First try with SkIndigo

Headroom · Post by **Headroom** » Fri Feb 15, 2013 1:08 pm

Don't use a glossy transparent transmitter for that situation. While that is probably the most intuitive selection as it matches the material most closely, it also slows down the render process.
Unless you want to make a close- up shot of the lamp I'd use a diffuse transmitter.

lape · Post by **lape** » Fri Feb 15, 2013 7:56 pm

Do you mean by "transmitter" the glass surrounding the bulb (which uses a diffuse material)? Or the emitter itself?

cotty · Post by **cotty** » Fri Feb 15, 2013 8:23 pm

lape wrote:attached is a picture of the lamp. Inside bulb with emitter (cylinder), outside cylinder with glass material.

How thick is that sourrounding cylinder? It is not clear visible in that picture.

lape wrote:Do you mean by "transmitter" the glass surrounding the bulb (which uses a diffuse material)?

This one, I think.

lape · Post by **lape** » Fri Feb 15, 2013 8:51 pm

thickness of glass is 5mm. I will try a different material for the glass.
thank you!

Pibuz · Post by **Pibuz** » Fri Feb 15, 2013 10:48 pm

Hi lape (..btw: is it Lape or Iape with an i?

)
The gravel is really really cool, a great tool to use as an obj: good choice!

Don't worry about CTZn's splat filters: in SkIndigo we cannot control those parameters directly. At the moment they're sort of hidden and surely predefined. I recently asked Dale aka Whaat to add them as optional parameters, so that some more "advanced" users could fiddle a little with them, if they want to

Generally, frosted glass enclosing emitters isn't a real good choice, efficiency-wise. I know that the purists of unbiased will kill me now, but I always suggest to make the glasses blurry afterwards in photoshop, just to shorten rendering times if needed. If you have one or more good machines, the most correct choice would be to make the glasses frosty as they should be in real life, that is true. If cheating or not, it's now up to you

A general rule you should remember about modeling and rendering with SkIndigo is: SMALL GAPS. Always detach the objects one from each other (as it happens in real life: nothing actually sticks 100% to something else), mostly when you deal with glasses and metals. I tell you because I cannot see any gaps between the light's glass and its body.

Light coming out of the house? I'd suggest you use a BiDir MLT method for sure.

For night shots, I'd use am evening EXR, but balancing light layer's values with "normal" emitters is a mess, and takes a lot of trial and error, at least chez moi. I usually get to absurd values for EXR maps such as 3.000.000 lumens and 1 lux for the indoor lights (that happens in my last project). I don't know Indiro RT too much, so I fear I cannot help you any further..

Headroom · Post by **Headroom** » Sat Feb 16, 2013 12:31 am

Every time the light rays have to interact with matter the computer has to calculate something. So as materials get moor complex the interaction is more complex needing more complex calculations.
Purists occasionally lack a sense for practical approaches.

You can either use Pibuz's approach and try to do things in Photoshop or use a diffuse transmitter for the glass, not the bulb.

Keep the bulb simple, particularly as you cannot see it anyway, and give, it a diffuse black surface ( going for minimum interaction )and make it a black body emitter for starters.

lape · Post by **lape** » Sat Feb 16, 2013 9:01 am

Thank you Pibuz and Headroom for your explanations

I remodeled the lamps and changed materials, now it seems to work much better. But I'm still trying to get some lights coming out of the building. Whenever I use bidir the inner building through the window stays black, though I tried different materials for the glass... It only works with path tracing.

@Pibuz: it's Lape

Headroom · Post by **Headroom** » Sat Feb 16, 2013 9:32 am

Unless you expect to see reflections in the window glass you can remove it.
If you have not done so already, try to enable glass acceleration. This only helps for flat glass so it will not do anything for the glass around the lamp, but it will help for the window glass should you decide to keep it in the scene.

lape · Post by **lape** » Sat Feb 16, 2013 9:11 pm

Here is my try for the nightscene after 11h of rendering, about 8000spp.
-path tracing
-glass acceleration

looks quite bad with all these fireflies...

CTZn · Post by **CTZn** » Sat Feb 16, 2013 9:22 pm

Have you tried MLT ? Like Pibuz I would recommend bidirectional with MLT over single path tracing for this scene. It's different but quite an all-round in itself.

You can reduce fireflies by using an higher supersample factor, but again MLT will help a lot with that.

lape · Post by **lape** » Sat Feb 16, 2013 9:33 pm

Ok, next try is bidir with MLT. Seems I've chosen quite a complicated scene for my first steps with indigo.

Pibuz · Post by **Pibuz** » Sat Feb 16, 2013 9:58 pm

It's ordinary life

lape · Post by **lape** » Sat Feb 16, 2013 10:26 pm

Ok, bidir with MLT started and I get 7k/s! this is so damn slow... after 30min 11spp. This will need days to get free of noise!

All I want to achieve is a nightscene with some warm light inside the building. Perhaps the material of then windows panes is the wrong one? I used the normal glass for the window (specular) by hitting assign preset. I saw the attached picture in the forum made by Jeff, how is this done? Glass material? Render method? Environment? question upon questions....

Zom-B · Post by **Zom-B** » Sat Feb 16, 2013 10:58 pm

Moinsen Lape,

good old Farnsworth House

Regarding the render modes MLT has his strengths, but this scene should be best with BiDir Pathtracing with GlassAcceleration I think.
In most scenarios you'll end up using pure PT (mostly with GPU turned on) and BiDir PT.
Pure PT is best on Studio setups and outdoor scenarios where Light doesn't bounce trillions of times inside some buildings etc., BiDir PT helps connecting more complicated light Paths like Caustics.
In This case all the light from the inner of the house are caustics, not fancy, but light going through a medium and getting some directional changes...

To optimice this scene further, take care that your light emitters inside the lamps are as big as possible and have as few polygones as possible.
Avoide pure white colors (chairs) and keep RGB range at max of 204 (80%) also for colors since everything above is unrealistic and also reflects light so much that calculation needs longer.
Using BiDir mode, you raise its effectiveness by keeping the scene "small", by deactivating objects and areas not relevant to your rendering.

Its really nice to see somebody new here who is already advanced in 3D!
Welcome, keep showing your stuff and asking questions

lape · Post by **lape** » Sat Feb 16, 2013 11:33 pm

Hi Zoom-B,

thank you for your kind words

I will try to follow your advices. But regarding gpu acceleration, I get an error when I activate it, see attached. What happens here?