OR since v3.00 is caustics later than v2.25 at same S/px

[nuno1980]

Hi, guys.

Dragon invisible test at resolution screen of 1002x602:
-- v3.01.1b:

-- v2.25:

I get caustics later using v3.00~v3.01.1b at same samples per pixel (S/px or spp).

PMC default settings except few things - see right of 1st image

Specular material for dragon default settings except:
- reflection and transmission 1.000
- índex of refraction 1.000


I hope the caustics will be better/earlier than v2.2x or this issue will be fixed to undo as v2.2x.

[abstrax]

Thanks for the report. We will look into it.

[roeland]

Hi,

Can you upload the scene you're rendreing?

And also, what effect does this have on the scenes you're rendering? Usually, if an object like a window is supposed to be invisible or transparent to light rays (for instance windows) the material will have the "fake shadows" option enabled. If you see caustics in the scene from an object, this object will have an IOR different from 1, and usually you set the caustic blur in the kernel settings to a small value, like 0.01, so the caustic will be rendered faster (but slightly blurred).

--
Roeland

[nuno1980]

Hi,

Can you upload the scene you're rendreing?

And also, what effect does this have on the scenes you're rendering? Usually, if an object like a window is supposed to be invisible or transparent to light rays (for instance windows) the material will have the "fake shadows" option enabled. If you see caustics in the scene from an object, this object will have an IOR different from 1, and usually you set the caustic blur in the kernel settings to a small value, like 0.01, so the caustic will be rendered faster (but slightly blurred).

--
Roeland

Yes, download orbx -

dragon invisible.orbx

(32.35 MiB) Downloaded 417 times

Wedding glass test:
-- v2.25

-- v3.01.1b

PMC default settings except:
- 151705 spp
- 50-diffuse and 52-specular maxdepth
- 10k-reject
- parallelism 1
- caustic blur 0.000

v2.25 is better caustic = more realism than 3.01.1b (look 3rd ring on right) but v3.0x is ~2x faster than v2.25.

See caustics "cleaned" for v2.2x only.

[nuno1980]

I've edited to change 2nd image in OP.

v2.25 at PMC kernel is normal.

I want unlimited samples per pixel in next release!! - I'm not worried despite long rendertime as days or as some weeks. I don't care animations but yes only few images about any scene.

[roeland]

What is the result with a caustic blur setting of 0.001 or 0.01?

--
Roeland

[nuno1980]

What is the result with a caustic blur setting of 0.001 or 0.01?

--
Roeland

I don't render, sorry but unfortunately, (very) hot weather during many days and it continues until saturday. Therefore, my room is 29 ºC in Porto (Portugal) and it was 31 ºC on yesterday.

--> Please you should fix PMC unoptimized (= appear caustics later) and increase from 256k to unlimited samples per pixel in next release after new v3.03.3 (the issue as my thread isn't listed yet).
Therefore, I keep to use v2.25.

[nuno1980]

Wedding test - caustic blur 0.001:
-- v3.03.3

-- v2.25

v2.25 continues to be better and more visible caustic (3rd ring on right). I prefer v2.25 until the issue fixed.

[nuno1980]

Look new 2 images

[nuno1980]

Ah... We get caustics later in v3.00~v3.03.3 due to maybe wrong number of rejects (10k).

IMPORTANT! Do not forget to increase maximum spp from 256k to unlimited (as 0=zero) in next release with this issue fixed.