# does adaptation only work for anti aliasing in

**URL:** <https://forums.chaos.com/t/does-adaptation-only-work-for-anti-aliasing-in/17291>\
**Category:** General\
**Created:** [March 5, 2008, 8:00am UTC](https://forums.chaos.com/t/does-adaptation-only-work-for-anti-aliasing-in/17291 "2008-03-05T08:00:39Z")\
**Posts on this page:** 8\
**Page:** 1

<div class="post-metadata">

**Author:** ![pg1](https://avatars.discourse-cdn.com/v4/letter/p/ac8455/32.png) [@pg1](https://forums.chaos.com/u/pg1)\
**Post date:** [March 5, 2008, 8:00am UTC](https://forums.chaos.com/t/does-adaptation-only-work-for-anti-aliasing-in/17291/1 "2008-03-05T08:00:39Z")

</div>

hey  
have been wondering about adaptation only and what its supposed to do.  
I see from this thread [http://www.chaosgroup.com/forums/vbulletin/newreply.php?do=postreply&t=39696](http://www.chaosgroup.com/forums/vbulletin/newreply.php?do=postreply&t=39696)

that there is a bug with how it effects samples in the irmap calculation.

Is there also a problem with how it effects the noise in the anti aliasing also.

As I understand hows it supposed to work when ticked there should be an increase in rendertime as it is now anti aliasing more in the darks.  
Now you can brighten later in post and not have noise in the dark areas.

But I am finding that it is having very little effect, slightly increasing the rendertime but not really doing much to the anti aliasing in the darks.

---

<div class="post-metadata">

**Author:** ![pg1](https://avatars.discourse-cdn.com/v4/letter/p/ac8455/32.png) [@pg1](https://forums.chaos.com/u/pg1)\
**Post date:** [March 11, 2008, 6:07am UTC](https://forums.chaos.com/t/does-adaptation-only-work-for-anti-aliasing-in/17291/2 "2008-03-11T06:07:58Z")

</div>

We have found a problem with the anti-aliasing settings and our current linear workflow (gamma 2.2). The problem is that when the render has gamma 2.2 applied the render appears to the anti-aliasing sampler as being much darker. Adaptive subdiv uses mainly the colour threshold to figure out how many subdivs the render applies to each pixel. Because the entire render is now darker the threshold is less effective and the darker areas of the image (in this case most of it) are only anti-aliased at the lowest setting. This generates a lot of noise in dark textures and on edges in darker areas.

the current way we get similar subdiv anti-aliasing as to non-linear workflow is to set the colour threshold to around half of the original setting (ie Gamma 1: Clr Thresh 0.1 --\> LWF: Clr Thresh 0.05) Even this however still produces more noise in the darks and edges in the shadows. It also renders slower than Gamma 1.

\* Nrm thresh:

Another option is to leave the Colour threshold at 0.1 and to add the normal threshold option. This option at default is 0.05 and causes the render to almost double in render time. However it will stop any flickering of trees, vertical thin lines etc. It will not prevent noise in textures expecially in dark areas.

My understanding was that the “dont affect colours adaptation only” button was to address this issue but after doing test images with samples turned on we found that there seems to be very little difference to the antialiasing in the dark areas of the image.

I will post some images to try and show what I’m asking about.

---

<div class="post-metadata">

**Author:** ![pg1](https://avatars.discourse-cdn.com/v4/letter/p/ac8455/32.png) [@pg1](https://forums.chaos.com/u/pg1)\
**Post date:** [March 11, 2008, 6:10am UTC](https://forums.chaos.com/t/does-adaptation-only-work-for-anti-aliasing-in/17291/3 "2008-03-11T06:10:40Z")

</div>

This is the base render made using LWF

 ![G2.2_Base_Render_ClrThresh_01.jpg](https://europe1.discourse-cdn.com/chaosprd/original/3X/0/e/0e011b23051c15339135d30069f6cdf73c5cac8b.jpeg)

---

<div class="post-metadata">

**Author:** ![pg1](https://avatars.discourse-cdn.com/v4/letter/p/ac8455/32.png) [@pg1](https://forums.chaos.com/u/pg1)\
**Post date:** [March 11, 2008, 6:12am UTC](https://forums.chaos.com/t/does-adaptation-only-work-for-anti-aliasing-in/17291/4 "2008-03-11T06:12:33Z")

</div>

This is the same image in a non linear work flow (no colour corrects gamma 1) as you would expect it is much darker

 ![G1_Base_Render_ClrThresh_01.jpg](https://europe1.discourse-cdn.com/chaosprd/original/3X/2/f/2f0774453ff17438426afacc5d94cc6c7a49b9e6.jpeg)

---

<div class="post-metadata">

**Author:** ![pg1](https://avatars.discourse-cdn.com/v4/letter/p/ac8455/32.png) [@pg1](https://forums.chaos.com/u/pg1)\
**Post date:** [March 11, 2008, 6:33am UTC](https://forums.chaos.com/t/does-adaptation-only-work-for-anti-aliasing-in/17291/5 "2008-03-11T06:33:45Z")

</div>

this is the anti aliasing (show samples ticked) for the gamma 1 image  
Adaptive subdivision clr threshold 0.1

 ![G1.0_ClrThresh_.1.jpg](https://europe1.discourse-cdn.com/chaosprd/original/3X/2/3/23f8f3ee9da52eef3627a03a5c0dd549b3a009dc.jpeg)

---

<div class="post-metadata">

**Author:** ![pg1](https://avatars.discourse-cdn.com/v4/letter/p/ac8455/32.png) [@pg1](https://forums.chaos.com/u/pg1)\
**Post date:** [March 11, 2008, 6:38am UTC](https://forums.chaos.com/t/does-adaptation-only-work-for-anti-aliasing-in/17291/6 "2008-03-11T06:38:48Z")

</div>

To get a similar (actually slightly better) result in anti aliasing the dark areas of the image in the gamma 2.2 image the colour threshold was changed from .1 in the previous image to .05 in the linear workflow render.

 ![G2.2_ClrThresh_.05.jpg](https://europe1.discourse-cdn.com/chaosprd/original/3X/7/4/745f8dc9ccbe1acca366c155d4af25300ee06b9b.jpeg)

---

<div class="post-metadata">

**Author:** ![pg1](https://avatars.discourse-cdn.com/v4/letter/p/ac8455/32.png) [@pg1](https://forums.chaos.com/u/pg1)\
**Post date:** [March 11, 2008, 6:51am UTC](https://forums.chaos.com/t/does-adaptation-only-work-for-anti-aliasing-in/17291/7 "2008-03-11T06:51:03Z")

</div>

In all of the above images the results are as expected  
This final image is  
gamma 2.2  
Adaptive subdiv clr threshold 0.05  
AND dont affect colour adaptation only ticked which is the only difference between this image and the previous image

and yet it is identical to the previous image where dont affect colours was left unticked

So the reason for this long post is since we have changed to a Linear workflow we are finding that there is a lot more noise in the darks. To get around this we have increased our anti aliasing setting (which of course means slower renders) which is antialiasing the whole image.

I thought that the dont affect colours tick box should tell vray to be clever and anti alias more in the darks as we are going to gamma correct the image later in Afterfxs but it seems to be having no effect at all or what are we doing wrong?

 ![G2.2_ClrThresh_.05 dont affect.jpg](https://europe1.discourse-cdn.com/chaosprd/original/3X/4/d/4df89e9718316fcc56e0615d5dd0a162925462ee.jpeg)

---

<div class="post-metadata">

**Author:** ![vlado](https://avatars.discourse-cdn.com/v4/letter/v/779978/32.png) [@vlado](https://forums.chaos.com/u/vlado)\
**Post date:** [March 11, 2008, 7:20am UTC](https://forums.chaos.com/t/does-adaptation-only-work-for-anti-aliasing-in/17291/8 "2008-03-11T07:20:01Z")

</div>

The bug with the irradiance map has been corrected for sp1.

For the antialiasing, you need to make sure that:  
(\*) Gamma in the color mapping section is set to 2.2;  
(\*) The Don’t affect colors option is on;  
(\*) Subpixel mapping is off (there was a bug here - if subpixel mapping was on, the don’t affect colors option did not work properly; this is corrected for newer builds);

Best regards,  
Vlado
