Using Fuji F-log 2 in Nuke with ACES

Discussions - Using ACES VFX
1

Hi,

So I’ve been having a tough time using Fuji F-log 2 footage in Nuke, that I shot on the X-H2s. Basically I’ve been able to successfully convert the log footage into ACEScg linear with an expression node, using the appropriate information from the F-log 2 datasheet from Fuji: https://dl.fujifilm-x.com/support/lut/F-Log2_DataSheet_E_Ver.1.0.pdf

I’ve also found how to properly add the datasheet values in Nuke using the expression found here: F-Log to ACES — PostWork.io

The only step I need to take now is to convert the ACEScg back to F-log 2, and this is where I’m stuck. Color grading is expecting F-log 2 and this is why it’s important to get the footage back to F-log 2 specifically, and I haven’t been successful in that regard.

Is there a way that I can like invert the expression in order for it to give me the footage back to F-log 2 from linear?

I’m not very experienced with these technicalities in color, usually I only work with Alexa, Sony and Red footage which is pretty straightforward to use in any software.

I appreciate any help I can get.

2

Hello and welcome here ! I think you already had everything with that paper. They specify the conversion from and to their log space.

couldn’t resist making a quick setup in nuke, which seems to work fine

set cut_paste_input [stack 0]
version 14.0 v1
StickyNote {
 inputs 0
 name StickyNote1
 label "from fuji"
 selected true
 xpos -88
 ypos -219
}
StickyNote {
 inputs 0
 name StickyNote2
 label "to fuji"
 selected true
 xpos -96
 ypos -41
}
push $cut_paste_input
Clamp {
 name Clamp1
 selected true
 xpos -31
 ypos -193
}
Expression {
 temp_name0 cut2
 temp_expr0 0.100686685370811
 temp_name1 xa
 temp_expr1 5.555556
 expr0 "r>=cut2?(pow(10, ((r-0.384316) / 0.245281)) / xa - 0.064829 / xa):(r - 0.0926864)/8.799461"
 expr1 "g>=cut2?(pow(10, ((g-0.384316) / 0.245281)) / xa - 0.064829 / xa):(g - 0.0926864)/8.799461"
 expr2 "b>=cut2?(pow(10, ((b-0.384316) / 0.245281)) / xa - 0.064829 / xa):(b - 0.0926864)/8.799461"
 name log2lin
 selected true
 xpos -31
 ypos -147
}
ColorMatrix {
 matrix {
     {0.9748949779 0.0195991086 0.0055059134}
     {0.0021795628 0.9955354689 0.0022849683}
     {0.0047972397 0.0245320166 0.9706707437}
   }
 name ITU_R_BT_2020__to__ACEScg1
 selected true
 xpos -31
 ypos -109
}
ColorMatrix {
 matrix {
     {1.025824748 -0.0200531908 -0.0057715568}
     {-0.0022343695 1.004586502 -0.0023521324}
     {-0.0050133515 -0.0252900718 1.030303423}
   }
 name ACEScg__to__ITU_R_BT_1
 selected true
 xpos -31
 ypos -9
 addUserKnob {20 User}
 addUserKnob {26 ref T "<a href=\"https://www.colour-science.org:8010/apps/rgb_colourspace_transformation_matrix?input-colourspace=ACEScg&output-colourspace=ITU-R+BT.2020&chromatic-adaptation-transform=Bradford&formatter=nuke&decimals=10\">source matrix</a>"}
}
Expression {
 temp_name0 cut1
 temp_expr0 0.000889
 expr0 "r>=cut1?0.245281 * log10(5.555556 * r + 0.064829) + 0.384316:8.799461 * r + 0.092864"
 expr1 "g>=cut1?0.245281 * log10(5.555556 * g + 0.064829) + 0.384316:8.799461 * g + 0.092864"
 expr2 "b>=cut1?0.245281 * log10(5.555556 * b + 0.064829) + 0.384316:8.799461 * b + 0.092864"
 name lin2log
 selected true
 xpos -31
 ypos 33
}
Clamp {
 name Clamp2
 selected true
 xpos -31
 ypos 65
}

image
image1828×1018 73.5 KB

3 Likes
3

It’s perfect!! Thank you so much!

Looking at your setup helped me better understand my “problem”, it’s so helpful. :pray:

I can finally finish my short film! :grin:

4

Hi! what colorspace do you read and write the footage as when you import and export it?

5

Hey there. Set the colorspace to ‘Utility - RAW’ both in the Read and Write.

6

Hello again,

Fujifilm has recently released their new F-Log2 C colorspace. According to the data sheet, the curve is exactly the same as F-Log2, what differs is the gamut.

F-Log2 C uses the F-Gamut C which is similar to ACES in green and identical in red. I’ve been using ACES gamut in Nuke and that has worked fine, but I have a handful of shots with harsh chromatic aberration and some values are starting to clip.

I’ve tried using other gamuts from other camera manufacturers but ACES is really the closest I can get to F-Gamut C. I’ve also tried to apply some gamut compression but ultimately I really want to avoid having to manually tinker with the image and just use the correct gamut.

I was wondering, is there an easy way to apply F-Gamut C to F-Log2 C footage in Nuke?

Here’s the data sheet from Fujifilm:

7

You could add this matrix node to convert F-Gamut C to AP0.

1 Like
8

Thank you, this is perfect!