Author Topic: Megascans/Bridge Exporter Updates + Workflow  (Read 75708 times)

2016-10-18, 09:36:22
Reply #30

ktulu

  • Active Users
  • **
  • Posts: 33
    • View Profile
Oh, good to know. I was under the impression that CoronaBitmap would speed things up, but I've learned to trust your tests now.
Will stick to regular Bitmap then.

2016-10-18, 10:02:54
Reply #31

Ondra

  • Administrator
  • Active Users
  • *****
  • Posts: 9048
  • Turning coffee to features since 2009
    • View Profile
I would be interested in a scene where corona bitmap gives worse performance than 3dsmax bitmap ;)
Rendering is magic.How to get minidumps for crashed/frozen 3ds Max | Sorry for short replies, brief responses = more time to develop Corona ;)

2016-10-19, 18:00:15
Reply #32

ktulu

  • Active Users
  • **
  • Posts: 33
    • View Profile
I went ahead and made a very quick comparison. Sample material was Rock 01 from real displacement textures volume one, which I have organized in Megascans Bridge library hence they get pushed to max via Dubcat's script.

I did two batches:
1. All maps in 1K resolution, Blur 1,0, Pass limit 25
Bitmap: 02:22
CoronaBitmap: 02:27

2. All maps in 8K resolution (except displacement 4K), Blur 0,01, Pass limit 25
Bitmap: 02:53
CoronaBitmap: 02:30

Is it the resolution, or the blurring that is responsible for the performance hit?

Edit: Added a screengrab of the shader tree...

2016-10-19, 18:48:14
Reply #33

romullus

  • Global Moderator
  • Active Users
  • ****
  • Posts: 8833
  • Let's move this topic, shall we?
    • View Profile
    • My Models
I suspect blurring. Back in the days of alpha, the difference in render time between different blurring values was much more severe. AFAIk that was one of the resons why CoronaBitmap was introduced.
I'm not Corona Team member. Everything i say, is my personal opinion only.
My Models | My Videos | My Pictures

2016-10-19, 18:58:24
Reply #34

karnak

  • Primary Certified Instructor
  • Active Users
  • ***
  • Posts: 76
    • View Profile
A good way to find that out would be to change 1 variable at the time. :)
Calculate a (1k, blur 1) (1k, blur 0.01) (8k, blur 1) (8k, blur 0.01) and then compare the results.
Corona Academy (May 2017)

2016-10-20, 11:28:31
Reply #35

ktulu

  • Active Users
  • **
  • Posts: 33
    • View Profile
OK, so I redid the 8K tests.

1. 8K resolution (except displacement 4K), Blur 1,0, Pass limit 25
Bitmap: 02:57
CoronaBitmap: 02:57

2. 8K resolution (except displacement 4K), Blur 0,01, Pass limit 25
Bitmap: 03:29
CoronaBitmap: 02:53

With default Blur value set to 1,0 the difference is negligible (or there is none). The moment one lowers the blur value to get a crisper image CoronaBitmap starts to shine.
Notice the difference in Rays/s amount:
6.013.059 (CoronaBitmap)
5.312.145 (Bitmap)

Now, on another note. Why does the same scene, with the same settings render 20-30 seconds longer than yesterday will remain a mystery to me.

2016-11-14, 11:36:39
Reply #36

JakubCech

  • Active Users
  • **
  • Posts: 126
  • jakubcech.net
    • View Profile
    • jakubcech
Hi Dubcat,
not sure if this is the place to post this issue but I think if its suitable you can forward it to Quixel team as you do the scripts for them.
I bought this huge 4K leaf (one big leaf + stem) and noticed when I exported textures to Corona 1.5 with your latest script I am getting some artefacts produced by converted displacement map (EXR option used for export):


If I switch it with the raw untouched EXR (leaving your Mix node untouched), problems are solved:


The untouched exr does the same thing but smoothly and has lower  file size as well (36MB vs 96MB).

Jakub

2016-11-14, 14:49:29
Reply #37

maru

  • Corona Team
  • Active Users
  • ****
  • Posts: 12752
  • Marcin
    • View Profile
To me it looks like wrong bit depth. But maybe let's wait for Dubcat's findings...
Marcin Miodek | chaos-corona.com
3D Support Team Lead - Corona | contact us

2016-11-14, 19:05:58
Reply #38

dubcat

  • Active Users
  • **
  • Posts: 425
  • ฅ^•ﻌ•^ฅ meow
    • View Profile
Hey

Megascans displacement maps are not real floating point maps.

These are the official remapping values.
Quote
White = 10 cm, Gray = 0 cm, Black = -10 cm

We need to remap 0 = -10 / 127 = 0 / 255 = +10

This can be done in the "Displace modifier" by enabling "Luminance Center".
Corona Mtl does not have this feature, so I played around with Output/Mix map to remap the values. Looks like it didn't turn out that well :P
Re-download the script from Google Drive.

We really need a "Luminance Center" option in Corona.
             ___
    _] [__|OO|
   (____|___|     https://www.twitch.tv/dubca7 / https://soundcloud.com/dubca7 ( ͡° ͜ʖ ͡°) choo choo

2016-11-14, 19:25:24
Reply #39

romullus

  • Global Moderator
  • Active Users
  • ****
  • Posts: 8833
  • Let's move this topic, shall we?
    • View Profile
    • My Models
If you set displacement min level to -10 cm and max level to 10cm it should work as intended without any modification to texture. Unless i misunderstood something.
I'm not Corona Team member. Everything i say, is my personal opinion only.
My Models | My Videos | My Pictures

2016-11-14, 19:37:24
Reply #40

dubcat

  • Active Users
  • **
  • Posts: 425
  • ฅ^•ﻌ•^ฅ meow
    • View Profile
That's what I reverted back to in the latest version.
There was a reason why I made the mix map instead, but can't remember.
             ___
    _] [__|OO|
   (____|___|     https://www.twitch.tv/dubca7 / https://soundcloud.com/dubca7 ( ͡° ͜ʖ ͡°) choo choo

2016-11-16, 16:47:46
Reply #41

JakubCech

  • Active Users
  • **
  • Posts: 126
  • jakubcech.net
    • View Profile
    • jakubcech
Thank you.
I see by using script the specular map is not used at all - why is it so?

2016-11-16, 16:58:41
Reply #42

dubcat

  • Active Users
  • **
  • Posts: 425
  • ฅ^•ﻌ•^ฅ meow
    • View Profile
The specular map is containing the reflectance values, you need to convert these values to IOR.
If you measure the specular maps in linear floating point, you will see that all the megascans materials have a base reflectance of 0.0358 = 3.58 reflectance = 1.47 IOR.

When I get some spare time, I will recreate my Specular to IOR LUT to work inside CoronaOutput.
This new LUT will most likely be bundled in Bridge, and will be applied automatically with the script.
For now you have to apply the LUT inside Photoshop and load the new IOR map as gamma 1.0.

Edit:
My sheet link
IOR spreadsheet
« Last Edit: 2016-11-16, 17:16:08 by dubcat »
             ___
    _] [__|OO|
   (____|___|     https://www.twitch.tv/dubca7 / https://soundcloud.com/dubca7 ( ͡° ͜ʖ ͡°) choo choo

2017-04-27, 20:52:13
Reply #43

Max Mucko

  • Users
  • *
  • Posts: 2
    • View Profile
Hi, guys!

I'm having a bit of problem with the Megascans Bridge exporter and the 3d assets in 3ds max.
I'm attaching different renders with different material settings for the stone object. The object is from the meadow pack from quixel and is a high poly ( 1mil ) fbx object imported from Bridge through Dubcats exporter! I don't think I am doing the job as I should ( I just leave the default exported material as it is after exporting it from Bridge, but the scans do not end out like they end on Quixel website. ) Please let me know if you find out what I am doing wrong, it would be of greatest help!

Attachment 01 - default geometry export from bridge with default material settings, Attachment 02 - material default settings

Attachment 03 - object with default displacement checked, Attachment 04 - Material settings with displacement

Attachment 05 - object with bump set to 10, Attachment 06 - material settings of bump at 10

Also tried applying a displace modifier but the result does not end up well although I tried to use luminance center at 0.5 as people mentioned.

Much Thanks for the help.

Max

2018-07-21, 05:27:46
Reply #44

GVRuslan

  • Users
  • *
  • Posts: 2
    • View Profile
hey guys!
where i can find maxscript.ms from megascans?