Use older color science version by default in filmic rgb

Software darktable
49

Thank you, I’ve got it. The paragraph 5:
Saturation: Your image will probably not look very colorful at this point. You can adjust the global saturation of the image using the color balance rgb module. The “basic colorfulness” should provide you with generally-reasonable defaults, but you are encouraged to experiment further with these settings as required.

The phrase “Your image will probably not look very colorful at this point” is not correct. It should be like that: “Your image colors look dull and unnatural at this point. You should adjust the global saturation of the image using the color balance rgb module. The “basic colorfulness” should provide you with generally-reasonable defaults, but you are encouraged to experiment further with these settings as required.”

The difference is obvious and shows that module should be enabled by default.

And the same is for the default filmic RGB settings.

50

Here is the issue: some people like to start from a pre-cooked image and some people do not. I happen to prefer to start with the captured data and gradually build my way towards a final result (I’m even considering disabling filmic by default). We shouldn’t change our approach just because “everyone else does it” or just because “new users give up quickly without it”. Eventually those new users will become experienced and IMO the best way of learning is doing – pre-baked presets discourage learning.

What are the practical ways to keep everyone happy?

We could have lots of configuration options cluttering up our global preferences window (to enable/disable modules by default), but honestly we have too many of those already (I think we plan to remove the sharpen preference in the next version) and most of them would only apply to RAW images (so extra code to handle non-RAW).

We could turn things on by default and make others create a preset to disable them but then, even if they never use that module it would still show up in their history in the disabled state (because that’s how presets work).

Finally, we could leave most modules off by default and let the user choose – those that want the modules on by default can create a preset and those that don’t can leave things as they are, with a clean history stack.

The last option is the current state-of-play and in my opinion it’s the best (or least bad) of the three.

4 Likes
51

If you deal with many images (a typical model shoot, daily street photography or landscape trip photos quantity is about 100, or even more by an order of magnitude), a good pre-cooked image is a must.

52

So you write a style for your pre-cooked image. But no need to impose it on everyone else.

4 Likes
53

Or prepare one image, and copy the settings to the others with similar conditions.

4 Likes
54

I can say for sure for many photographers - and I know a lot of photographers - most of them prefer having good looking image out of the box. If you want, I can ask about 40-50 persons and white down here the voting results. They do not use darktable, but it is the common idea.

You see, even choosing from several “duplicates” requires their adjustment into some good picture to judge. And it all requires some “default” style.

55

Out-of-the-box means jpeg, not raw (the box is the camera, not darktable).

3 Likes
56

I’ve started and deleted about 5 replies over the weekend, none sounded very constructive. But abruptly I remembered my doctoral dissertation from 25 years ago on “software volatility”, the topology of how software changes over its life, and it occurred to me…

darktable is a fully functional raw processor, contains all the tools to convert raw camera data into pleasing renditions. And, most of those tools are state-of-the-art now, with no equal in most of the other available softwares, commercial or open-source. This thread is about configuration, how all those tools are ordered and parameterized, particularly a “default” configuration. It may be there is a need for the development and “marketing” of darktable configurations, named files that set up default ordering and parameterization.

I don’t know if/how darktable supports configuration files, but in my linux-oriented thinking, it’d be a directory, e.g., ~/.local/share/darktable/defaultconfig, and one could drop any file named *.cfg in that directory and darktable would start up ready to process files in the workflow defined therein. A few possiblities come to mind:

  • ACES.cfg for an ACES-oriented IDT->ODT workflow,
  • scene_referred.cfg for the “thinkers” (see Thom Hogan’s most recent missive, at https://bythom.com/newsviews/thinkers-versus-non-thinker.html)
  • LR-like.cfg, for the “nonthinkers” (read his post, it’s not as offensive as you might think)
  • etc…

That way, anyone with fundamental understanding of their computer’s file directories can readily determine what default configuration in which their darktable will start. Heck, that file name should be displayed in the dt banner or somewhere; in rawproc I show the loaded rawproc.conf path in the About box.

My dissertation was about how to find the things that will change in a software’s life and making them changeable in the original design (spoiler: it’s the user-facing things that change the most, go figure). From a distance, it appears to me that darktable is readily organized to support such; really, this thread is not about changing darktable, its about configuring it…

6 Likes
57

I shoot RAW from my first digital camera - Minolta G500, which was enabled through the hack - a combination of buttons entered engineering menu, and RAW was there. And I participated in decoding its RAW file format. Since that time I have been shooting RAW only.

I’m pretty sure shooting RAW is not the goal itself. The goal is to obtain an image you (and others) would like. RAW allows one to obtain maximum from an image. It allows changing WB, exposure (even applying -EV to overexposed images: RAW is underexposed comparing to resulting jpg for about 1 stop), use tonal adjustments on a high bit image without banding etc. It allows using better image processing algorithms than the built-in camera are. It allows using LUTs for creative effects. And so on.
But. All the above mentioned actions are required if one is not satisfied with the image he obtained with RAW processing software “out of the box”. If one obtain results with some default RAW processing settings better than out of his camera, maybe spiced with some additional plugins like NR, local contrast, etc., all that images doesn’t require additional actions. They are already good. And they are so because one shot RAW, not jpg.

58

It might be possible to come up with a reasonable set of defaults that works well for your camera, but of course the same default might not work well for mine, so it’s hard to do default settings well. Even the default exposure setting is a compromise which might work well for some non-existent “average” camera but in practical terms needs tweaking for most real-life cameras. Default settings that are better than your camera can do are unlikely without a lot of effort, and afaik companies like Adobe only do it with significant effort and a lot of co-operation from camera manufacturers. Canon, for example, won’t even release the details of their RAW file formats to the FOSS community without a highly-restrictive NDA, so I’m not sure what we could do easily.

59

I understand that… actually I have a set of presets myself, one (or more) for each camera I use. Although recently, I’ve discovered that I can view and select the images in lighttable before doing any processing. Which means (in my case) that lighttable will display the embedded jpgs in the raw files. So I get to do my first selection using ‘pre-cooked’ images, then I start ‘developing’ that selection.
But I suppose if there was a default ‘out of camera’ look maybe I wouldn’t bother with this workflow.
On the other hand, if I wanted that I might just use RawTherapee, ART, or even… a commercial alternative (aaarrgh!)
All of which do give a ready-made ‘cooked’ look.
I guess what I’m saying is that while I do see the @BigSerpent 's point about a good starting point, and also it’s value for beginners, I’m not sure that it’s something that fits in to darktables ethos. On the other hand, maybe if darktable had an option be configured this way it wouldn’t be a bad thing, but then I’ve no idea how much time and effort from the developers would be needed to make a change that might have no great benefit for the majority of it’s user base.

Good luck!

60

This discussion in some form happens over and over. It always comes down to expectation for some. They look at raw files as JPG+. Without the camera that shot the raw there is no idea even what the JPG “should” be only the camera maker has that formula. Even if that information was available it would then have to be mapped to the DT tools. Others look at the image as a blank canvas and are happy to go from there. I think with the ability to use preset and styles you can speed up a custom starting point or go the base curve route even make your own. Maybe a point of capture for users to share their idea of good defaults for a particular camera might be of use and comfort to some.

Filmic is just another module. We can use it or not but it seems like people talk like filmic=DT rather than seeing it as just another module. I don’t apply it by default on purpose. I want to see and evaluate the image without any particular manipulation so that I can see what happens when I do apply it. I often find I can get better highlight details if I don’t have to use it but I also often like the global result I do get with it so I make a decision. I don’t think defaults will help that decision. Knowing the tools and how they behave and the various limitations and caveats is all I can do as the user. With use and practice the workflow evolves and the tools become familiar…

2 Likes
61

Do you deal with a lot of images in one sitting? Say low hundreds up to a thousand? I’m a medium type shooter who only gets into the thousands a few times a year but frequently deal with a few hundred images that should become say 25 finished images all in the same style.

I mention this because your post above makes it seem you’re more of a low volume image crafting type of photographer. Now just saving some styles, and personal default will enable a quicker workflow but understanding the issues people tackle is important to understanding what solutions are appropriate and why they ask for certain features and workflows.

62

I don’t and so you are right I can’t speak to that sort of volume… I would say though guessing if it was me I would be shooting raw and jpg and identifying my keepers from the jpg images and then diving deeper to get my 25 keepers and then doing my processing … Do photographers actually process that many raw files or do they at the very least look for key images and then apply that editing to the sequence??

Perhaps it would be good to hear from people that shoot that many images in a session using any software…

63

That is a reasonable goal.

The problem is that it is a technically difficult goal. I mean very, very difficult. There is a big difference between light measurements made by a camera, and colours as they are perceived by our human visual system. For example, the physical light we receive from an object is not affected by the background behind the object, but the perceived colour of the object is affected by the background. A common editing goal is to change the image so it resembles what we saw with our eyes and brain. We transform physical measurements so that the image is perceived to be like the real-world scene as we perceived it.

A much simpler goal is to make the image prettier (for some definition of “pretty”).

Even simpler: make the image like the OOC JPEG. This is fairly simple for individual images, but difficult to reverse-engineer the OOC algorithms and translate them into darktable (or other system) operations.

64

Kinda a hard statement. You say Darktable produces bad results out of the box , while the filmic defaults where for me a reason to take note and start using Darktable.

So, don’t confuse fact and opinion here.

I’m amazed with all the questions about ‘getting in camera look’ , while i often just need to set exposure , hit auto in filmic and look if i want to tweak saturation/colorfulness in some way to be ‘done’. Posts where people ask ‘how to get this look’ for me seem to always answer with ‘load, set exposure , set colorfulness and done’ ??.

Now, to be honest , for me it isn’t really the V5 / V6 change . It’s the change to maxrgb that trips out most people . It now really tries to preserve color even in very bright parts. And this is just a case of 'sometimes you want this , sometimes you don’t '. It’s awesome with skies and portraits , it’s maybe not what you want if you want specular highlights to show contrast and turn to white.

Yes, i do have some pictures where I like the V5 defaults more compared to V6 defaults… But they are few , and when noticing it afterwards I can almost get something out of V6 that i likeas well (most often changing preserve chroma mode , and / or playing with the latitude slider ) .

I didn’t read it properly, but ‘in my mind’ V6 is now even better than V5 in preventing hue shifts and staying true to the resulted color . This changes how highlights look , which can trip people up . The tool is made by the author to fix certain issues and to be a technically correct tool, that you need to master to get a desired output . It never is and will be a one click solution.

V6 is technically more correct , so it’s the new default. That you (and others , and maby most) like other results with fewer clicks … Setting up a default more suited to your liking is easy enough.
(But like priort i never noticed filmic has special auto settings , i always hit the auto button to get a start point after my exposure and tone equalizer has done its thing )

1 Like
65

I shoot occasionally couple of hundred pictures and process small set from those. I have been using DT for this under one year now. Before that I used it only every now and then.

My current process for this is following with scene referred workflow (this is still evolving as every day I learn more and more how to use darktable more efficiently):

  1. Created presets to enable only for my Canon cameras: demosaic (AMaZE + VNG4), diffuse or sharpen (sharpen demosaicing (AA filter), color balance rgb (basic colourfulness: standard), denoise (profiled), chromatic aberrations
  2. While on lighttable I hit ‘f’ for sticky preview mode, and scroll through photos. Set one star for photos I want to look closer in the darkroom and reject bad photos.
  3. Next I filter photos for one star and go darkroom and hit ‘F11’ to go full screen, Tab to show only the image, and ctrl+b for color assesment view.
  4. Couple of weeks ago I bought Loupedeck+ to make my life easier for this step. I have set exposure and master tab sliders from the color balance rgb to it (had to create own labels for knobs). Using only these 13 knobs I can change exposure and all chroma, saturation and brilliance sliders, while I look one photo at time in full screen color assesment view. I have also set up color zones tabs on the loupedeck+, if I need to make some lightness, saturation or hue changes for specific colors.
  5. I use some 30 sec to 1 min turning on knobs in loupedeck+ for each photo and if it looks like a keeper I set two stars for that.
  6. Next step is hit to tab and set filter for two star photos.
  7. Now I have only candidates for keepers and go in lighttable to see what kind of set images I have selected for keepers. Might drop some photos back to one star if I have too many similar photos or go to look for some more if something is missing.
  8. Then I go back to darkroom and make final tuning for the photos. First I check white balance in color calibration and after that I usually fix horizon with rotate and perspective, crop photo to my liking, set highlight reconstruction if there are blown highlights, probably make some local fixes with masks and might remove some distracting spots with retouch.
  9. Final editing step is to go filmic rgb. There I usually select first auto tune levels and after that make slight tuning to white and black relative exposure. After fine tuning I set three stars for the images I want to export.

For me this works much better than my previous process, where I made RAW conversion with Canons DPP and final photo editing and exporting in Gimp. I have also used Lightroom previously. For me darktable gives much better final images than those tools I have earlier used. It is just amazing how powerful darktable has come in resent years.

Here is link to example set I have done few days ago: Fin5 2022 orienteering photos

6 Likes
66

Then provide those settings. You said you’ve contributed before and it if this task is so easy, as you say, then please provide those settings to the project.

I’d bet that it will cost you at least some time, but what do I know?

4 Likes
67

Nice photos…very “realistic” I really like the one with the graffiti on the wall…

1 Like
68

Here is a sample style for FilmicRGB V6. It produces results as close to reality as I managed to get. It was tuned for my Sony A99, so it can differ in exposure for other camera manufactures.
Filmic_V6_sharp_highlights_Standard_matrix.dtstyle (1.7 KB)

1 Like