[PlayRaw] Something different: Match the JPG!

For a while now, I have been struggling with recreating the colors and tones of my Fuji camera in Darktable. So how about we turn this into a challenge: The explicit goal of this PlayRaw is to match the JPG colors straight out of my cameras.

Edit: Ideally, you should achieve a similar rendition without local adjustments.

To this end, the following JPGs are unedited, but shot with different camera settings:

DSCF6840.JPG (11.3 MB)
DSCF6840.RAF (19.1 MB)

DSCF6960.JPG (14.3 MB)
DSCF6960.RAF (20.2 MB)

DSCF7812.JPG (15.4 MB)
DSCF7812.RAF (23.9 MB)

Trying this challenge myself, I have already discovered a few secrets. But I’ll post those later; let’s first see what you can come up with!

These files are licensed Creative Commons, By-Attribution, Share-Alike.

6 Likes

I missed the goal to match the JPG colors intentionally because I did not like that colors. They looked somewhat muddy and had a strong red/purp tint which seemed to be too intense for a sundown scene. Especially the bright clouds looked unnatural for my taste.
My result for the first photo:


The sky is more blue and the purp haze above the horizon and in the lower clouds give the feeling of a sundown.

I decreased the saturation in the clouds and increased contrast in the thin clouds that are slightly visible in the background.
As you can see in a 100% crop there is much more contrast/detail and less noise in the clouds. And the red haze is gone.

In the lower part the blue in water and sky looks more natural and you see the difference in noise reduction. Afterwards I think I could have shapened it a bit more.

DSCF6840.RAF.xmp (15.4 KB)

3 Likes

I picked the city image, #3. in rawproc:

With the linear RGB as a start, I applied a filmic curve and scooched it around to match the tone, but found the colors still slightly desaturated based on the camera JPEG, so I added just a bit of HSL saturation.

Oh, finally got a raw file with a CFA black subtract to test my new logic; works! Thanks for that opportunity…

2 Likes

Hi,
Here are my attempts…



1 Like

DSCF6840.RAF.xmp (8.0 KB)

DSCF6960.RAF.xmp (7.0 KB)

DSCF7812.RAF.xmp (12.9 KB)

It actually was a rather red evening. But I like your edits nontheless!

That’s pretty close! Oddly enough, on my screen the original JPG looks a bit less saturated than yours, or the ones by @age and @agriggio. Maybe that’s a color space issue?

1 Like

Are you colour managing your apps, OS and screen?

The JPGs I simply compared in the browser. But as they’re both JPGs, in the same browser, so that shouldn’t be an issue, right?

When editing, I’m using a calibrated screen and color managed apps (darktable and digikam, mostly).

The Agriggio’ s versions are very close to the camera jpg (have you used the dcp profile?).

The camera jpg looks with less saturation because of 8 bit processing and extreme chroma compression

Just a really quick one borrowing the X-E2 color profile in RawTherapee:

I’d call it acceptable. The colors still look a bit distorted to me, but so does the jpeg. :slight_smile:

Could you explain that in more detail?

DSCF6840.RAF.xmp (23.4 KB)
My attempt at matching the jpg, used heavy exposure correction, some contrast equalizer ton adjust the cloud and mainly “filmic rgb” and “haze removal” to tweak the dynamic. Colorwise I tried to fine tune the color balance and tweaked the sky with color zones and velvia with a mask.

3 Likes

Impressive! Which side is yours?

Thanks :slight_smile: , The right side is my edit, the clouds are a little bit too saturated and the boat/sea a little too dark but I could fix it I think

This is utterly fascinating. I must at this point confront that my initial thesis “Fuji colors are hard to recreate” is wrong. Even from my own experimentation for this thread, they seem actually pretty straight-forward to recreate.

So I wonder how I formed that impression? Perhaps it was prompted by one of the other color profiles? Or by “more difficult” high dynamic range images? Or simply by sub-par pictures that I couldn’t rescue, which had nothing to do with colors?

At any rate, thank you all for your help! This has been very interesting already!

4 Likes

It goes to almost any mechanism: once you peel back the layers of “make-nice” so the essential dynamics are evident, then most mechanisms are easy to understand. Except quantum mechanics; Lord knows what’s going on in there… :grin:

Funny, the very best advice I’ve recently seen in this regard is in a post in the thread I read immediately before coming here, attributable to @Soupy:

1 Like

I think a whole lotta good old fashioned marketing may have something to do with it. That’s all I heard about when researching Fuji before I took the plunge. “The colors! The colors! You just can’t replicate the colors!” Yeah, I get the same results with my D700 and even my old D90 files. All of it is hype with very little substance.
I know I’m late to the party, but here’s my quick go:


DSCF7812.RAF.xmp (15.0 KB)

I think mine is a bit more contrasty, but whatareyougonnado?

3 Likes

At this point? I think this is now fully open to creative reinterpretation! And yours is nice, for sure!

2 Likes