Rerendered at 800x600 with slightly improved lighting.

Sorry for starting a new thread but Thunderbird insisted on
sorting my original "Rainbow" post into a thread from 2008
using the same title.

While testing I also had the curious effect that I got additional
caustics from the glass when *removing* the window pane. I really don't
quite understand how this can affect anything, the window pane uses
photons {pass_through refraction off reflection off collect off}.
The water glass is the only photon target in the scene.

Post a reply to this message

Attachments:
Download 'rainbow.png' (506 KB)

Preview of image 'rainbow.png'

Christian Froeschlin wrote:

> Rerendered at 800x600 with slightly improved lighting.
> 
> Sorry for starting a new thread but Thunderbird insisted on
> sorting my original "Rainbow" post into a thread from 2008
> using the same title.
> 
> While testing I also had the curious effect that I got additional
> caustics from the glass when *removing* the window pane. I really don't
> quite understand how this can affect anything, the window pane uses
> photons {pass_through refraction off reflection off collect off}.
> The water glass is the only photon target in the scene.

I've had something similar with photons, blobs, and a certain level of
transmit on the textures on the blobs. Pov'd go into an infinite loop while
casting photons or render a completely black scene. Somebody on here said
it was a known bug?

Maybe you are seeing a manifestation of it. As far as I know this is not
fixed in 3.7 beta either.
-- 
Stefan Viljoen

Post a reply to this message

> Rerendered at 800x600 with slightly improved lighting.
> 
> Sorry for starting a new thread but Thunderbird insisted on
> sorting my original "Rainbow" post into a thread from 2008
> using the same title.
> 
> While testing I also had the curious effect that I got additional
> caustics from the glass when *removing* the window pane. I really don't
> quite understand how this can affect anything, the window pane uses
> photons {pass_through refraction off reflection off collect off}.
> The water glass is the only photon target in the scene.
> 
> 
> ------------------------------------------------------------------------
> 
It could be a max_trace_level isue. Try adding 2 to it and use the glass 
pane.



Alain

Post a reply to this message

Christian Froeschlin schrieb:

> While testing I also had the curious effect that I got additional
> caustics from the glass when *removing* the window pane. I really don't
> quite understand how this can affect anything, the window pane uses
> photons {pass_through refraction off reflection off collect off}.
> The water glass is the only photon target in the scene.

What's your max_trace_level (for photons) set to?

Post a reply to this message

Stefan Viljoen schrieb:

> I've had something similar with photons, blobs, and a certain level of
> transmit on the textures on the blobs. Pov'd go into an infinite loop while
> casting photons or render a completely black scene. Somebody on here said
> it was a known bug?

No, that one is specifically related to blobs.

Post a reply to this message

clipka wrote:

> Stefan Viljoen schrieb:
> 
>> I've had something similar with photons, blobs, and a certain level of
>> transmit on the textures on the blobs. Pov'd go into an infinite loop
>> while casting photons or render a completely black scene. Somebody on
>> here said it was a known bug?
> 
> No, that one is specifically related to blobs.

Oh!

Ok, I thought it had to do with photons? Must have been mistaken then.

-- 
Stefan Viljoen

Post a reply to this message

Alain wrote:

> It could be a max_trace_level isue. Try adding 2 to it and use the glass 
> pane.

I had max_trace_level set to 20 with no special value for photons,
as far as I understand photons should then also use 20. I tried 22
both in global and photon block but it didn't make any difference.

I attached two test renders, the version without pane shows a
bright line to the right of the rainbow. Also, it gives wildly
different photon statistics (count is set to 20.000, with the
water using 7 dispersion_samples).

// With window pane

Number of photons shot:           39761
Surface photons stored:          197605

// Without window pane

Number of photons shot:           31731
Surface photons stored:           64749


I noticed the bright line goes away if I change the light source
from "parallel" to a normal point light. The statistics remain
similar however.

Post a reply to this message

Attachments:
Download 'rainbow_pane.png' (86 KB) Download 'rainbow_nopane.png' (87 KB)

Preview of image 'rainbow_pane.png'

Preview of image 'rainbow_nopane.png'

> Alain wrote:
> 
>> It could be a max_trace_level isue. Try adding 2 to it and use the 
>> glass pane.
> 
> I had max_trace_level set to 20 with no special value for photons,
> as far as I understand photons should then also use 20. I tried 22
> both in global and photon block but it didn't make any difference.
> 
> I attached two test renders, the version without pane shows a
> bright line to the right of the rainbow. Also, it gives wildly
> different photon statistics (count is set to 20.000, with the
> water using 7 dispersion_samples).
> 
> // With window pane
> 
> Number of photons shot:           39761
> Surface photons stored:          197605
> 
> // Without window pane
> 
> Number of photons shot:           31731
> Surface photons stored:           64749
> 
> 
> I noticed the bright line goes away if I change the light source
> from "parallel" to a normal point light. The statistics remain
> similar however.
> 
> 
> ------------------------------------------------------------------------
> 
> 
> ------------------------------------------------------------------------
> 
The window pane have a frame that cast a shadow near the base of the 
glass. The caustics in that area are still there without, but are less 
visible on a brighter surface. Try removing the glass itself, but leave 
it's frame.
Even with your settings, the window can still affect the photons coming 
from another object. At least, that's what the documentation seems to imply.
With the window pane, the rainbow is brighter and cover a whider arc.

Without the pane, the bright line that you mention is very hard to see 
here, even with the brightness pushed to the max. I also need to sellect 
the two images and unsellect the window to be able to see it. Tested on 
two different monitors.



Alain

Post a reply to this message

Alain wrote:

> The window pane have a frame that cast a shadow near the base of the 
> glass. The caustics in that area are still there without, but are less 
> visible on a brighter surface. Try removing the glass itself, but leave 
> it's frame.

Tried rendering with frame but without glass but the result on
the floor seems no different than without both. I wasn't worried
about the caustics near the glass itself.

> Even with your settings, the window can still affect the photons coming 
> from another object. At least, that's what the documentation seems to 
> imply.

Which part of the documentation? I must say I didn't quite understand
how the amount of photons is determined. From the algorithm I thought it
should mostly depend on the target object. As far as I understand, an
object with pass_through should not affect photons before they hit the
target object. And those photons which get refracted on the glass so
much they actually return to the window shouldn't affect caustics
on the floor either way.

> Without the pane, the bright line that you mention is very hard to see 
> here, even with the brightness pushed to the max. I also need to sellect 
> the two images and unsellect the window to be able to see it. Tested on 
> two different monitors.

Hmm ... I have a 5 year old TFT screen which did come with a color
calibration tool but I couldn't actually get that to *do* anything,
it goes through the calibration procedure and afterwards everything
looks exactly as before. So maybe my gamma is off the scale as I
see it distinctly (I certainly don't have correct gamma as far
as all the recent eye squinting is concerned).

The actual difference in intensity seems to be about 5 levels of 255.
The attached render has radiosity disabled so the contrast is better.
You can also see the line is actually composed of a very few splotchy
photons. Probably some off photons coming exactly off the rounded
edge of the glass or something like that.

Post a reply to this message

Attachments:
Download 'rainbow_frame_without_pane.png' (50 KB)

Preview of image 'rainbow_frame_without_pane.png'

Christian Froeschlin schrieb:

> Which part of the documentation? I must say I didn't quite understand
> how the amount of photons is determined. From the algorithm I thought it
> should mostly depend on the target object. As far as I understand, an
> object with pass_through should not affect photons before they hit the
> target object. And those photons which get refracted on the glass so
> much they actually return to the window shouldn't affect caustics
> on the floor either way.

Well, that's not the /exact/ truth, as far as I understand the source 
code. Instead, pass_through prevents an object from /totally/ blocking 
photons before they hit their target (which would otherwise be the 
case); however, other effects - such as refraction - still seem to be 
applied.


> The actual difference in intensity seems to be about 5 levels of 255.
> The attached render has radiosity disabled so the contrast is better.
> You can also see the line is actually composed of a very few splotchy
> photons. Probably some off photons coming exactly off the rounded
> edge of the glass or something like that.

You might try placing the glass half inside the wall, so that you get a 
"cross-cut" of the photons path on the wall, to find out where that 
stripe actually comes from... if you can decipher what I mean.

Post a reply to this message