Indigo Renderer Forum

somewhere between "glass cylinder 10, alcohol-water column 11, glass bulbs 12" you need some "air" over the alcohol-water otherwise it gets filled by the "glass cylinder"

Going back to the first render where I at least had some color, I had an "AH!" moment. I think I'm beginning to grasp precedence. If the mesh in question doesn't actually reside inside another mesh, set it's precidence to the same as other objects at it's level, or even lower! So in my case, the bulbs being glass with thickness, the bulbs and the liquid do not occupy the same space. That was why they keep coming out near-black. D'oh. If I don't give the bulbs an inner wall of faces, then I'd have to specify a higher precedence.

Hey CCJ, re: the speed of glass renders, nope! Not for me anyway, a 320x240 vs 3200x2400 render the same speed on this particular scene - 37k/sec. (2x2.2GHz 4200+, 64-bit linux, wine-0.9.43... it's just a lot of transparencies to deal with.)
Time elapsed: 10h, 44m, 25s
Done 1450050000.00000 samples (3020.93750 samples per pixel)
37502.43017 samples / second (26.66494 micro-seconds / sample)

The color saturations are wrong, (the bottom yellow should be green and the middle red should be orangey) but I think I'm on the right track, I just wish I could get the glass reflections to be as strong as the first surface they pass through. <rgb>0 0 0 </rgb> should not be attenuating the light at all, only bending it due to IoR... but I guess there's more to it than that.

(don't ask re: the UV grid, I have no idea, it's cleanly unwrapped. Such is my rendering life. )

zuegs wrote:somewhere between "glass cylinder 10, alcohol-water column 11, glass bulbs 12" you need some "air" over the alcohol-water otherwise it gets filled by the "glass cylinder"

Good point!

edit: No, I take that back. That would be silly unless things aren't physically modelled. If it were just a solid column with another column inside it with different material, then yes you would need to "create" air. But this is entirely physically modelled. Every bit has a front and back and voids are air... or air is, uhm, air.

Oh well. I guess I'm done. I'm bored of trying to make it look right. I think the render server would be put to better use as an ET:QW server. Game on.

Thanks anyways.

//closed

Heavily Tessellated wrote: Hey CCJ, re: the speed of glass renders, nope! Not for me anyway, a 320x240 vs 3200x2400 render the same speed on this particular scene - 37k/sec. (2x2.2GHz 4200+, 64-bit linux, wine-0.9.43... it's just a lot of transparencies to deal with.)
Time elapsed: 10h, 44m, 25s
Done 1450050000.00000 samples (3020.93750 samples per pixel)
37502.43017 samples / second (26.66494 micro-seconds / sample)

man that makes no sense. there are physically less pixels to be rendered!
sample per second pixel rate won't change but the samples per pixel rate will be much higher

26.66494 micro-seconds is fast, you don't know what slow is....my glass heavy scene is rendering at 180+ microseconds per sample....
rendering for 15 hours now.. now at 2000+ samples, and that's at 640x480....


---

re reflections - maybe your light source and environment is the problem

Heavily Tessellated wrote:

zuegs wrote:somewhere between "glass cylinder 10, alcohol-water column 11, glass bulbs 12" you need some "air" over the alcohol-water otherwise it gets filled by the "glass cylinder"

Good point!
edit: No, I take that back. That would be silly unless things aren't physically modelled. If it were just a solid column with another column inside it with different material, then yes you would need to "create" air. But this is entirely physically modelled. Every bit has a front and back and voids are air... or air is, uhm, air.

yes, if physically modelled you don't need air... only for non-boolean modelling... i thought you done this for your second render, sorry.

Keep on with this work, it's very NICE