We have a problem - what is it about Ansel and AP? (Edit - concern fully resolved via respectful dialogue)

Software
21

@OK1 if AP creates some new modules or improvements in Ansel I for one would like to hear about this. He is obviously talented and may have some more editing tricks up his sleave. Since AP has chosen to leave this forum I depend on someone like yourself to inform me of any new tools he creates. For me Ansel is missing some of the important tools that other developers have created because one person has decided what tools I should or should not have in my editing tool kit. AP could have an Ansel update here but has chosen not to. Let’s not be critical of those that have chosen to remain here. There are many people that are appreciative of the modules, tools and concepts AP has created, but equally there are others working very hard to make DT the great program that it is. For me the single most important feature in DT that won me over from other worthy contenders such as Rawtherappee is the drawn and parametric masking capabilities. So my thanks go out to the creater and maintainers of those features.

This quote seems at odds with the concept of FOSS. I for one don’t want to pay Adobe a subscription to use DT. But if AP branched out and created his own commercial product or wanted to go work for Adobe I wish him well and believe he would be a valuable contributor to those commercial enterprises.

1 Like
22

If they’re that good/useful, I’m sure someone would port them to darktable.

3 Likes
23

That is my hope. There is no doubt that he has created some excellent modules and the maths and coding behind it is beyond my understanding. BTW, initially I liked the more intuitive RawTherapee layout. RawTherapee also has some very nice tools. DT seemed overwhelming but somehow I fumbled through with getting my head around the basics by watching some videos, reading the manual and just somehow learning. Once I discovered the masking capabilities of DT I was soon devoted to the program. The masks are so much better than Adode’s Lightroom. I am also seeing the evolution of the image management capabilities in the lighttable tab. Some of this is just better understanding on my part, but some is improved design by the developers. Initially I looked at DT as a edit only, but now I am looking at it more and more as a way of importing and sorting my images. It is not quite as good as Adode’s Lightroom in my hands, but with time I hope to get better at using it as a photomanager.

24

One way to find out without relying on others is to create an account on Github and following him: GitHub - aurelienpierreeng/ansel: A darktable fork minus the bloat plus some design vision.

I do this for many projects I’m interested in. It’s a great way to stay up to date on developments – you don’t need to know how to code, just read the issues and comments that catch your interest.

2 Likes
25

For example a long time ago, organisations like Oracle invested in Java, and much of what they have done since is still available as open source, and Linux would not be what it is today, without the contributions of commercial businesses like Microsoft/Oracle/IBM etc who have teams of people who work on aspects of Linux, for their own benefit, which spillover into contributions to open source.

Linus Torvalds - open source developer earns 10 million dollars a year, from his involvement in open source.

So there is a potential for open source software and ethos to remain open source yet commercially viable enough to earn the developers, a good comfortable income, and devote quality time and thought to it, without having to worry where their next meal or rent will come from. Best of both worlds - open source and a source of good living, at the same time.

26

Now back to pixls.us. Hope we can open up threads to discuss the merits and challenges with Ansel itself, the software, not the developer, as I think this may have value. For sharing thought amongst ourselves, and it does not matter that AP may not be on the pixls.us forum, as I think that is not relevant.

We have threads here which discuss imaging software that has no support anymore or has not had any development in many years, i.e extinct, and the developers/contributors are not probably aware that the software is discussed on pixls.us, and even if they are aware, they do not join in the discussion. Who knows, the developers may even be dead, for all we know. !!, but its still worth chatting about these tools.

So can we open threads to talk about Ansel itself, the open source fork of darktable.

While its possible to do this in the web site setup by AP, I cherish the opportunity of a more open objective discussion here. For bug fixes, I’ll do the right thing and highlight these on the relevant Ansel github site, but nothing wrong with talking about these here also.

So Ansel the software becomes just another product we discuss, like Darktable or RawTherapee or Gimp

27

Yes

28

The creator of Ansel does not have to be present on pixls.us forums , for us to discuss the tools they create.

I have no idea who the developers of RawTherapee are, because the discussion of the tool has not to my knowledge ever had a bearing on the tool creators, i.e the discussion never seems to get down to the merits and merits of what pathway the tool should take, in the manner in which AP, had done, by making his opinions widely available outside of the darktable github.

So we do not need AP to be here on pixls.us, to discuss Ansel the fork of darktable.

On that note would it not be so nice to have a fork of Ansel that adds back in the Sigmoid module. !!, so we have the best of both worlds.

When I have enough time, I must learn how to at least be able to do something like this, oh that would be the 11th wonder of the world. Ansel fork with Sigmoid added back in…

As the creator of Sigmoid, it would be so easy for you to do - in the spirit of “openness” so your effort remains more available to all users of open source image processing tools, rather than restricted to darktable only.

The availability of such a fork of Ansel, may encourage AP, to keep Sigmoid included, in Ansel, so so that such a fork of Ansel, is no longer required.

29

I periodically check-in on the Ansel git. I saw some musing from AP about filmic v7. I think he is removing the norms and moving to a slider or set of sliders for hue preservation if I recall correctly so basically maybe something akin to how @Jandren handled the adjustment… I might be recalling wrong and until it appears he may do something else.

What should be noted here for anyone not following the DT git is how much “non-sexy” invisible work is being put in by devs like @hannoschwalm to enhance and clean up things like the opencl code and the pipeline processing. Also the great job on the HLR aligo to handle scene referred data modern WB and filmic combination… The default clip highlights and other methods created a host of artifacts in highlights and color with this workflow that so many experience and complained about and these tools have all but cleared up those crazy issues. Several others are tackling bugs and removing and improving the code… DT went at a break neck pace and now there is a focus on refining and optimizing the code. Maybe its not leading to a sexy module or fancy math but hard work and very important . Thanks to those that have been pounding away at this for some time now… I don’t understand much of it but I can only image how much time and effort goes in to the process… Thx

8 Likes
30

https://community.ansel.photos/

Why do it here when a dedicated forum exists? IMO it makes no sense to discuss Ansel here all things considered.

3 Likes
31

I wonder which companies would commercially benefit from darktable, sufficiently to devote full time developers to coding it. Lots of commercial entities benefit from things like Linux, Java etc. but a Raw development tool has (IMO) limited commercial appeal unless a company wanted to fork it and repackage it as a paid platform.

Yes of course we can, but this thread was started to talk about alleged mistreatment of a person and whether or not a previous thread (/threads) should have been locked. Again, start a new thread that’s just talking about the software, please. Just bear in mind that if you want to talk about Ansel with its developer, you cannot do so on pixls.

Aurelien’s objections were the primary reason for sigmoid not being included in darktable – it was only merged after he left the project. I don’t think there’s any chance he will ever include it in Ansel.

5 Likes
32

In the audio world there are two digital audio workstation apps - Mixbus and Mixbus32C, based predominantly on Ardour the most popular DAW on Linux. And they contribute code to the development of Ardour, and their own main code for the mixbus products, is available on github.

Thanks for educating me. I was not aware of this. I was a bit active on the thread where sigmoid was proposed, but that was a while back. I even for the 1st time in my life, bothered to install and compile Windows executables from github, just to check out sigmoid. But had to abandon this, cos I would have needed to repeat the process every few weeks, and it’s hard to maintain

Hmm it does sound unfortunate, but that is life… interoperability is a good feature to have. I come from more intimacy with the world of music production, where its so normal to have different plugins for things like equalisation, which all serve the same purpose and you can choose which one you prefer. So there is no need to be previous about these things. This is in spite of the fact that every plugin host in the audio world already comes with its own equaliser.

This is how sigmoid should have been viewed, as an alternative to filmic rgb and base curve. Looking back filmic did not replace base curve in dt, just became an alternative, for those who wanted to use it. So nothing wrong with having all three filmic, sigmoid, base curve and any others available.

The beauty of the menu management system in dt and ansel, is, the end user can configure/customise to leave visible, only those modules they use regularly, and the inclusion of sigmoid was not taking anything away from the benefit of having filmic.

Anyway it is what it is. We cannot roll back time. If I may be honest, I think the way to go will be to stick with dt.

I took a look at Ansel, and neither sigmoid or base curve is available. I prefer an inclusive approach, and admit carrying everyone along has its penalties, but this is a good decision, to add sigmoid to dt.

33

This is one of AP’s fundamental objections, and a common complaint from users: There are too many modules that do the same thing. I’m sure this is why he has removed base curve and won’t implement sigmoid – because they’re both trying to do the same thing as filmic

34

I do not think I will be starting any thread on Ansel here, and having understood the history that led to Ansel’s fork from Darktable, as a veteran of the IT profession, my thoughts are - support is 99% of the reason for choosing any tool. All the features in the world, without support - is nothing.

And this is the fundamental challenge that darktable solves - a team of people who will support the tool. I value this and for this reason, I’ll be sticking to darktable. Ansel would take me into a risk which I would not like to face.

Fundamentally without AP, there will be no further work done on Ansel… To commit to anything I need assurances that my investment in time and the learning curve will not be suddenly brought to an abrupt end because of the lack of continuity in the development of a tool, with dependencies on a single individual.

Thanks everyone.

2 Likes
35

Hello @OK1

If I may be honest, I think the way to go will be to stick with dt.

IMHO, it might be the best route to follow, in the end.

I am a RawTherapee user but I do admire the work done by Aurélienne, over the years, on darktable.
This being said, with dt you have much more developers working on the code whereas with Ansel, at present, you only have 1 (AP).

On top of this, if you read some previous posts of AP on this forum he himself considered VKDT (GitHub - hanatos/vkdt: raw photography workflow that sucks less ) as the most promising software for future developments. Therefore, it is a bit strange he is still working on his fork of dt instead of forking Vkdt… :slight_smile:

1 Like
36

I have been spoiled by the audio world, where there must exist well over 200 different parametric equalisers, and some of them as freeware, and some would argue they fundamentally do the same thing, yet they can all live happily in the same ecosystem.

I’m sticking with darktable, cos there is a far greater probability that whatever AR develops for Ansel, that has value to the wider community will get ported back to darktable, than the other way round where Sigmoid and the Base curve have already been removed.

37

I agree 100%.

38

Things have gotten worse after the split.

There’s a general thing with ethics in open source - taking someone’s contributions without appropriate credit is considered a no-no by almost everyone.

As a result, when the darktable team cherry-picks something from Ansel, they will preserve commit authorship so that credit is retained.

On the other end, apparently AP has routinely grabbed changes from darktable and not preserved authorship, effectively taking credit for someone else’s work. I’m sure one of the curreny dt devs can point out examples of this again (there were examples provided in other threads)

Funny that he keeps on ranting about “cargo cults” in open source, yet he’s on par with the worst of the cargo-cult-inspiring kangers in the Android custom kernel/firmware scene a decade ago. (People who gain a following of worshippers based on “OMG that kernel is so great”, when in reality, 90%+ of the kernel are things kanged without credit from others.). That’s another reason there is so much animosity towards him - he’s frequently guilty of some very severe hyporcrisy.

Of note, he’s one of the few people who ever solicited direct monetary contributions. Which reminds me of a statement made by one of the admins over at xda-developers lamenting the donation-seeker/cargo-cult culture that was an endless battle to try and rein in - “compeeeete for donations!”

9 Likes
39

Amusing. Entertaining actually. To become aware of the human side of techies. I’m sticking with DT, not for any personal reasons, but it’s the best choice.

I had a recent experience where I must have invested over 24 hours across several days, researching options to replace an HDMI cable used in my church to display messages, which had been run across the floor, and therefore a hazard. Then in a few moments, in conversation with my pastor, he had a bright idea, which could be implemented in no more than 30 minutes, which involved changing the position of the screen. And the cable would no longer run across the floor.!

Problems should be shared, and other minds can bring so much value, when we allow others to contribute to any challenge.

On that note I find it very instructive that when the technology was not “invented” or profited by them, great minds like Elon Musk want to put a halt to Artificial Intelligence. At the heart of all humanity are these self centered traits, which I dare say the open source community is one of our most altruistic examples of human collaboration, in spite of any shortcomings such as you have pointed out, the failure to properly attribute the contributions of others.

I love the way you write, wonderful use of language, you have.

1 Like
40

The problem is that a color pixel pipeline is orders of magnitude more complex than an audio pipeline. Audio is a one dimensional signal with only two parameters attached, the sample rate and the sample resolution (or bit depth or number format, however you like). Pixel data is spatially two-dimensional with typically 3 or 4 color dimensions, which are not necessarily independent. For the color dimensions there are many different fundamental representations (RGB(A), CMYK, YUV, …), and for each of these, many representations that detail the actual meaning of the data, such as color spaces and “gamma” curves. All this on top of the resolution per channel. All of this makes it IMHO orders of magnitude more complex to have a universal plugin system for image operations. Furthermore, different software puts different latency standards on a single image operation, which also prevents such a universal system – in audio, filters have way less data to process and therefore are more easy to design to a certain real time requirement. And there are more reasons. I was dreaming of a universal plugin system for a while, but understand that these issues make it close to impossible.

3 Likes