Problem getting nice colors, maybe problem with LED spots

All my files, which I post here, are licensed Creative Commons, By-Attribution, Share-Alike.

I had problems again and again to upload the original files, so you can get it at cathedral – Google Диск amd you are welcome to post them here.

I have 4 similar examples and you can choose which you like to discuss.

All files have problems with the color and I am not able to correct them, so the color looks nicely. The goal is to see a “bright golden” color

If you try something, please post your xmp-file. Please criticize what I have tried.

Sorry, the source is not raw, but jpg at my download-link, raw does not exist.

I know the source is very bad.

These are my results.

Sorry forgot to write the name below, you see it with mouseover. It are 1-4.

Just looking at your history and offering constructive criticism for you to consider as part of your learning process. You are encouraged to disagree with what I say here as it is your image and you know what you want it to look like.

Why the contrast equalizer. If used for sharpening this was unnecessary in my view as your phone already sharpened the JPG.

With local contrast the values set high making the contrast too harsh in my view.

RGB levels appears to be doing no change to the picture

AgX was not of benefit to this JPG and made the shadows too dark

Tone equalizer was used to lighten the image but not the dark shadows, which also needed lightening.

Color balance rgb used to increase global saturation and global brilliance. No specific criticism of this choice as adjustments were modest. Modest adjustments are often better than sledge hammer adjustments.

My edit and I am not saying this is necessarily the best way to edit the image. I hope others can post better edits than mine.
1. Activate the color assessment condition which places a white border around image to assist with decision making for color, contrast and brightness.
2. Applied local contrast first at default settings as this influences the final look and I use it on most of my images.
3. Color balance RGB to brighten the shadows as this module is very easy to use and works well with jpgs. I used an inverted drawn mask to exclude the alter from this correction.
4. Additional instances of color balance rgb using masks to brighten or darken selective portions of the image.
5. Two instances of shadow and highlights module only brightening the shadows.
6. Moved AgX above the shadow and highlights modules. Then used the auto tune levels and the picker for pivot target output. Then I lowered the contrast a little
7. Rotate and perspective module to improve the vertical lines.
8. Crop to personal taste, but chose a square crop.
9. I applied no sharpening to this image as I believe the phone has done that sufficiently already.
10. I did not touch color as I was happy enough with the color.


cathedral1_01.jpg.xmp (32.0 KB)

3 Likes

As these are jpg not raw it might be better to use gimp or krita instead of darktable. You can use darktable to edit jpg, but all the scene referred modules will not be applicable, since the jpg already has a tone curve, and scene referred modules are used before the tone curve.

cathedral1_02.jpg.xmp (15.0 KB)

cathedral1.jpg.xmp (8.5 KB)

cathedral2_01.jpg.xmp (10.7 KB)

2 Likes

I would have thought that DT will transform the image data to linear space when opening an image. I’ll have a closer look into the docs once I’m back at a desktop PC.

IIRC the main issue with jpgs is mostly that a lot of information is irrecovably thrown away, and later edits will highlight that missing data and produce artifacts easily. You cannot push a jpg as far as a RAW.

1 Like

But just if you are comfortable with those tools.
The changes the TO expects can be done easily with darktable since it’s not about drawing, painting…

1 Like

See processing-modules/input-color-profile:

This module takes the color space used by the image source (e.g. camera, scanner) and converts the pixel encodings to a standardized working color space.

That one is linear rec.2020 by default, and will be shown in DT:

Yes the colour space can be set to linear, but jpegs still have a tone curve baked in whether the colour space is linear or not, thus all your adjustments will be on non linear data. That’s display referred, not scene referred. Scene referred only applies before the tone curve is added, for which raw is required.

For example, it is unnecessary to use agx, which is a tone curve, on a jpeg, since jpeg already has a tone curve. If you like the look artistically that’s fine, but technically it’s not advisable.

In darktable then, all the modules that come after agx (or filmic, sigmoid, etc…) in the pipeline are display referred. Those would be suitable for editing jpeg.

I want to know what I can do with darktable. The goal is not to reach 100% with editing, better a little bit is enough.

I’m not really sure if we are talking about the same thing. I should have quoted earlier what my response all was about:

Yes, the tone curve is baked in, as are many other transformations (most easily seen, I think, are lens-corrections, sharpening, white-balance and color-science, and of course the tone curve we are talking about). But there is no reason why scene-referred modules will not be applicable because of this.

Consider this counterexample: I have one instance of tone equalizer, compressing shadows and highlights, improving contrast in midtones, just doing so by hand.

From now on, for all later modules in the pixel-pipe, we are working on data with a tone-curve baked in.

But I still can use another tone equalizer instance, or exposure, or color balance rgb. The existence of a non-linear tone-mapping step does not make it “technically inadvisable” to later do further scene-referred operations.

Of course I can perform steps early in the pixelpipe that will cause artifacts later. And in the case of jpegs, such transforms have been done, and in such a way that they cannot be undone (I cannot uncompress the highlights, for instance, and as such S-curve-like adjustments like filmic, sigmoid, AgX are usually not helpful).

But whether a module works or not is less to do with “scene-referred” or “display-referred”, and more whether the specific transformation can be applied as intended, with little enough artifacts.

Thanks a lot again, Terry!

I have to edit my default presets, especially limit them to raw.

Same problem with default presets.

Yes, I would say 99.99% RGB levels does nothing with jpg, but there was 1 image where it became a lot better. So is there a reason to deactivate it if nothing changes? It is trial and error.

With raw photos RGB levels often change nothing too, but some are a lot better then.

That happens often with these cathedral pictures. Is it a bad idea to use the tone equalizer to correct it then? Or is it better to deactivate agx?

It is a lot better than mine, and that is what counts

I need time now to play with the things you are doing.

Is there a misunderstood? Darktable processes the modules in a certain way and doesn’t care how it is edited?

I am not familiar with masks. I saw some videos with parametric masks. I fear I will use masks when 10000 photos are corrected. Personally I think better to correct a lot photos than a few perfect.

Learning masks is on my todo-list, but later.

This is an image for learning, the alternative is to delete it, it is not impartant.

How?

I need a format for 16:9 TV and 2:1 mobile phone. I know the perspectuve correction looks sometimes strange, but I think it is better than a black frame.

Exactly, there is no need for drawing. I want to know what it is possible with darktable. The result counts. Important is to know which modules are more or less useless with jpg, if they are not blocked already.

“useless” is depending on your editing - you just need to be aware that you won’t get more detail info from a limited input even it’s handled in an unbounded pipeline.
As during raw processing you need to know the purpose of a module - if you expect a module to be a jack in all trades, then this is a good way to shoot yourself into the foot :wink:
darktable modules are designed to solve specific topics:
a tone mapper doesn’t make sense if there is no demand to map a wide tonal range into a limited - even if you can use it as a quite limited curve tool …

2 Likes

I am aware that raw is always a lot better than jpg. I don’t expect things from jpg which are great with raw.

But the fact is, the image is available in jpg only. This is the big discussion between me an my family.

At the end I have to explain my family what can be done easily and fast. They prefer to delete a photo before they do spend a lot time with editing. It is always a question of quantity and quality.

So the goal is to get a better result, but not perfect.

You talk about default presets, but I am not sure what you mean. I apply a style to my raw files which I have designed to suit my camera. Are you applying a set of modules by default to all images? In this case the modules needed for a RAW file would be unsuited to a JPG from your phone.

I moved AgX by dragging and dropping it above the shadow and highlights module. I only do this with JPGs and not all the time. If I place shadows and highlights before AgX I will never touch the highlights but will brighten the shadows. If I am using shadow and highlights module to do both highlights and shadows I keep it above AgX. I am really only using AgX similar to levels to get my black and whites set nicely. I was surprised this module works so nicely with JPGs since it is primarily designed for RAW files.

I apply local contrast early in the editing because it will always have a strong influence on the final look. Yes, DT processes it late in the pipeline but that doesn’t mean I have to work in the same order. I like the extra clarity it brings to most of my images. It also adds a degree of perceived sharpness.

Do yourself a huge favor and learn to use drawn and parametric masks. That is the reason I selected DT over RawTherapee which is also an excellent program. Learn to drawn a mask around a specific region of a picture to lighten or darken it, but feather the edges to get an invisible transition. Load my xmp file and look how I used drawn masks on a number of modules.

As for comments by others about GIMP, it is an excellent program for editing JPG images if that is all you are doing. DT and RT are the programs of choice for RAW files, but they are also very suited to JPGs if you are comfortable with DT. However, if you are only ever doing JPG images I would recommend GIMP. I personally use GIMP for photo restoration work using Tiff files from my scanner. But all my RAW files are in DT.

Color balance RGB is one of the best modules to use with JPGs as well as RAW files.

Non-raw images

While DT can edit non-raw images such as JPGs many of its modules are optimized for RAW images. JPG images are unsuited to large exposure adjustments, and this is why RAW files are preferred for image editing.

  • The exposure module is generally unsuited to non-raw images.

  • The RGB curves and the tone curve can be used to adjust exposure by raising or lowering the middle of the curve.

  • RGB levels can adjust exposure by moving just the grey point handle left or right.

  • The tone equalizer module and/or the shadow and highlights module can be used to adjust exposure zones such as the shadows or highlights.

  • Color zones module can also be used to adjust lightness of individual color zones or the whole image.

  • The AgX module works very well on non-raw images to adjust brightness and contrast.

  • The shadow and highlights module works well with non-raw images

  • The color balance rgb module is good for exposure, contrast and saturation adjustments with non-raw images.

I’ve read your post in this thread and others. It seems that you want your family to value the pictures they take with their cellphone as much as you do. You want to in an automated way (auto presets) to further improve a camera jpg. But, even the folks that took the image, don’t care that much about it.

I would consider just letting Google photos do some AI magic. I don’t think dt is not the right tool for what you are after.

2 Likes

Working with this scene I encountered many firsts for me, so I don’t have a full edit, just experiments w.r.t. color and lighting. I learned a bit.

If I understood correctly, the ask is how to reduce the worst offenders in a non-raw image, without having a lot of editing to do.

The prime issue (for me) is that this scene has different light sources in different parts of the image, challenging in any case, even more so in JPEG (or so I heard, because I’m a newbie and so far may play-raws submissions were only in landscapes or nature, without artificial light).

My attempt:

  • non-artistic crop, just to focus on one problem area
  • coarse drawn mask (would need refinement) for top area, inverted in other modules for front
  • parametric masks for the front, to select the gold and the grey parts, respectively
  • The corrections themselves I used only color balance rgb.
    • A bit of vibrancy at the top to highlight the gold, and than work on the front so that the whites and gold colors match (better).
    • A bit of half-hearted contrast reduction in some parts.

Only partially successful, or maybe not at all, honestly. I could spend another hour tweaking it, but I definitely lack experience to get good results, so I stopped.

cathedral1_02.jpg.xmp (27.4 KB)

I think your recommendation is a good one.

A lot of photos have no real function other than be a visual diary entry, calling good and bad times into memory when we want to. As long as it invokes the context for those who were there, any technical aspects are irrelevant.

I myself have a few such photos, too bad to print on even a postmark, but cherished because of the moment when they were captured. And I only regret not having more such “bad” photos.

2 Likes