[hdesbois]

Here it is :

The thing with colored dots is the regent diamond model. Default diamond mat. Magnifying glass dielectric is ND 1.75 abbe 200 with a faint green tint.
One emitter on the left. The book is inside a room with sunlight entering trough a window on the right (no dielectrics). Other materials are diffuses or plastics.
Things are getting worse as time increases.
HD

[tom]

WOW this is the most beautiful bug I see...

[Intuition]

This is a cool idea for a render.

Makes you really see how unbiased MW really is.

Bugs notwithstanding of course.

[j_petrucci]

is maybe some conflict between emitter and sunlight?

I hope developers take a look at our (exaustive!) tests, just because we are pushing Maxwell to the limit, but we are also helping them finding bugs!

[hdesbois]

I'm doing a rendering without the magnifying glass, with same setup : so far (level 14) seems ok. I will let it cook a few more hours and run more tests.
This one :
http://www.maxwellrender.com/forum/viewtopic.php?t=3723
might be the same bug. having several different dielectrics in the same scene might be the problem...
HD

[spekoun]

I have black dots in every my render. Except some simple tests...

[Siegfried Kircheis]

This is a kind of test that I wanna do for a very long time, align a series of lens with different geometry and design, and see what I can see through the lenses.
Bounces ~ more than 50, maybe....time....

[hdesbois]

And if you remove the magnifying glass :



Diamond looks better than with alpha, although it lacks some caustics, even after 72 hours lvl20 render. I thought it would look sharper too, at fstop : 22.

HD

[hdesbois]

And now with beta 1.2.1 :

better, but not that good after 17 hours, lvl 16.
I'll let it cook overnight (if I have no overheat crash) but I think their is room for improvment with these dielectrics.
HD

[psanitra]

Now it`s not so bad, just the dispersion ...

[Aldaryn]

Nice scene idea, imo a dielectric behind a dielectric is a hard case to solve, but, this image worth the rendering time.

And the caustics under the magnifying glass looks almost noise free, and this is good news.

[tom]

sure...this must be about trapped bounces..
so it may take lots of time but works all right.

[hdesbois]

49 hours, and it's rather getting worse :

I suspect there is a problem here
This seems related to the color dots of the previous version : no more dots, but the colored noise is just at the same place. It seems the dots have not totally disapeared : they have shrinked, becoming really tiny.

HD

[Aldaryn]

What is the abbe value of the lens? Or only the diamond dispersing the light?

[hdesbois]

lens abbe is 200, and ND 1.75 : this is a 20 cm focal length biconvex spherical lens, optically correct except for dispersion. Diamond is default diamond mat value.
HD