Author Topic: Tonemapping - Plz Halp  (Read 115790 times)

2020-05-05, 13:12:46
Reply #195

James Vella

  • Active Users
  • **
  • Posts: 542
    • View Profile
Yes you are correct Fluss. Textures are converted to aces cg space before rendering.



And the sun and sky (internal Vray colorspace)?

Using Raw to AcesCG (but it doesnt make any difference in this example)



2020-05-05, 13:17:31
Reply #196

Fluss

  • Active Users
  • **
  • Posts: 553
    • View Profile
For the sun and sky you do not need OCIO nodes, you need to change the Vray internal colorspace and it's only accessible by script.

Execute this in the maxscript listener :

Code: [Select]
renderers.current.options_rgbColorSpace = 2
If you don't do this, you are rendering ACEScg textures with sRGB sun and sky
« Last Edit: 2020-05-05, 13:21:07 by Fluss »

2020-05-05, 13:18:11
Reply #197

James Vella

  • Active Users
  • **
  • Posts: 542
    • View Profile
For the sun and sky you do not need OCIO nodes, you need to change the Vray internal colorspace and it's only accessible by script.

Execute this in the maxscript listener :

Code: [Select]
renderers.current.options_rgbColorSpace = 2

Thanks! Ill give it a shot

2020-05-05, 13:24:20
Reply #198

James Vella

  • Active Users
  • **
  • Posts: 542
    • View Profile
Great tip thanks - its not true what they say about you Fluss ;) (jk), fixes the blue color cast.



2020-05-05, 13:33:48
Reply #199

Fluss

  • Active Users
  • **
  • Posts: 553
    • View Profile
Glad it helped. I'm testing Vray 5 beta and this will be easier!


2020-05-05, 13:44:52
Reply #200

James Vella

  • Active Users
  • **
  • Posts: 542
    • View Profile
Glad it helped. I'm testing Vray 5 beta and this will be easier!




This checkbox "auto RGB primaries etc..."

This does the utility rgb - acescg conversion for vraybitmaps without needing the OCIO node?

2020-05-05, 13:49:12
Reply #201

Fluss

  • Active Users
  • **
  • Posts: 553
    • View Profile
Exactly! And it's way quicker than the OCIO conversion. It's based on the filename tho, so you'll still have some work to do.

2020-05-05, 13:52:19
Reply #202

James Vella

  • Active Users
  • **
  • Posts: 542
    • View Profile
Seems like there could be a better way. Using the tag inside the VrayHDRI for sRGB or Linear. Anyway Ill stop derailing this thread with minor topics.


2020-05-05, 13:55:09
Reply #203

Fluss

  • Active Users
  • **
  • Posts: 553
    • View Profile
Seems like there could be a better way. Using the tag inside the VrayHDRI for sRGB or Linear. Anyway Ill stop derailing this thread with minor topics.

It's also planned

2020-05-05, 14:33:30
Reply #204

James Vella

  • Active Users
  • **
  • Posts: 542
    • View Profile
Test #2 with Fstorm.

For this section I kept tonemapping OFF, just to see the results of purely increasing the exposure (keep your eye on the exposure setting).








Now we reached the point of blow out. This is where it gets interesting. I turn tonemapping on (defaults)



Now decreasing the burn below 0 (which I expected not good results, but its actually ok). Not sure what to make of it to be honest.



« Last Edit: 2020-05-05, 14:37:48 by James Vella »

2020-05-05, 15:00:57
Reply #205

James Vella

  • Active Users
  • **
  • Posts: 542
    • View Profile
The way I look at it is, color space makes the difference in terms of lighting - its obvious when comparing default Vray with Aces Vray - both have default settings just adjusting exposure value (no tone mapping).

We can then see Fstorm renders without tone mapping, and the light behaves differently to exposure than say default corona/vray. Tone mapping is the last piece of the puzzle here - important yes but not as important as getting a 'photo real' result - might as well use a composite software if you arent happy with the tone mapping controls (not saying it shouldnt be improved but just saying its not going to make a world of difference which you use once you have the correct rendered result).

As others have noted about small details, bumps, materials etc these all help with the bigger picture. My 2c

2020-05-05, 15:06:17
Reply #206

lupaz

  • Active Users
  • **
  • Posts: 951
    • View Profile
I once proposed devs what I consider could work as workaround. To allow Corona to save directly into .dng format. Adobe's DNG has both 16bit and 32bit floating point options. And I thought perhaps this would confuse raw editors less, effectively "trick" them.

This would be great! Honestly, I think if we could treat Corona as a camera (with the .dng format) and use any post production tool we choose, the tone mapping conversation would get much easier to follow and understand.


2020-05-05, 15:13:54
Reply #207

lupaz

  • Active Users
  • **
  • Posts: 951
    • View Profile

Download the scene from here (created in a student version so none of it can be used for commercial purposes, only learning):https://share.getcloudapp.com/yAu20wZX

It doesn't include the textures? There are only Max files in that zip file.

Edit: I see a problem with this test scene: It uses corona sun and sky. May be we should use an HDRI to make comparisons easier.
« Last Edit: 2020-05-05, 15:17:48 by lupaz »

2020-05-05, 16:29:22
Reply #208

lolec

  • Active Users
  • **
  • Posts: 179
    • View Profile
Haha fair play, I'll concede that I was indeed duped by this.

Cool that you took it the right way :)

WOW! thanks for downloading the scene and playing around with it. I think there is a lot we can learn from each other.

My objective was to see other people's Corona workflow, so that's why I used Corona tools exclusively, It wasn't meant to compare different engines.

HDRs are loaded at a different starting position and materials would need to be manually converted, it is not a straightforward process.

However, it is very interesting to see different workflows even within Corona! Hope more of you can download and try the scenes, I will later compile all of the images and see what we can learn about it.

2020-05-05, 18:11:36
Reply #209

lupaz

  • Active Users
  • **
  • Posts: 951
    • View Profile
Not sure if anyone is still interested in this. If not, let me know.

Below I rendered a test with gray materials. I removed the glass. No DOF still.
I did Fstorm, Corona with cache and Corona brute force.
Corona with PT never cleared the noise, even after over 300 passes.


It seems that Corona with path tracing is closer to Fstorm in terms of depth of shadows than with cache. Probably expected though.
So may be if we want more depth like with fstorm we need to use path tracing exclusively.

Look at the shadows inside the profile:


With cache, the shadow inside that tube is not as defined as with brute force and fstorm. I believe this is in part why people think tone mapping is the problem. Because we try to get more contrast but there's just not enough detail to begin with.


Corona Cache:




Corona Path Tracing:



Fstorm: