Author Topic: Corona Color Correct: Roughness -> Glossiness  (Read 5339 times)

2020-03-18, 23:42:43

cjwidd

  • Active Users
  • **
  • Posts: 1077
    • View Profile
    • Artstation
When converting Roughness to Glossiness, it is not correct to 'invert' using the Corona Color Correct map, but you can 'flip' the white and black points using the curves(?)

EDIT: 'flip' the white and black points using the curves, but you can 'invert' using the Corona Color Correct map(?)
« Last Edit: 2020-03-19, 20:13:24 by cjwidd »

2020-03-19, 10:56:20
Reply #1

romullus

  • Global Moderator
  • Active Users
  • ****
  • Posts: 8779
  • Let's move this topic, shall we?
    • View Profile
    • My Models
Why it is not correct to invert using Corona CC node and what is the difference between "inverting" and "flipping black and white points in curve editor"?
I'm not Corona Team member. Everything i say, is my personal opinion only.
My Models | My Videos | My Pictures

2020-03-19, 20:16:51
Reply #2

cjwidd

  • Active Users
  • **
  • Posts: 1077
    • View Profile
    • Artstation
'correct' might be a loaded term in this case, but the results are different - I just don't understand why(?)

Curves


Invert

2020-03-19, 20:34:31
Reply #3

romullus

  • Global Moderator
  • Active Users
  • ****
  • Posts: 8779
  • Let's move this topic, shall we?
    • View Profile
    • My Models
Indeed, you're right. I've compared it to the max's native output node and it looks like invert colours in Corona CC node is broken. Moving topic to bug reporting.
I'm not Corona Team member. Everything i say, is my personal opinion only.
My Models | My Videos | My Pictures

2020-03-19, 23:07:20
Reply #4

TomG

  • Administrator
  • Active Users
  • *****
  • Posts: 5434
    • View Profile
Haven't tried the native Max, but I did this - took a bitmap, put it through two Corona Color Corrects, one using Invert, one using the "reversed curve". I then screengrabbed, copy-pasted into Photoshop, duplicated, and inverted the duplicate there. The Corona Color Correct invert, once inverted again in Photoshop, matches the original (and the Photoshop Invert matches the Corona Color Correct invert).

The Corona Color Correct curve change, however, looks very different from the Photoshop invert result. So, to me, it looks like the Invert in the Corona Color Correct is working as expected, and either there's a problem with the curve, or there's just a misunderstanding as to what the curve does. As to what Max native invert is doing, not sure :)
Tom Grimes | chaos-corona.com
Product Manager | contact us

2020-03-20, 03:17:41
Reply #5

cjwidd

  • Active Users
  • **
  • Posts: 1077
    • View Profile
    • Artstation
Perhaps that is the question, what does it mean to 'flip' the curve handles as I did in the original post? I thought it would just be an invert(?)

2020-03-20, 10:46:14
Reply #6

maru

  • Corona Team
  • Active Users
  • ****
  • Posts: 12708
  • Marcin
    • View Profile
I will take a closer look at this, but my first thought is that most likely this is related to this phenomenon:
https://forum.corona-renderer.com/index.php?topic=19997.msg144961#msg144961

So it's probably gamma-related and/or filtering-related.
Marcin Miodek | chaos-corona.com
3D Support Team Lead - Corona | contact us

2020-03-20, 10:50:11
Reply #7

romullus

  • Global Moderator
  • Active Users
  • ****
  • Posts: 8779
  • Let's move this topic, shall we?
    • View Profile
    • My Models
Tom, when you test it in photoshop, don't forget that inevitably there's gamma involved. I did my test in linear and i think, it's pretty obvious which node is broken (see attachement).

Perhaps that is the question, what does it mean to 'flip' the curve handles as I did in the original post? I thought it would just be an invert(?)

You are correct - swapping curve's endpoints = invert.
I'm not Corona Team member. Everything i say, is my personal opinion only.
My Models | My Videos | My Pictures

2020-03-20, 10:52:18
Reply #8

Juraj

  • Active Users
  • **
  • Posts: 4743
    • View Profile
    • studio website
The issue here could be that Roughness map by standard, is authored at Gamma 1.0 (technically glossiness should be too but that isn't any standard and everyone uses gamma arbitrarily).

So if you want to convert Quixel/Substance Roughness map into correct and expected glossiness result, you need to also use gamma outside of invert, because the invert alone is only in sRGB space (it's in sRGB space tab).
Please follow my new Instagram for latest projects, tips&tricks, short video tutorials and free models
Behance  Probably best updated portfolio of my work
lysfaere.com Please check the new stuff!

2020-03-20, 12:31:10
Reply #9

maru

  • Corona Team
  • Active Users
  • ****
  • Posts: 12708
  • Marcin
    • View Profile
Juraj's explanation seems to be correct. It seems that "invert colors" and "invert curve" operate in different gamma/color space.
I am attaching an example with a possible workaround.
I will now report it to the devs, but keep in mind that this may be the expected behavior.




(Internal ID=480940238)
« Last Edit: 2020-03-20, 12:38:54 by maru »
Marcin Miodek | chaos-corona.com
3D Support Team Lead - Corona | contact us

2021-01-16, 11:48:21
Reply #10

cjwidd

  • Active Users
  • **
  • Posts: 1077
    • View Profile
    • Artstation
Man, this came back to bite me *again*. Maybe there should be a warning in the curve editor(?) Was there some consensus from the devs about this, or is it too niche of an issue, or already well understood by most of the user base?