[Play Raw] Processing another very high dynamic range raw

This is frustrating for me. I downloaded RAF and sample XMP files. And those are OK. But when I try to mimic flow done on film I can’t get close to what I see. I’m out on first exposure where my sample is much darker (Even when I set +3.0EV) and having sky much lighter.


So I gave up and tried to process with masks one of my photos I did some time ago in LR. And here is a the problem. No matter what I try I can’t recover highlights any way close to what I achieved in LR. Getting out shadows using mask is brilliant, but can’t get anything close in highlights:

On the left is LR, on the right DT working on highlights only. What am I doing wrong?
This is the original RAW:
P1440493.RW2 (18.8 MB) licence by

This is what I managed to get in LR (pushing highlight, whites, exposure and using ordinary graduated filter):

What I have to do to get the view through window close to what I get in LR?
PS. I’m a new here (and forums) so I don’t know should I do to license RAW file to users.
Thanks .

Hi,

is this a general question, or do you want a solution specifically for darktable?

Maybe make this a new play raw thread. just license the image as CC0 as seen in the start of this thread.

This should probably be a new thread.
Maybe move it? (@Morgan_Hardwood, @houz ?? )

Agreed, and done.
Topic split off and moved here from [Play Raw] Processing a very high-contrast raw - #65

Using RawTherapee 5.3-606-g00e5e2b6 (same as 5.4-rc1 which will be released later today).

P1440493.jpg.out.pp3 (10.9 KB)

P1440493 perspective.jpg.out.pp3 (10.9 KB)

3 Likes

I still like the result from LR in terms of the view through the window but I am sure that RT could do better.

@afre are you aware that the LR version basically has a super sharp and strong gradiuated filter, which ends right over the altar?
I wasn’t aware of it until I saw @Morgan_Hardwood 's version, where you can see, that the stones are actually beige everywhere (and not black on top as the LR version implies).
So @Gobo this should easily be duplicated using RT or DT. I’ll give it a try soon…

@Morgan_Hardwood, that is an impressive result!

@McCap Definitely. Just dead tired, although

basically says what you said in less words :wink:.

I also prefer window view from LR. I was thinking to learn GIMP but first time a saw a tutorial with masks in darktable made me think about it. The more tutorials I see the more I want it but I’m stuck on such simple adjustments. @McCap I know this is an extreme exposure corrections. Sharp graduated filter is +4EV right above altar. Highlights (for window) are -50. I exposed photo not to overblow highlights with a hope to bring out shadows. This is one of photos I expect to properly develop using DT. With masks this should be breeze.

Firt pass exported from rawtherapee with a very low contrast style


P1440493.RW2.pp3 (11.3 KB)

Second pass in rawtherapee


P1440493.tif.pp3 (11.4 KB)

In this case it sometimes helps to consider using a layer-based image editor for pixel-level manipulation like you’d need here (even the LR version sucks with that sharp gradient).

I gather that your main concern here is the highlight recovery outside the window. In that case, I have this first pass where I focused only on the window:

This is the profile used: P1440493.RW2.patdavid.pp3 (10.6 KB)

I used the Color Propagation method for highlight reconstruction (that thing is magic). Using Blend as the method fixes this, but washes out some of the highlights a bit (which may not be a bad thing).

For comparison (keeping in mind all I worked on was highlight recovery, not contrast or WB or lens corrections):

This is pretty much a worst-case scenario for any image processing, and I’m frankly amazed at what modern programs pull off… :slight_smile:

Personally, I would create a version of the image exposed for the window as best as possible, and a separate one for the rest of the room, and blend using GIMP or enfuse.

3 Likes

Here is an attempt using RT 5.3 (Non-dev) The dark parts below the windowsill was lightened using the negative vignette from the Transform Tab with Y-axis shifted by -76.
P1440493.jpg.out.pp3 (11.4 KB)

The only problem I see in the above image is the red fringing on the edge of the highlights on the golden dome. That was removed using cloning in GIMP (one can do better) plus a dash of Levels tool.

IMHO, This looks far more natural than LR attempt of @Gobo wherein, the walls have become unnaturally dark above the windowsill.

1 Like

Wheh, this was a properly difficult one! Awesome image :smiley:
I even had to go as far as setting the white point manually and tweak the highlight reconstruction (which I otherwise never do), and definitely go crazy on masks and color corrections. But, the end result looks quite nice I think.

Edit: while I could have pushed the brightness of the altar further, I thought it would be better to actually some of the contrast intact. It highlights the cross imo.


P1440493.RW2.xmp (30.2 KB)

2 Likes

On the left is LR, on the right DT working on highlights only. What am I doing wrong?

It is difficult to say without having the sidecar for your edit but I think the main thing you have missed in your processing with darktable is to fine-tune raw white/black points and highlight reconstruction. These steps are useful in such cases that you have clipped highlights/shadows when you turn on the raw-based overexposure indicator:

Here is a the result I could get for the highlights by using this method.


P1440493_01.RW2.xmp (6.1 KB)

And further processing for the entire photo:


P1440493.RW2.xmp (9.1 KB)

1 Like

I have a few more like this if you like it :wink:
I see that I have to put GIMP on my list of programs to install (and learn). What surprised my that most of you are using RT while I thought DT with masks should be easier to develop my sample. I’ll have look at XMP tomorrow. If I a make a step forward and start using GIMP should I also try RT or stay with and learn DT?
I remember reading “Visualizing the raw…” but as everything is new for me in DT I forgot about it. I’ll get back to that.
@msd and @Thanatomanic - your process looks best to me. I’ll analyze them with the XMP.
Thanks to all.

I think everyone’s recommendation will be slightly different, but after having a go at this with just darktable and getting some not stellar results, I’d probably make two images, one for the window, and one for the room, then blend them in gimp.

Both are excellent programs and with practice you can achieve excellent results with both. The main difference is that dt has local adjustments, while the official RT version doesn’t (there is a development version, which does local manipulation, search for local lab in this forum). So if you started with dt just keep on practicing…
:grin:

This is the correct answer! Whatever feels better, just go with it. Both are capable tools.

This is processed with darktable. Using LCH-mode for highlight reconstruction brings back a lot of structure but rather no color. In this case it worked to pull up the clipping threshold and let the highlight turn magenta. I used the color zones module to shift the magenta into a blueish tone, pulled back the saturation of the magenta tone. Additionally, I placed a drawn mask roughly over the window to avoid negative impact on the remaining image.


(looks ok, at least on my screen… maybe there is still a touch of purple in the sky)