<br><br><div class="gmail_quote">On Thu, Nov 17, 2011 at 11:42 AM, Brecht Van Lommel <span dir="ltr"><<a href="mailto:brechtvanlommel@pandora.be" target="_blank">brechtvanlommel@pandora.be</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
Hi,<br>
<div><br>
On Thu, Nov 17, 2011 at 8:05 PM, Lucas <<a href="mailto:wsacul@gmail.com" target="_blank">wsacul@gmail.com</a>> wrote:<br>
> I don't quite understand why the caustics doesn't work easily if the light<br>
> is passing through parallel glass planes- it seems like as many paths ought<br>
> to make it back from a diffuse bounce to the light source as they do in free<br>
> space, unless some shortcut/'cheating' is helping out with the free space<br>
> path, or the probabilistic path termination is really aggressive after a<br>
> single free space to glass transition. Can I get correct results (like the<br>
> light source passing through the glass ball<br>
> in <a href="http://www.cs.berkeley.edu/~sequin/CS184/TOPICS/GlobalIllumination/Gill_0.html" target="_blank">http://www.cs.berkeley.edu/~sequin/CS184/TOPICS/GlobalIllumination/Gill_0.html</a> and<br>
> projecting and interesting light pattern on the wall and floor) if I crank<br>
> up the render cycles (going to try it myself but it may take a while).<br>
<br>
</div>If Glass IOR is exactly 1, it could indeed do the same as it does for<br>
the Transparent. For other values, this requires a different<br>
algorithm. I think it's good to keep different sampling behavior for<br>
different nodes, suddenly doing something when IOR is 1 is not really<br>
predictable, and it would give strange results when texturing this<br>
input.<br></blockquote><div><br></div><div>No I wouldn't want special case code either. </div><div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
I guess you can get the results from that scene, if the light source<br>
is a mesh with an emission shader, it's just going to take a long time<br>
for the noise to clean up.<br>
<div><div></div><div><br>
Brecht.<br></div></div></blockquote><div><br></div><div><br></div><div>I did more rendering and now I can see the transparency is working as it ought to, I just didn't know it would take so many cycles to get this far:</div>
<div><br></div><div><a href="https://picasaweb.google.com/103190342755104432973/Blender3D#5676124095067668818">https://picasaweb.google.com/103190342755104432973/Blender3D#5676124095067668818</a></div><div><br></div><div>
Now I think I'll have to resort to more tricks to simulate this in less renders, maybe by setting a camera to capture light coming through a transparent object from a certain point of view and then re-emitting that light from the back of the camera. For that and other purposes it would be really useful to have a non-lambertian emission shader, so I could send light rays from a texture only in narrow lobes along the surface normal and use it as a projector (maybe have a 0 - 1.0 slider to scale inbetween pure lambertian and pure normal). If that doesn't exist maybe that's a good excuse to learn how to make it myself.</div>
<div><br></div><div>Thanks!</div><div><br></div><div>Lucas</div><div><br></div></div>