Cut Crystal Tests

zsouthboy · Post by **zsouthboy** » Fri Oct 12, 2007 1:57 am

CoolColJ wrote:can't the regular AMD/Phenom desktop chips run in dual CPU?

That's the way I would go myself - save some bucks

Nope. No dual or quad-cpu action, from either Intel or AMD, without paying for a "2" or "4" capable cpus (Xeon or Opteron)

WytRaven · Post by **WytRaven** » Fri Oct 12, 2007 3:35 am

I just compared my Cauchy curve for BK7 against the graph on wikipedia. My curve behaves the same vs the measured data as the one on wikipedia. I am now 100% sure I have the A and B calcs perfect now. I've only changed one thing since last post of the spreadsheet and that was switching the 'n' in the A calcs from the median of nD and nd to the mean of nD and nd (which is what I originally intended but had a brain fart

).

EDIT: Decided to go back to meassured nd for the A calcs, made more sense. Updated link to version 13.

Link removed, new link further down.

WytRaven · Post by **WytRaven** » Fri Oct 12, 2007 4:47 pm

Render update:

bidirectional: 1
width: 1920
height: 1200
super_sample_factor: 2
hybrid: 0
metropolis: 1
num_threads: 2
large_mutation_prob: 0.40000
max_change: 0.01000
max_num_consec_rejections: 1500
maxdepth: 15000
ray_origin_nudge_distance: 0.00001

num_samples: 25915570000.00000
render time elapsed: 601386.28205
samples_per_pixel: 11248.07726
samples_per_sec: 43093.05146

It's actually getting pretty close to convergence now

zsouthboy · Post by **zsouthboy** » Sat Oct 13, 2007 12:28 am

Well, that beats me for longest render - longest i've gone is a week (then again, no cauchy in that scene)

WytRaven · Post by **WytRaven** » Sat Oct 13, 2007 1:06 am

It's really facinating watching this scene evolve. Every time I think it's moving toward convergence indigo finds some new detail to start working on. Its over 175hours now and all of a sudden some brand new caustics just started showing up

BbB · Post by **BbB** » Sat Oct 13, 2007 1:07 am

You guys are mad

WytRaven · Post by **WytRaven** » Sat Oct 13, 2007 1:13 am

I'm actually really peeved that I din't put more effort into the scene setup now. What was supposed to be a simple test has grown into aa render marathon and unfortunately the end result whilst being technically cool will make for a flawed render due to the crapy scene.

I promise my next 200 odd hour render will be a good picture as well as being technically cool

WytRaven · Post by **WytRaven** » Sat Oct 13, 2007 3:09 am

Here's a pretty comparison of my calced values for BK-7. The comparison from wikipedia shown first:

Wikipedia values:
A: 1.5046
B: 0.00420

My Values:
A: 1.50457412559239
B: 0.00422072479206567

Considering that wikipedia doesn't show the values they calculated their A and B from it's no surprise that the numbers aren't eactly the same but you can see from the graph that the curves are behaving the same. I think it's safe to say from this that the method I am using is correct.

Stur · Post by **Stur** » Sat Oct 13, 2007 4:01 am

WytRaven wrote:It's really facinating watching this scene evolve. Every time I think it's moving toward convergence indigo finds some new detail to start working on. Its over 175hours now and all of a sudden some brand new caustics just started showing up

All of a sudden you're going to have an electricity cut while Indigo is saving and your .igi will be fucked up

You should backup your .igi sometimes if you don't already do

WytRaven · Post by **WytRaven** » Sat Oct 13, 2007 4:23 am

Dude! Don't say things like that *touches wood*

Kram1032 · Post by **Kram1032** » Sat Oct 13, 2007 10:31 am

But he's right xD
you should copy your igi into an other folder, so that it's saved, there

Not all that long, to finish your 200, any more

CoolColJ · Post by **CoolColJ** » Sat Oct 13, 2007 11:43 am

maybe you should make sure the pic actually looks the way you want at a smalll res before you render out a massive one

WytRaven · Post by **WytRaven** » Sat Oct 13, 2007 8:43 pm

As per request Ono, n vs λ^-2 for BK7 using my calculated A and B coefficients.

WytRaven · Post by **WytRaven** » Sat Oct 13, 2007 8:55 pm

BK7 meassured n vs λ^-2. Meassured n wavelengths are from 1060nm -435.8343nm

Kram1032 · Post by **Kram1032** » Sat Oct 13, 2007 9:06 pm

why are the wavelengths in 1/λ² ?
What about a 3D-net comparison? I think, Excel has that, too, doesn't it?

then, you can compare 3 datas at once ^^