Ray Depth Help

[PhilBo]

I am rendering some glassware from my laboratory. I have black areas in my render that in other software indicate that I'm not shooting a large enough ray depth. In Yafray or Blender I have gotten these areas to look correctly by increasing the Ray depth, so I know it can be done with this model.

I have tried tweaking the <max_depth> parameter, but it doesn't appear to be changing anything. I started at 1000 and have tried 2000 all the way up to 10000.

Is there another parameter that I need to adjust at the same time? What am I missing?

Thanks for any help that can be given.

Phil

[TjGk]

Indigo is unbiased, so there is no ray depth option. I think that some of your glassware may be intersecting the table it's on or something like that.

[PhilBo]

So what does the max depth parameter do then?

I have a render that I have been working on for a long time. It is a rack of glass panels that are all about 1/2 inch apart. There is a black square that shows up after the 6th panel or so because the depth was not set deep enough. I could not get it to go away with Blender, but after I increased the depth high enough, Yafray was finally able to get past the 6th plate of glass and go through to the end (15 panels of glass.)

I have tried to render this setup with Indigo, and I get the same artifact. If I move the camera so that I am not looking through all 15 panes, then there is no issue.

It just seems that the light isn't getting all the way through all the panes and is making it so that after a certain number, the rest are simply black.

[PhilBo]

I figured it out!

I disabled the bidirectional setting (using Blender -> Indigo script) and now it works perfectly.

If I had 7 very thin cubes upright spaced 1/4 inch apart, they were all clear. If I added number 8, it would render black.

Hope this helps someone who struggles with something similar in the future.

[manitwo]

hi PhilBo

try to reenable bidir-pt and set "max_num_consec_rejections" up to "1000" in the infile.txt.
I make similar tests atm - i'll post them later

[manitwo]

I have done some test with the max_num_consec_rejections
option and 2 mirrors:







Anim gif:


A question to nick: what exactly is the difference between
"max_num_consec_rejections" and "max_depth"?
they behave similar with this scene - but they do not the same.

[Kosmokrator]

excellent test!!!

thnx

[eman7613]

which version was that? i was trying to pull off the same thing in .5 and i couldent get it to work (knowing me i could have easily faced the polly the wrong way but..)

[manitwo]

indigo 0.6t4

mirror:

Code: Select all

    <material>
        <name>mirror</name>
        <phong>
            <fresnel_scale>1</fresnel_scale>
            <diffuse>1 1 1</diffuse>
            <specular>1 1 1</specular>
            <exponent>1000000</exponent>
        </phong>
    </material>

[eman7613]

indigo 0.6t4

mirror:

Code: Select all

    <material>
        <name>mirror</name>
        <phong>
            <fresnel_scale>1</fresnel_scale>
            <diffuse>1 1 1</diffuse>
            <specular>1 1 1</specular>
            <exponent>1000000</exponent>
        </phong>
    </material>

thank you!

[PhilBo]

Thanks for the hard work and the mirror example. Unfortunately, it did not help my situation at all. It is exactly the same with 1000 or 10,000 max rejections.

Is there anything else that I should be changing in the material or in the scene to ensure that the camera can go through all 15 glass plates (besides turning off bidirectional?)

[Zom-B]

@ manitwo

nice test dude...

how about including this pictures together with some more detailed info
(If nick dropes some lines about your question) to the dokumentation?

Think about it

[manitwo]

Yes i also thought about that.
Should i change something ?
If you think it's ok i'll rerender this pics (render all with the same
amount of time) for a better comparison.
Also ... if someone is interested in i would share this simple scene.

[PhilBo]

Here is what I'm talking about.

Here is with Bidirectional on and 1000 max rejections


Notice the black area in the middle?

Here is the same scene with 10,000 max rejections


Notice that the black area didn't go away?

Here is the same scene with 1000 max rejections but with bidirectional set to false.


Now the black areas in the middle are gone and the glass looks correct.

Can anyone shed light on this?

[eman7613]

i could be wrong, but wouldent the first 3 examples be more corect in the real world? it seems taht you want something that is just transparent, not reflective.