This whole scene/display referred thing is the point where DT became usable to me and where I was happily able to switch over to Linux full time. I’ll not pretend I even half understand it all but the new modules and following peoples advice gets me nice results which I couldn’t get before very easily, so it would be interesting to see you expand on your comments.
3 Likes
He’s just saying that they’re both complicated pieces of software that you have to take time to learn how to use.
4 Likes
Sorry, I was a bit vague…
I was just comparing apples with bananas. They’re both fruit!
Yes!
I loved that description about darktable, particularly about the possibility of sending the image to an external editor or zooming with the mouse.
It would be excellent if users would take some time to learn about the work philosophy, the program interface, what functions each icon has, read the manual and then draw conclusions.
Anyway, I am also an average user who learns a little more about this program every day.
2 Likes
I’ve always found RawTherapee to be pretty straight forward — perhaps I’m not using it right. 
I’ve never used darktable, but for no other reason than RawTherapee does everything I need it to do.
If life’s taught me anything, though, it’s this: everything is complex until you learn how to master it (you have to walk before you can run — but before that, you have to learn to walk).
2 Likes
darktable is so good I no longer need photoshop
But it did take much longer than a week to get comfortable with, and after a year I’m still picking up new things.
1 Like
To answer your requests:
- It would require a thorough rewrite of our processing pipeline from fixed to dynamic. Not going to happen anytime soon I think, although on this particular point I think RT is behind darktable/vkdt
- Currently only possible in a select few modules. No general masking is planned. Not sure how hard it is.
- Same as above. I believe ART has a good implementation of different masks, but I would have to check.
Darktable folk: let’s keep this topic focused on RawTherapee (even if you feel Andy judged darktable without knowing it in depth). I’m sure if he ever becomes more interested in darktable, he’ll open a thread for that.
3 Likes
Hello , and excuse my (very) bad english.
What Andy puts forward in his contact with DT – as well as RT can be summed up in 2 points, which I will explain next.
a) As we say in French: "Une question n’est difficile que si on ne connaît pas la réponse " : “A question is difficult only if you don’t know the answer”
b) Complexity (I will come back to this concept) is essentially a matter of habits and learning.
As some probably know - this is personal, but every developer/user will find similar quirks:
- I am old (75 years old), very ill, this last point has been a big handicap for the last 3 or 4 years.
- Moreover I am not a computer scientist but self-taught, my code is bad (it is necessary to optimize it…), and I have trouble understanding the code of others (that of DT, to a lesser degree ART…) .
- On the other hand, my training as an engineer and manager, then as a sociologist and ergonomist brought me 2 things:
** a permanent need for innovation
** knowledge of individual and collective reactions to usual or new situations.
The Need for Innovation:
- it is a psychological and sociological behavior that we have or that we do not have - it is neither better nor worse than the “application” behaviors where the individual (the group) feels more at home ease in respecting procedures, a known environment.
- Rawtherapee was an opportunity for me to flourish. I seek no profit, no glory, no ego. I do this for fun and as a hobby. This led to modules which at the beginning and still now have shocked or been rejected (most often by cognitive misunderstanding or ignorance of my purposes). I would cite Retinex, Wavelet, Ciecam, Abstract profiles, White balance auto by temperature correlation, Local Adjustments, etc. These modules may seem complex, they are essentially for 2 reasons:
- the vocabulary which is that of the researchers, for example that of Ciecam which is partially found in that of DT
- an ill-adapted GUI.
Once the user has passed the a priori, understood the concepts and learned to use them (with the help of documentation, videos, forums, etc.), he often becomes an unconditional supporter.
Researchers (sociologists, ergonomists, psychologists) distinguish 3 stages in the behavior of individuals faced with a system (this is particularly useful in nuclear and railway security, etc.)
1 - the gestures and behaviors are automatic, the person has acquired experience and masters the system well, … but what happens with a new procedure?
2 – gestures and behaviors require calling on the memory of previously learned but known situations. Again, what happens when the system has not been used for several weeks or months?
3 – gestures and behaviors require the use of knowledge (when it exists). For example when faced with “wavelets” the reaction might be “what the hell is this” / “I don’t understand anything”, or “I know but why are we using this here”, or "I know, I’m going to dig deeper . »
In general, faced with a new product, the user is in 3) above… And we often end up in a rejection. Experts speak of unlearning and new learning. This implies a desire for change and work…not easy. The system control step is 1)
What further complicates things is the graphical interface (in the case of software). For RT I took what was there – without understanding, I work like a monkey copying what works. The GUI is based on programming knowledge, the innovation has often been made elsewhere, but the language, the programming are abstruse.
A beginner will obviously notice that the GUI of RT, DT, Lightroom is different. This is obviously an obstacle to the newcomer.
The other obstacle is the understanding of the system: a systemic analysis (which has been one of my specialties to understand the functioning of companies, organizations, …) could facilitate this understanding, but the implementation with isolated people is almost impossible.
Going back to RT is to the underlying requests of Andy and others (requests which I respect), if I had been able/knew how to do them a long time ago it would have been done….Difficulties mostly due to the GUI . But also to the weakening of the team for a few months…
The comparison between the 2 products is also a matter of knowledge and understanding (to arrive at 1). The ideal would be a combination of the two… almost impossible. I know that RT has been inactive (in terms of official publications) for about 2 years, but that doesn’t fully explain why innovations in DT (scene refere, Sigmoid, new color science,…) are hotly debated on the forums and that similar processes on RT (Ciecam, Sigmoid, Munsell correction, Auto WB…) are currently in the shadows;
Jacques
14 Likes
How difficult would it be to implement 10 bit output? Is it necessary to port to gtk4?
re: 10bit. disclaimer: i last tried this more than a year ago with gtk3 and cairo. it was possible but with a lot of issues. some icons would render in broken (the reason why also libreoffice and gimp were borderline unusable in a 10-bit xorg) and cairo would slow down to a crawl. like draw a simple gradient in a not even fullscreen window would take 30 seconds (or use an 8-bit codepath). at the time i concluded for all practical intents and purposes it was not possible with these old libraries. i have no idea about gtk4.
With the exception of a couple of modules dt is actually very fast on my Win11 system, faster than DxO (which is very snappy) and about the same as Capture 1. I don’t think speed is the problem.
I think it’s more what Andy is complaining about: dt is so confoundingly complex and there are so many intertwined mysterious modules that a newcomer, even one like Andy who has years and years of experience and proficiency with many other developing programs, just sees a dense incomprehensible mess when he sees dt for the first time.
To really use dt near to its potential requires a person to spend a LOT of time and effort evaluating dozens of modules to find just the ones that he/she can understand, that are compatible with each other, and that do what the person needs and wants, all in an efficient workflow. Is this really necessary? Is this fun yet? No wonder Aurelien left.
Is dt this convoluted only because the master plan is to retain backward compatibility at any cost? If so, I think the powers-that-be should take a poll and see if that’s really what the majority of users want.
3 Likes
Well I heard that one some Mac systems there is no gpu acceleration for dt at all (no opencl), otherwise I can’t say anything about Mac because I am not a Mac user.
I think I am one of the few people that tests software on both Windows and Linux and even though dt speed may seem ok on Windows, at least a few months ago there was a serious problem with some modules such as denoise profile in combination with Nvidia drivers - those modules were faster without gpu acceleration.
ok. seems like that opencl bug in Windows was fixed…
Precisely summed up there Eddie, thank you 
As I mentioned before, this latest edition/release flies on my Mac. So I’m embarking on a fast learning curve, but immediately I can see things that really make no sense to me - as a photographer. And that’s the important thing, photographers are going to use it, but ‘over-complexity’ causes headaches, and headaches cause loss of patience with many photographers.
Just found that you can get DT to read external camera profiles, but only .icc and .icm profiles. It can not read the more useful .dcp profiles, which is a shame as Raw Therapee has no problem reading them.
1 Like
On the module side, should the base configuration shipped with DT show upfront less modules and more thoughtfully selected one (that make sense as general purpose solution) ?
Being on the other side of the fence (already used to DT and having already selected my tools/modules) I do not feel this nuisance as the modules I do not use are masked and I’m just occasionally bothered by some modules being deprecated (when tone-mapping module for example went away) or grand new developments : when A.P introduced filmic, colour balance and colour calibration, this took me quite some time to get used to, even though now I find these wonderful…
I must state to contextualize that I practice photography as a hobby and that DT is the first development software I ever used (since like 10y ago) and the main one I still use. I’ve been very puzzled and felt constrained when I tried a few years ago Lightroom (no modules duplication, no masking tools) as it was at that time presented to be the industry standard. I heard that it as been added since.
1 Like
Lightroom does indeed have masking features which are functional and very easy to use.
You can utilise as many masks as you like, and add, subtract and intersect masks/selections as much as you like.
It does not allow module duplication.
The only reason I started to use Raw Therapee was Adobes faulty demosaicing algorithm which adds pre-sharpening halos to certain boundaries in an image even with all sharpening turned off.
You can see the problem in my YouTube video here https://youtu.be/-Zz_t_pRdp4
1 Like
While it would be nice to have multiple graduated filters in the standard module, note this can already be done with LA : there’s graduated filters in the advanced version of several modules (tone equalizer, color & light).
I think it would be nice to have a dedicated graduated filter LA tool. Same with the tone curve, I doubt many people realized you can use a tone curve locally.
https://docs.darktable.org/usermanual/4.0/en/overview/workflow/process/
This page exists to avoid exactly that. There are many examples in the forum of simple processing workflows, and in the end after a bit of reading, it’s not even close to “dense incomprehensible mess”. Seems like criticism in bad faith, given how people already know darktable is an unorthodox editor. Any user can reach this page with 3 clicks in the manual. Overview > Introduction to dt’s workflow > Processing.
3 Likes