Indigo 2.2.1

Zom-B · Post by **Zom-B** » Fri Aug 21, 2009 3:11 pm

I have a scene with lots of ray bounces (room with shiny floor + shinny things) that is some real pain in the a** for the 2.2.1 path tracing... :/

Should I upload something for you??

Post by **OnoSendai** » Fri Aug 21, 2009 3:28 pm

ZomB wrote:I have a scene with lots of ray bounces (room with shiny floor + shinny things) that is some real pain in the a** for the 2.2.1 path tracing... :/

Should I upload something for you??

Does it converge slower than 2.0 stable path tracing?
Or do you mean that MLT is faster in this case?

Zom-B · Post by **Zom-B** » Fri Aug 21, 2009 4:45 pm

To admit, I haven't tested on 2.0 stable with path tracing, but this one seems to be a MLT darling anyway.

Post by **OnoSendai** » Fri Aug 21, 2009 4:47 pm

I'll have another crack at a new hybrid algorithm sometime soon, to get the best of both worlds

Zom-B · Post by **Zom-B** » Fri Aug 21, 2009 5:12 pm

This would really Rock my world :D

Jambert · Post by **Jambert** » Fri Aug 21, 2009 8:33 pm

ZomB wrote:This would really Rock my world :D

My world too

pvbeeber · Post by **pvbeeber** » Sat Aug 22, 2009 6:02 am

Just a thought: rather than specifying the hybrid threshold, would it be possible to specify a stsrting threshold and then having Indigo iteratively converge on the optimal threshold by adjusting the value, measuring the change in samples per second, and then adjusting again accordingly (i.e. adjusting in the other direction if spp decreases, adjusting in the same direction if spp increases)?

Post by **OnoSendai** » Sat Aug 22, 2009 12:10 pm

There is no hybrid threshold currently, not at least to do with the hybrid we're talking about (MLT/QMC hybrid)

Pibuz · Post by **Pibuz** » Thu Sep 03, 2009 4:42 am

OK GUYS GREAT BUG.
I was rendering a scene, it's been cooking since two/three days.
I have igs, igi etc etc.
I had to stop the rendering, and ok.
Now I try to re-launch.
Everything loads fine, but towards the end Indigo fails to allocate memory.

The scene is exactly the same that has been rendering for 3 days giving NO errors.

Now, THIS is really annoying. Cos the scene was absolutely not completed, and I had to stop it from calculation to do my work. I was planning on re-launching it this evening and let it cook overnight, but now it seems that I cannot. DAMN.

So you tell me what's going on. And please find a way to solve this. I have to get on with this render.

Jambert · Post by **Jambert** » Thu Sep 03, 2009 5:06 am

Same bug here

Maybe find somebody to send a pigm to render with you. If you blend to file you can erase a bit the noise with calculating 2000spp ;)

I can't help you now, I have to go but if you share youre file and send me a PM I could have a look when I'll be back... I'm running with i7 and it's quite faster)

Post by **fused** » Thu Sep 03, 2009 5:06 am

try to reboot and/or kill other memory intensive apps

Pibuz · Post by **Pibuz** » Thu Sep 03, 2009 5:28 am

Thank you Jambert for your kind offer. But I really think this is something that NEEDS to be fixed.

Also because there is no reasonable cause which explains why this is happening.

Fused, Ive closed ALL possible RAM consuming applications. Thank you BTW. I'll try the reboot option.

PureSpider · Post by **PureSpider** » Thu Sep 03, 2009 5:43 am

I think it's happening because the IGI file also needs some RAM that isn't used when rendering from scratch... I may be wrong here, just asumptions

Pibuz · Post by **Pibuz** » Thu Sep 03, 2009 7:39 pm

I don't know either..

I'm just waiting for some anwers from the pros here..

BTW, I was trying a great wax material downloaded from the database, and I came up with this idea: should be possible to somehow link the mat prevew directly to the igm or pigm file, so that when it is loaded into the modeling application we automatically have a clean mat preview image (the one posted in the database by the mat contributor)?

PureSpider · Post by **PureSpider** » Thu Sep 03, 2009 8:13 pm

Thought about that idea the other day when rendering another mat preview in Cindigo, too.
+1

Indigo 2.2.1

Re: Indigo 2.2.1

Re: Indigo 2.2.1

Re: Indigo 2.2.1

Re: Indigo 2.2.1

Re: Indigo 2.2.1

Re: Indigo 2.2.1

Re: Indigo 2.2.1

Re: Indigo 2.2.1

Re: Indigo 2.2.1

Re: Indigo 2.2.1

Re: Indigo 2.2.1

Re: Indigo 2.2.1

Re: Indigo 2.2.1

Re: Indigo 2.2.1

Re: Indigo 2.2.1

Who is online