rods, dark mixed colors show instead of the Red, Yellow, Orange, and Blue. The difference between the 2 strands is that the one with the beads doesn't penetrate the actual spiral clear glass colored portion of the strand; however, on the strand that has the rods--the rods themselves are penetrating the clear glass colored portion of the strand. I think that I'm doing something wrong with the Precedence and/or the Gain settings of Indigo (see settings below).


Please take a look at my attached .blend file and tell me how to make the strand with the rods look as good as the one with the beads.

Thanks in advance,
scatman

[scatman]

Opps, here's the .blend file compressed with 7zip...

[Headroom]

You may want to work on the scale of your objects. The default cube in Blender is 2m x 2m x 2m in Indigo by default.

Unless you have changes the scene scale in Blendigo these helixes are roughly 74 meters tall with the minor diameter of the helix being 90cm. The light has to travel through meters of glass. No surprise it's too dark.

The helix in the image below still is about one meter in diameter and lit by the St Peters EXR for some nice subtle caustics. Nothing too dark there and if you scale that down to diadem size it'll look even better.

[scatman]

Thanks, Headroom for replying.

I've tried scaling (twice) down the objects, as you've suggested, but it's still rending the same--just smaller each time.
Is it possible for you to take the actual .blend that I posted and modify it so that the colors are not dark on the rods. I've only been able to reduce the darkness of the rods by changing their Gain values to 1.0. Unfortunately this also changes their colors, and they still mix...so the Orange & Blue look Green from the side. Weird??

scatman

[galinette]

What do you find wrong exactly? Your post image looks good (at least physically!)

Etienne

[scatman]

If you look at the strand with the beads, starting and the bottom of the image (where the spiral glass is shaped like an 'X') you'll see yellow beads, then as you look up the the next 'X' you'll see blue beads, then the next 'X' you'll see red beads, then the next 'X' you'll see orange beads.
If you look at the corresponding 'X' part of the rod strand, you'll see a dark color followed by a red color (these colors repeat) as you look at the strand starting at the bottom and make your way up.

The rod dna strand should show the same (or similar) color variations as the bead dna strand at the 'X' sections.

scatman

[galinette]

I think what you see is a physically correct effect.

In the rods, light entering by one end is guided by total reflection like in a glassfiber. Which means light is forced to bounce inside the rod and cannot escape until the other end.
Additionnally to this, you set two almost complementary colors to each half of the rod. Which means light has to cross both colors without escaping. Since these colors are quite dense, and since the substractive combination of two complementary colors is black, you see black.

On the other hand, stacked spheres will not guide light like in a glassfiber. So the light exiting one end mostly crossed the first sphere only. Which means it has roughly the same color as one sphere viewed from the side.

I'm almost sure this is not a modelling issue or renderer problem. It's a physically correct effect. This means you have to change your scene!

To avoid absorption being too dark after some distance in a medium, you can use the fact your renderer is spectral. To make a blue color, instead of using RGB absorbance, use a square bandpass filter spectrum in the blue wavelength. This allows to make dense absorption colors while avoiding black opacity at longer distance. And this is something a RGB raytracer will never render correctly!

I'll try to make some demo of this some day...

Etienne

[CTZn]

use a square bandpass filter spectrum in the blue wavelength

If I understood correctly, Galinette mean a spectrum definition of the 'peak' type, with the peak starting at about 450nm for a width of 45nm approx.

[scatman]

Thanks for you help, galinette. I understood you up until the following statements:

"To avoid absorption being too dark after some distance in a medium, you can use the fact your renderer is spectral. To make a blue color, instead of using RGB absorbance, use a square bandpass filter spectrum in the blue wavelength. This allows to make dense absorption colors while avoiding black opacity at longer distance."

I think I'll try paring each of the 4 colors (Red, Orange, Yellow, Blue) with a clear rod instead and see how that looks...

scatman

[galinette]

I will bake some suited designed spectra for you if you want. Example is always better to understand!

Etienne

[scatman]

Thank you, galinette. Please bake the spectra that you mentioned.

scatman

[dag]

After downscaling you need to adjust the gap or overlap on the rods.

[smashpapst]

One think I remember from my good ol' Blender 2.49 days is not to forget to Apply Scale (something like Ctrl+A, I think, on object level), when it comes to physical correct scales.

Just a thought..
Smash IV.

[scatman]

Hey dag, thanks for your help. Could you please upload the .blend file from which you got the renders for you post with the 0.2mm gap?

scatman

[dag]

Don't have it anymore, but how hard can it be?
Scale everything down 100 times and add a gap.