if new users expect darktable to be a tool like the green rectangle of the camera - they need to learn: With great power comes great responsibility
darktable gives a lot of room to do things other tools won’t give - including the ability to shoot into one’s foot 
1 Like
but using a sports car instead of a automatic gear car where you just need to steer and control speed with limited speed or a bus where you just need to step in and let others drive for you
Yes, I heard that. That’s not a proper excuse for not managing light in a user friendly way.
what is user friendly if control is fully given to the user? Is a knife to be seen not user friendly if it cuts into your finger? Or is it you that doesn’t make proper use of it.
1 Like
Fair enough. Then again, my initial remark was in reply to the quote. If you feel the interface or behaviour of dt is incorrect, you have two options:
- convince one or more of the developers something has to change (and ideally, how it should change)
- code the changes yourself (up to forking the program), this is basically what Aurélien Pierre did when he developed filmic and several other modules (and forked to create Ansel).
In this particular case, the best you can hope for is a small change in color balance rgb. Even that is not guaranteed, as the different parts of that module can interact, so a simple soft limit on a slider may be either not enough or too much.
There is (in my opinion, of course) no reason to change filmic in any way to solve this problem.
At the end, darktable gives you a lot of control over what you can do with your images. That also means that you can use settings that are theoretically incorrect, but in some cases allow an effect you want. Or it blows up in your face…
Martin, your illustrations are so vivid. I love fast cars too, but if the nuts to one of the front wheels aren’t tightened, it will kill me at one point. But if I care about my well being I am urged to take the bus.
I am out.
I think I made this point: Even if the user obey the visual guidelines given with DT’s waveform, inappropriate light emission can occur. That would be recognized as a problem, if it happened with my new sports car (Martin).
2 Likes
it’s not about untighted bolts - you’re just driving too fast into a narrow curve without experience how to control your car if there’s no electronic stabilisation system aboard…
Yes, you have made your point and your frustration is understandable. The trouble is, this is a user forum, not the darktable issue tracker (even though developers do hang out here).
To propose/request the change, please use the following link:
Better yet, comment on the already open issue mentioned above:
With enough info and convincing arguments, the developers may take it up(if you are not willing to take it up yourself). Please understand they may decide against it, or may have other properties.
Their solution/workaround was to disable filmic’s highlight reconstruction by default.
You can see that some have picked up the discussion there, prompted by this very topic:
I’m involved in motorsports so I like these analogies. I’m familiar with darktable by now, and I do think that it has some quirks - but I accept those in the same way that my brother accepts a harsh ride in exchange for handling, lots of engine vibration in exchange for optimum response from the drivetrain, and so on.
Ideally, yes, one would have everything just right, but it’s not a perfect world…
I think it should also be remembered that darktable hasn’t been around all that long - dunno when 1.0 came out, but in comparison with many things I use it’s pretty recent, and of course it’s developed by by essentially volunteers - same as many of the rally cars I’ve been around…
Don’t know where I’m going with this really - just my 2c worth!
1 Like
I played around with a recent image I had from a vacation.
My normal edits with filmic HR off.
Changed the brilliance highlights to 32.61% (until it showed super nova, no issue at 32.50) and turned on filmic HR (default settings).
Adjusting the white fulcrum in the color balance rbg to +15 had not effect. Going -10 made it worst. Auto picker picked +2.84, but it looks the same.
But the change that did managed to had a positive effect was reducing the filmic HR transition to 2.
I can even use 100% in brilliance without super nova.
While not a fix, it might further help reduce the occurrence super novas(if someone turns ON flimic HR and goes too high on brilliance highlights). I tested in a few images I had. If someone can confirm that using 2 for transition helps on their images too, then I can do a PR.
3 Likes
10 years.
1 Like
LOL watch it, we could wind up hijacking this thread with analogies!
1 Like
My problem with this “feature” is that it doesn’t always show in the fit-to-screen preview.
You have to zoom in at 100% to catch those nasties. I ain’t no pixel peeper, so rather often I don’t look at the full preview.
As a result I have exported hundreds of images that suddenly had white blotches seemingly out of nowhere.
Well – if software does not behave predictable, then change your workflow. That’s what I did and I haven’t seen a supernova for lightyears. I get my brilliance somewhere else.
Feature – yes, sure.
Broken – totally.
One extra wrinkle on this issue is, as you’ve noticed, that what you see in DT may or may not be what you get when exporting. Seems that the filmic HL reconstruct algorithm is different on export (probably for performance reasons).
That would be my guess, too.
Data gets scaled down, then sent through the pipe.
Depending on wether that one pixel to create a supernova – and one is enough, I’ve tried – is among the scaled down data will determine the outcome of the preview. But scaling algos that emphasize extreme values usually look really ugly and pixelated, so the choice is not that wrong.
It’s just the tool chain that allows for hidden nasties.
As I see it, the technical aspect is correct while the usability aspect is all wrong. But usability was never really a strong suit of AP, anyway. Probably wasn’t scientific enough. I had discussions about that with him way too often.
Having to zoom in to see an image matching what will be exported is not specific to darktable.
http://rawpedia.rawtherapee.com/Editor#The_Preview_Panel
Note that the effects of some tools are only accurately visible when you are zoomed in to 1:1 (100%) or more.
That is of course true.
Although with the supernovas we are talking being off by a number of magnitudes.
A single pixel can blow up to a thousandfold of its original size.
Which makes the whole thing highly unpredictable.
Not what most of us would consider a reasonable margin of error.
Keep the Brilliance grade below 20 seems reasonably predictable.
DXO PR3 files open directly in DT - at least on my Mac anyway https://youtu.be/cSbVs-W5B7o just don’t allow DXO to do any sharpening, because for my taste, even a “soft” setting is too harsh.
The trouble is that downscaling implies a kind of averaging, and it’s often single supernova pixels that explode. Those single pixels, when averaged with their neighbours, no longer produces the effect.
Limiting brilliance to 1.2, and setting the correct white fulcrum prevents pixels from going supernova.