Author Topic: Contradictory information at freshdesk for "What is GI vs. AA balance?"  (Read 6081 times)

2017-02-14, 19:47:53

SharpEars

  • Active Users
  • **
  • Posts: 103
    • View Profile
Reading the documentation at the Fresh Desk for one of the more difficult concepts in Corona to wrap one's head around, What is GI vs. AA balance?, I noticed contradictory information. Namely, at the top we have the statement that:

If "Light samples multiplier" is set to 2 and "GI vs. AA balance" is set to 16, this means that 16 GI samples and 32 direct light samples will be used for each pixel per pass. This implies that:

32 = 16 ⋅ 2   # Or more generally:
Direct_Light_Samples = GI_VS_AA_Balance ⋅ LSM


Then further down the page, we have:

Example 3
GI vs. AA = 64
LSM = 0,25
64 samples are used for GI and 16x0,25=4 samples are used for direct light. There is visibly more noise in direct light pass than in GI pass. This implies that:

4 = 16 ⋅ 0,25   # Or more generally:
Direct_Light_Samples = 16 ⋅ LSM    # i.e., the GI vs. AA balance setting of 64 has no affect on the pre-adjusted "# of direct light samples" starting setting, which is fixed at 16 and cannot be changed by the user


Only one of the above can be correct. Which one (and please update the freshdesk to reflect the correct answer)?


« Last Edit: 2017-02-14, 20:00:59 by SharpEars »

2017-02-14, 19:54:37
Reply #1

PROH

  • Active Users
  • **
  • Posts: 1219
    • View Profile
As I understand it, example 3 means that there are 64 x 0,25 = 16 directlight samples.

2017-02-14, 19:55:20
Reply #2

SharpEars

  • Active Users
  • **
  • Posts: 103
    • View Profile
As I understand it, example 3 means that there are 64 x 0,25 = 16 directlight samples.

That is the question! Is this a typo or is this correct?

2017-02-14, 20:00:02
Reply #3

PROH

  • Active Users
  • **
  • Posts: 1219
    • View Profile
I'm sure I'm correct :)

There's no "fixed" 16 value. The LSM is a multiplier of the GI/AA value, giving the amount of direct light samples.

Hope it makes sense

2017-02-14, 20:01:49
Reply #4

SharpEars

  • Active Users
  • **
  • Posts: 103
    • View Profile
I'm sure I'm correct :)

There's no "fixed" 16 value. The LSM is a multiplier of the GI/AA value, giving the amount of direct light samples.

Hope it makes sense

I believe you, in part because that makes the most sense. They should update the freshdesk article to fix the incorrect info, because it makes an already confusing topic even more confusing.

2017-02-14, 20:15:10
Reply #5

PROH

  • Active Users
  • **
  • Posts: 1219
    • View Profile
I can't see what's wrong in the fresh desk writing, exept the fact that you somehow misunderstood it. So maybe it's more about trying to clearify the meaning further, and not a "correction of incorrect info".

2017-02-14, 20:24:49
Reply #6

SharpEars

  • Active Users
  • **
  • Posts: 103
    • View Profile
I can't see what's wrong in the fresh desk writing, exept the fact that you somehow misunderstood it. So maybe it's more about trying to clearify the meaning further, and not a "correction of incorrect info".

What is there to misunderstand? The underlined values are incorrect and should be 64 and 16, respectively:

Example 3
GI vs. AA = 64
LSM = 0,25
64 samples are used for GI and 16x0,25=4 samples are used for direct light.


2017-02-14, 20:25:25
Reply #7

romullus

  • Global Moderator
  • Active Users
  • ****
  • Posts: 9283
  • Let's move this topic, shall we?
    • View Profile
    • My Models
Yes, that is typo. Moving to bug reporting.
I'm not Corona Team member. Everything i say, is my personal opinion only.
My Models | My Videos | My Pictures

2017-02-14, 21:18:10
Reply #8

PROH

  • Active Users
  • **
  • Posts: 1219
    • View Profile
Ahhh, yes now I see it :) Thanks for pointing it out.

2017-02-14, 22:59:23
Reply #9

cecofuli

  • Active Users
  • **
  • Posts: 1578
    • View Profile
    • www.francescolegrenzi.com

2017-02-15, 07:33:28
Reply #10

Christa Noel

  • Active Users
  • **
  • Posts: 911
  • God bless us everyone
    • View Profile
    • dionch.studio
this discussion talk about values of some parameters in rendersetup, it reminds me to vr4y haha! as the cecofuli said, we'll have automatic GIvsAA & LSM feature and we will no need to fix the GivsAA freshdesk page, it just need to be removed :) and there will be no any other discussion like this anymore :D
95% of my projects never change the default render settings, except the GI solver, ondra knew that in many situation we can call "laziness" as a different form of efficiency

2017-02-15, 11:49:08
Reply #11

cecofuli

  • Active Users
  • **
  • Posts: 1578
    • View Profile
    • www.francescolegrenzi.com
I think that it will be good to maintain the possibility to change manually this parameter.
But yes, please.. I don't want to see anymore these formulas in Corona Forum.
I was vomiting to think the hours spent to optimize the old V-Ray Ahahaha =)

2017-02-15, 16:14:23
Reply #12

SharpEars

  • Active Users
  • **
  • Posts: 103
    • View Profile
Automatic/adaptive settings are great, but always leave the ability to manually tweak, just in case (for atypical scenes). This is the direction that VRay is moving as well - ability to automate and defaults, but allow for manual override for difficult scenes/lighting.

It's just like metering and focus in a digital SLR. It's great 90% of the time, but that 10% of the time when it guesses wrong on a difficult scene/subject, I am glad that I can manually override.
« Last Edit: 2017-02-15, 16:25:50 by SharpEars »

2017-02-16, 13:19:58
Reply #13

maru

  • Corona Team
  • Active Users
  • ****
  • Posts: 13722
  • Marcin
    • View Profile
Thanks for reporting the typo. It is fixed. Also the screenshot is old - we will update it.
We will be super grateful for catching any typos, obsolete images, and other errors in the helpdesk.
Marcin Miodek | chaos-corona.com
3D Support Team Lead - Corona | contact us