Altus denoiser

[super gnu]

Another try. 20 minutes per seed + 5 minutes to denoise. Full size png here http://rghost.ru/8s9PQcb9r
[ATTACH=CONFIG]26736[/ATTACH][ATTACH=CONFIG]26737[/ATTACH][ATTACH=CONFIG]26738[/ATTACH]

btw i get dire warnings when i download your rar file.. chrome refuses to open it, insisting its malicious.

[Paul Oblomov]

Oh, feel free to scan in with antivirus But there's none inside.

Btw the blotchy result is a surpise to me too. Maybe, they should introduce some bias to noise removal reoutine. So it won't be completely denoised. Tho, 10minutes in PS makes even this result quite promising (blending two pic with 50% and then mix it altus result with some masking applied ).

[pingus]

Thanks Lele for script !

Works pretty well, even on a furry character (can't upload TGA :s).
It make the fur a bit soft for the moment, but this king of technology is quite important and promising.

I need to test it on an little animation, but having some problems customizing the .bat file.
Are the render elements required ?
What are the rules for namings files ? baseName_b0.RBG_color.XXXX.ext ?


Do you guyz have any information on a commercial release & pricing ? thanks

[^Lele^]

Oh, feel free to scan in with antivirus But there's none inside.

Btw the blotchy result is a surpise to me too. Maybe, they should introduce some bias to noise removal reoutine. So it won't be completely denoised. Tho, 10minutes in PS makes even this result quite promising (blending two pic with 50% and then mix it altus result with some masking applied ).

[I][B]altus.exe --help
[/B][/I]
 -r [ --radius ] arg (=10) Filter radius. default value is 10.
 --kc_1 arg (=0.449999988) This value is sensitive to image detail, but less
                           aggressive towards noise. Default value is 0.45.
 --kc_2 arg (=0.449999988) This value is the same as kc_1, but for a second
                           pass. Default value is 0.45.
 --kc_3 arg (=1e+010)      This value determines aggressiveness towards noise,
                           but may cause loss of detail. Default value is
                           1e10f.

There ARE parameters, which i chose not to expose.
Btw, i'm waiting for some feedback, and then i'll post the new version of the script, which has max passes file saving, (simplistic) BB submission, and animation support (and that was quite the pain to chase.).

[Paul Oblomov]

Oh. I thought this is the rgb sensitivity... Gonna try to refilter.

Also - gonna try volumetrics. How they will handle that.

[innobright]

[I][B]altus.exe --help
[/B][/I]
 -r [ --radius ] arg (=10) Filter radius. default value is 10.
 --kc_1 arg (=0.449999988) This value is sensitive to image detail, but less
                           aggressive towards noise. Default value is 0.45.
 --kc_2 arg (=0.449999988) This value is the same as kc_1, but for a second
                           pass. Default value is 0.45.
 --kc_3 arg (=1e+010)      This value determines aggressiveness towards noise,
                           but may cause loss of detail. Default value is
                           1e10f.

There ARE parameters, which i chose not to expose.
Btw, i'm waiting for some feedback, and then i'll post the new version of the script, which has max passes file saving, (simplistic) BB submission, and animation support (and that was quite the pain to chase.).

The first two values determine the aggressiveness of the primary filter, and the last setting is how aggressive the overall result will be. We recommend to only change kc_1 and kc_2 between a range of 0 and 1 as these settings will greatly effect the filtering process as it will change how the first and second passes are weighted against each other. kc_3 can be experimented with, but is a very sensitive value as it determines the overall aggressiveness of the filtering process. We are working on fixing the help and some user experience issues currently as the k_red k_grn k_blu no longer work and are not actually exposed inputs at this point.

[^Lele^]

Here's the latest version of the sample script.
It supports animation (MINUS nth frames), Max file per seed pass plus altus batch file(s) saving without rendering, Backburner submission (VERY simple. don't use that if you're in a serious production environment.).
In general, i would very much avoid gaps in sequences.
Although supported, there MAY be issues with sub-chunks of 1 frame of length (as they remain treated as animations, but start and end are coinciding).
I'm referring to the very bottom animation option, where this is more likely to happen: the default string of "1,3,5-12" would create three Altus batch files, one for the range 1-1, one for the range 3-3, one for the range 5-12. The behaviour of Altus in the first two cases is unknown to me.
Notice that normal gaps should be well supported: "1-25,33-40,27-29" should work fine.

The script has tooltips: read them!

altusMax_v012.zip

[M.Max]

Thank u very much Lele for the script!!
It makes testing way easier
But I have tried the v012 and once the render finish I get this error :
"Unable to convert: undefined to type: FileName"
and it point to the line 88
Then I have tried the v004 and it worked !
I can zip/upload the file for you to test it if you want ?

[M.Max]

I have another issue also as only the the last third from the bottom of the image became cleaner and the rest stay noisy! I have tried to exclude the POS Pass but still the same result ! I am using Vray 3.20.02 if that matter ?
Anyone facing the same problem ?!

[^Lele^]

Thank u very much Lele for the script!!
It makes testing way easier
But I have tried the v012 and once the render finish I get this error :
"Unable to convert: undefined to type: FileName"
and it point to the line 88
Then I have tried the v004 and it worked !
I can zip/upload the file for you to test it if you want ?

It looks like you opened the scene file, then the script, and then changed to VRay as renderer from something else, is that possible?
It's the only case i can think of for which the split file path could result undefined instead of an empty string.

Regardless, it's now checked against undefined and empty.

altusMax_v013.zip

[M.Max]

Thanks for the quick answer Lele
No that's not the case .. I have just found the problem ...When I use a number - "2" in my case- to name the rendered image the error show up and when I have tried the V012 on the same file and with just changing the output name to some thing like "test" then it worked again
now with the V013 it works even with a number as a name , Thanks again for the update .

[^Lele^]

very odd. glad it's fixed!

[innobright]

I have another issue also as only the the last third from the bottom of the image became cleaner and the rest stay noisy! I have tried to exclude the POS Pass but still the same result ! I am using Vray 3.20.02 if that matter ?
Anyone facing the same problem ?!
[ATTACH=CONFIG]26752[/ATTACH]

Can you email your inputs to support@innobright.com so we can run some tests?

[pingus]

For animation mode, why do you render twice every frame ?
Can't we denoise between frame n+1 and/or n-1 ?

[^Lele^]

For animation mode, why do you render twice every frame ?
Can't we denoise between frame n+1 and/or n-1 ?

Err, no.
It's based on frame variance.
Temporal variance too, for animations (which takes N frames into account as well).
It does need two images per frame, right now.