absorption and scattering coefficients for liquids

General questions about Indigo, the scene format, rendering etc...
User avatar
Whaat
Developer
Posts: 1827
Joined: Fri Dec 22, 2006 6:15 am
Location: Canada
Contact:

Post by Whaat » Tue Aug 14, 2007 3:09 am

OnoSendai wrote:As far as I can tell, the coefficients given are actually for the media arrived at from mixing the given volume V with 23-V liters of water.
So the fractional concentration is V/23-V.
So to get the coefficients for the 100% concentration, you need to multiply by (23-V)/V.
I have done many liquid tests with Indigo 0.9 using values from Jensen's paper and I can confirm that the values given in Jensen's paper are for the given volume V diltuted in 23-V litres of water. To get values for 100% concentration I multiplied the values by 23/V since that is the method described in the paper to extrapolate to higher concentrations. This seemed to give correct results for absorption and scattering. Gamma must be set to 1 for absorption, scattering, and Henyey-Greenstein.
However, there still seems to be a bug in version 0.9 with SSS because the milk materials would not render correctly (basically black, no scattering).
Using the same values and redering in version 0.8, the milk material seemed to render correctly. Ono, can you confirm that this is a bug? I was using RGB scattering values of approx. 18,000 for the milk material with gamma of 1.0.

User avatar
Kram1032
Posts: 6649
Joined: Tue Jan 23, 2007 3:55 am
Location: Austria near Vienna

Post by Kram1032 » Tue Aug 14, 2007 3:46 am

mrCarnivore:
Magnitudes shouldn't be different.
Just make sure, (all of you), to set Gamma to "1" ;)
Last edited by Kram1032 on Wed Aug 15, 2007 2:19 am, edited 1 time in total.

User avatar
CoolColJ
Posts: 1734
Joined: Mon Jun 25, 2007 1:47 pm

Post by CoolColJ » Tue Aug 14, 2007 8:25 am

now we just need a gamma setting in our exporters.... :wink:

you see it's the little things like that that we need to be told, that would help the frustration level a lot

what is the gamma setting of 2.2 for in the first place? visual and texture colour?

User avatar
CoolColJ
Posts: 1734
Joined: Mon Jun 25, 2007 1:47 pm

Post by CoolColJ » Tue Aug 14, 2007 5:37 pm

well here is my test with v0.9 test 6

I used the SSS Indigo values directly from that attached PDF with the Indigo values a couple of pages back

I made 3 SKindigo .IGM files for Specular material with values for Milk, Coke and Apple Juice
put the liquids inside 3 glass containers with a 30cm base tapering up to 700cm high. Made sure the liquid objects had higher precedence than the glass container material

Modified the Skindigo exporter .rb file so it exports out the scattering coefficient RGB for Specular and Glossy Transparent materials with a gamma value of 1. I left the absorption RGB values at the original exporter 2.2 gamma values as they are supposed to work with the RGB sliders, and this will insure the the imported values from the .IGM files work as they should with SKindigo

exported out the scene and rendered it....looks pretty good to me 8)
made me go woah - realistic!

One thing I noticed was that with the RGB scattering value gamma set to 1, SSS actually developed pretty quickly now... perhaps there is no need to set max bounces down to 5 anymore...

How does it look values wise? maybe the values needed to be boosted up to account for the dilution as mentioned in the original topic PDF? Only milk seems to thin much at the top
Attachments
CCJ_SSSLiquids.JPG
Milk, Coke and Apple juice

rendered up to 200 samples with MLT mode
CCJ_SSSLiquids.JPG (35.34 KiB) Viewed 2800 times

mrCarnivore
Posts: 517
Joined: Sun Mar 04, 2007 6:20 am
Location: Stuttgart, Germany

Post by mrCarnivore » Tue Aug 14, 2007 7:25 pm

I think the sss values are a little too low. It is only clearly noticable in the milk, because milk almost exclusively uses sss (almost no absorbtion) the coke could be a little less black, don't know if it's too much absorbtion or too little sss. the applejuice looks fine, a little more sss wouldn't hurt, though.

Bottom line: more sss in every material should make it perfect. Don't know how much, though... :-(

User avatar
CoolColJ
Posts: 1734
Joined: Mon Jun 25, 2007 1:47 pm

Post by CoolColJ » Tue Aug 14, 2007 8:03 pm

ok, yeah dropping the gamma for Absorbtion RGB from the current 2.2 to 1.0 should fix the Coke

Multiplying the original scattering values by 23 should fix the density, as the article states the values were from mixing the liquids with 23V litres of water. Or should I multiply the Indigo numbers by 23?

User avatar
CoolColJ
Posts: 1734
Joined: Mon Jun 25, 2007 1:47 pm

Post by CoolColJ » Tue Aug 14, 2007 10:32 pm

did tw more renders - first with the same Indigo values from the PDF, but now with gamma of 1 for both absorption and scattering, phase_function is uniform

the colours look better now with the change of gamma from 2.2 to 1
but as you said the liquids look too diluted.

so in the 2nd render I doubled up the scattering values, since they are already x10 from the original numbers, then doubling up, makes them closer to the 23x increase to make up the original concentration

should I double up the absorption values as well????
Attachments
CCJ_SSSLiquids_allGamma1.JPG
Milk, Coke, Apple juice with original indigo values and RGB gamma of 1 for both
CCJ_SSSLiquids_allGamma1.JPG (37.78 KiB) Viewed 2760 times
CCJ_SSSLiquids_doubled scattering.JPG
doubled scattering values
CCJ_SSSLiquids_doubled scattering.JPG (37.63 KiB) Viewed 2760 times

User avatar
CoolColJ
Posts: 1734
Joined: Mon Jun 25, 2007 1:47 pm

Post by CoolColJ » Tue Aug 14, 2007 11:05 pm

ok with both values doubled up and gamma set at 1 - looks closer to the real things now

the Henyey-Greenstein Parameter "G" with RGB values will make up the rest of the look most probably, but Sketchup exporter doesn't support that yet, only uniform

I'll try and hand edit them in for milk to see
Attachments
CCJ_SSSLiquids_alldoubled up values.JPG
CCJ_SSSLiquids_alldoubled up values.JPG (39.03 KiB) Viewed 2753 times

User avatar
Kram1032
Posts: 6649
Joined: Tue Jan 23, 2007 3:55 am
Location: Austria near Vienna

Post by Kram1032 » Wed Aug 15, 2007 1:01 am

very nice mats, now :)

Milk looks too bluish, though, I'd say... add a bit yellow fat :P

mrCarnivore
Posts: 517
Joined: Sun Mar 04, 2007 6:20 am
Location: Stuttgart, Germany

Post by mrCarnivore » Wed Aug 15, 2007 2:14 am

The milk still doesn't look right. But that can be due to the low render time and depth. Try to render with better quality and let's see the result.

User avatar
carbon
Posts: 231
Joined: Mon Feb 19, 2007 11:26 am

Post by carbon » Wed Aug 15, 2007 2:19 am

check your white balance again :D

User avatar
Whaat
Developer
Posts: 1827
Joined: Fri Dec 22, 2006 6:15 am
Location: Canada
Contact:

Post by Whaat » Wed Aug 15, 2007 3:04 am

CoolColJ,

Your approach is still incorrect. Try these values:

Coke:
Absorption = 100.1 165.0 246.8
Scatter = 2.544 2.99 0
G-Spectrum = 0.965 0.972 0

Apple Juice:
Absorption = 12.957 23.741 52.184
Scatter = 2.568 3.105 4.127
G-Spectrum = 0.947 0.951 0.974

Milk:
Absorption = 1.533 4.6 19.93
Scatter = 18205 20383 22370
G-Spectrum = 0.75 0.714 0.681

Set Gamma to 1.0 for ALL of these values. The values from the PDF are for LOW CONCENTRATIONS. You have to multiply all values by 23/V or c2/c1 (see equation 8 from Jensen's paper).

The Milk will NOT render correctly in Indigo 0.9. I think this is due to a SSS bug. Try rendering with these values in Indigo 0.8 and you will see that they are correct.

User avatar
Whaat
Developer
Posts: 1827
Joined: Fri Dec 22, 2006 6:15 am
Location: Canada
Contact:

Post by Whaat » Wed Aug 15, 2007 6:46 am

OK,

I have made a new PDF of Scattering properties for use in Indigo 0.7 and up. Gamma must be set to 1.0 for these values to work!
Also included is the document in Excel format that allows you to input any concentration and it will calculate the Indigo properties.

Important: Indigo properties are now given for 100% concentration (except for the powders). Properties for the diluted media are given for reference.

Since SSS seems broken in Indigo 0.9, make sure you try the values in Indigo 0.8 before you complain to me that I got them all wrong. :P

Cheers,
Whaat
Attachments
Indigo Properties for Liquids.zip
(24.19 KiB) Downloaded 223 times

User avatar
Zom-B
1st Place 100
Posts: 4697
Joined: Tue Jul 04, 2006 4:18 pm
Location: ´'`\_(ò_Ó)_/´'`
Contact:

Post by Zom-B » Wed Aug 15, 2007 7:17 am

Great Work Whaat :!: :!: :!:

Such a PDF should be distributed directly with Indigo,
means in the manual!
polygonmanufaktur.de

User avatar
CoolColJ
Posts: 1734
Joined: Mon Jun 25, 2007 1:47 pm

Post by CoolColJ » Wed Aug 15, 2007 10:00 am

thanks!

Can you attach a text version of this PDF so we can copy and paste the values into the material much easier! :)
It's a bit of pain to to type it all out etc...


man I didn't think the scattering values would be that high?!


--

edit - it seems you can copy and paste text from a PDF anyway

ok will try some new renders with these values. BTW my white balance was set at D65 which is what you want for outdoor lighting :)

Post Reply
109 posts

Who is online

Users browsing this forum: Semrush [Bot] and 4 guests