Unified sampling Quality Issues (Maya 2013 SP2)

Become a member of the CGSociety

Connect, Share, and Learn with our Large Growing CG Art Community. It's Free!

Thread Tools Search this Thread Display Modes
  02 February 2013
Question Unified sampling Quality Issues (Maya 2013 SP2)

Hi guys,

Im encountering strange problem when rendering using unified sampling approach (Maya 2013 SP2, Win7 64)

I really dont understand whats happening there....but my pictures has obvious sampling errors/low quality...

Actualy lower quality settings gives me better antialiasing on edges, more noise in area lights (shadows) and higher quality settings give me opposite (worse edges better shadows). See pictures with info about my settings and output quality....I also cut in alpha channels so you can see it more clearly (red circles) In yellow circles there are shadow areas... (they work, better settings = better quality / less noise in them)

Does anybody know whats happening there?? Use right click and see the full res pictures...

also screengrab of my scene light setup:

P.S. I use 2 area lights (mentalray type of course, no decay, no mray custom lightshader) + simple IBL with rampshader for color + FGather with just primary rays (0 secondary bounces).
  02 February 2013
What are your filter settings?
My opinions are always my own...and maybe a friend's, but never my employer's.
  02 February 2013
thx for interest! Its really strange, other than aa problems on the bench planks the image looks fine (nice mblur, low shadow noise from areaLights) even on lower settings like the qual 1 (min 1 max 100)

my settings are :

Gauss, filter size 3.0 3.0, Sample Lock ON

Ive also experienced these problems however when using Gauss, filtersize 2.5 2.5

P.S. regarding unified sampling settings: additional sampling / error cutoff left at 0.00 Im rendering to 8bit tiffs but even when used linear EXR these aa problems showed up (of course 32bit float buffer)
  02 February 2013
so Ive made further testing with following findouts...

* the AA filter type (ie Gauss) doesnt matter (changing filter size and or filter type)
* comparing unified sampling / progressive unified produce much better quality with the same settings and is even faster!

The aliasing problem occurs whenever I go with quality over 1.0 (1.5 produces worse picture than 1.0 settings)

If need be I can send the scene for your testing....

Thanks again for your interest into this problem!

  02 February 2013
I think what Bitter was wanting to know is when you applied the custom string options to miDefaultOptions, what did you set the values for the following to:

"samples min"
"samples max"
"samples quality"

cause in the screen shots you show, a max of 300 is really high for the scene IMO... I've had excellent results between 64. and 128. float...
-- LinkedIn Profile --
-- Blog --
-- Portfolio --
  02 February 2013
For production rendering we use

min 1
max 250
Quality 1.5 to 2.5

200 or more samples a pixel isn't too bad, you should be using near brute force settings so it's fast. It also works well for motion blur and DOF

Do not do renders with Progressive for final images. It doesn't follow the Quality setting very well at all and it doesn't output passes (framebuffers)

It only seems faster because it samples the frame progressively a single sample per pixel at a time. Unified with Progressive off samples multiple times per pixel before stopping. Progressive is meant to be used with IPR for tuning only.
My opinions are always my own...and maybe a friend's, but never my employer's.
  02 February 2013

thanks guys for your input but any idea whats happening there?

How is possible that quality 1.0 gives better edge antialiasing than for ie quality 2.5 ? Of course speaking of the same min/max sample setting!

Pls send me privatemessage with your email and I ll send you maya scene (small size) so you can do some testing...

Its really pain to cannot rely on this unified approach
  03 March 2013
I never could reproduce this effect.
My opinions are always my own...and maybe a friend's, but never my employer's.
  07 July 2013
Keep the sample quality to 3 or 4.
Max sample to 100 and Done. Actually low max sample will give better edge result. Why? I don't know.
  07 July 2013
Had some issues with unified sampling and alpha channel, but that was long time ago when unified sampling was just implemented. They fixed it with next release.

For production rendering i'm using 200 max samples and 1 - 1.5 quality. But if deadline is very tight, max samples can go to 64 or even 32 with 1 for quality if scene is not very complex. Low render time without much quality loss.

Bitter, you are like mental ray expert here. Can i ask: what would be your workflow for production quality interior lighting with mental ray (not iray). What would be your render settings. Would be very greatful for your answer. And sorry for offtopic.
  07 July 2013
Originally Posted by PurpleRainbow: They fixed it with next release.

Please tell which release. For those who don't have new version they can use above method by PurpleRainbow.

Last edited by Issac : 07 July 2013 at 04:24 AM.
  07 July 2013
Thread automatically closed

This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.
CGTalk Policy/Legalities
Note that as CGTalk Members, you agree to the terms and conditions of using this website.
Thread Closed share thread

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

vB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Society of Digital Artists

Powered by vBulletin
Copyright 2000 - 2006,
Jelsoft Enterprises Ltd.
Minimize Ads
Forum Jump

All times are GMT. The time now is 03:10 PM.

Powered by vBulletin
Copyright ©2000 - 2017, Jelsoft Enterprises Ltd.