[CONFIRMED] Epsilon value

[Deus]

Indigoworld had a beach issue as seen in this image.

http://www2.indigorenderer.com/joomla/i ... emId=16574

The reason is epsilon related because when i scaled up everything a factor 10 the issue dissapeared. My original model was about the size of 1x1x1 unit.

That leads to the followup question. What is the optimal size for a model into indigo for numerical precision reasons. What is the maximum size?

/ Deus

[Kram1032]

1x1x1?
Don't forget, to hold in the size!
1 unit = 1 meter

[Deus]

Yeah indigoworld is a small project

I knew for a fact that indigo had a max size (300m iirc) so "actual" real size was never an option. I took unit size because it was conventient.

[Kram1032]

this issue is solved!

maximum size is (more or less) infinite, now
before, just the SKY (and a bg-lighting) had this problem, but now, it just rezises to 100x the world size

[Deus]

Ok then i will start working in real world units.

Still this IS an issue for people that would like to model real small things.

Besides I cant use milk material for fog no more.. It will be too thick

[manitwo]

wouldn't real-world units (for your scenes - which are big) result in problems with floating-point precision?
i have not enough technical-knowhow to say that for sure, but afaik scenes shouldn't be too small(mm) and also not too big(km).

[Deus]

Actually it might. There are always issues.

What happened with the epsilon it is there to not cause self relection/shadowing/intersection but in my case that lead to that the ray hit the "close" beach surface. Same thing can hold true on large scenes. That rays that hit the water would self intersect.