Page 1 of 1
[REQ] Random Seed for Render-Q Files ** in 3.6.x **
Posted: Sat Jun 16, 2012 8:12 pm
by Zom-B
Since I do quite some animations in Indigo I would be glad to see a option to have a new random seed for each file rendered by the Render-Q.
At the moment Indigo only does a random seed for each instance of Indigo started, and keeps if for all files opened and rendered in that.
The result is that noise stays quite static over the image for animations and is very ineffective to get rid of by temporal denoisers that check multiple frames and identify noise by rapid pixel changes from frame to frame...
So please create some IGS tag like <random seed>true</random_seed>
Re: [REQ] Random Seed for Render-Q Files
Posted: Fri Jan 18, 2013 10:19 pm
by Zom-B
bump -.-
Re: [REQ] Random Seed for Render-Q Files
Posted: Fri Jan 18, 2013 10:23 pm
by fused
Zom-B wrote:bump -.-
Hi Zom-B,
we did introduce a per item seed in 3.6.x.
It was either that or a global random seed option and we decided in favor of the per item seed.
Re: [REQ] Random Seed for Render-Q Files
Posted: Fri Jan 18, 2013 10:37 pm
by Zom-B
Seems it simply wasn't mentioned in the Changelog, but after checking, I found a seed value in the Render Q settings of the GUI
Now only some Exporter support is needed for that feature

Re: [REQ] Random Seed for Render-Q Files
Posted: Fri Jan 18, 2013 10:40 pm
by fused
Zom-B wrote:Seems it simply wasn't mentioned in the Changelog, but after checking, I found a seed value in the Render Q settings of the GUI
Now only some Exporter support is needed for that feature

I am pretty sure you were given some version with some kind of support for it
For everyone else, sorry to keep you waiting. Will be back to doing exporter work very soon

Re: [REQ] Random Seed for Render-Q Files
Posted: Sat Jan 19, 2013 4:13 am
by CTZn
Zom-B wrote:Seems it simply wasn't mentioned in the Changelog
Nor elsewhere :/
By the way did I make it up or were the
*_spectrum tags rendered obsolete ? Got rid of'em.
Re: [REQ] Random Seed for Render-Q Files ** in 3.6.x **
Posted: Fri Jul 12, 2013 4:46 pm
by CTZn
Is there a similar seed tag for the render_settings ?
Re: [REQ] Random Seed for Render-Q Files ** in 3.6.x **
Posted: Fri Jul 12, 2013 6:50 pm
by Zom-B
afaik its only been a command line attribute since day 1
Code: Select all
START /B /LOW indigo.exe -n wm -p 7100 -seed 1234
best way is now to create a Q file for animations now, define seed there and load it in Indigo after export is done.
Re: [REQ] Random Seed for Render-Q Files ** in 3.6.x **
Posted: Sat Jul 13, 2013 1:07 am
by CTZn
The command line it will be then, thank you Arthur.
Re: [REQ] Random Seed for Render-Q Files ** in 3.6.x **
Posted: Sat Jul 13, 2013 1:21 am
by CTZn
What's Indigo's default seed ? If i'm correct by default Indigo is using the same seed isn't it ? (I set it to 1337).
The exporter will be shooting by default a random seed in the range 0-10000 for each render, can this be improved somehow ? Is this even a good default behaviour ?
Re: [REQ] Random Seed for Render-Q Files ** in 3.6.x **
Posted: Sat Jul 13, 2013 2:17 am
by Zom-B
indigo default is 0, and each core gets a +1 seed, check the log after render started.
your range in random seeds seems enough... temporal denoiser usually don't take more then 5 frames into account for denoising, as long as these 5 frames in a row have not the same seed it will work nicely

Re: [REQ] Random Seed for Render-Q Files ** in 3.6.x **
Posted: Sat Jul 13, 2013 3:44 am
by CTZn
thank you, much appreciated !
Re: [REQ] Random Seed for Render-Q Files ** in 3.6.x **
Posted: Sun Jul 14, 2013 2:42 am
by pixie
Zom-B wrote:indigo default is 0, and each core gets a +1 seed, check the log after render started.
your range in random seeds seems enough... temporal denoiser usually don't take more then 5 frames into account for denoising, as long as these 5 frames in a row have not the same seed
it will work nicely 
Video or it didn't happen!

Re: [REQ] Random Seed for Render-Q Files ** in 3.6.x **
Posted: Sun Jul 14, 2013 3:33 am
by Zom-B
pixie wrote:Video or it didn't happen!

The chance to have at least two same seeds in a following row of 5 random picks out of 10000 should be quit easy to calculate, but my math skill is actually non existing -.-'