Darktable has a "neural restore" algorithm [0] in the development version (intended for midsummer release). Note:
- It appears to be an out-of-band pre-processing stage (run the image through denoise to produce an intermediary TIFF), unlike most other parts of the program.
- All AI features are gated behind compile-time flags which default to off.
darktable has supported Fuji raws since 2014! It currently supports the classic "uncompressed" RAFs, as well as the newfangled "lossless" (compressed) RAFs. I do not believe that it supports the "compressed" (lossy) format. So setting "recording type" appropriately on your camera is necessary.
I'm curious where the notion comes from that there is no support for Fujifilm RAF files, as I see this in a cousin comment as well.
> Together with all of American higher education, Cornell is entering a time of significant financial uncertainty. The potential for deep cuts in federal research funding, as well as tax legislation affecting our endowment income, has now been added to existing concerns related to rapid growth and cost escalations. It is imperative that we navigate this challenging financial landscape with a shared understanding and common purpose, to continue to advance our mission, strengthen our academic community, and deepen our impact. [0]
At the moment, this is about a week's work by eight authors. Others cycle/in out, of course -- this is a spot sample. They range from bugfixes to performance improvements to documentation to translation work. All what one would hope for in a software project headed to its bi-annual release next month.
There are many ways to develop, and it may be a bit cruel to compare a one-man show to a long-term international collaboration. But here are the recently merged pull requests from the software which is posted about in the blog post:
Page 1 of Ansel commits is by its mono-author from the last week. Page 2 takes us back to August. Page 3 back to June. I totally understand that good developers need to work carefully and sit on things, then release them in due time.
If we take a moment to page back to page 3, one can note that we're back to two weeks ago (rather than June). Steady work by a committed community matters. The log of work done is may be quite worth looking at, rather than incendiary blog posts.
What we're seeing here is one embittered ex-developer spending time/effort saying toxic things about their former collaborators in a public forum. At the same time the former developer is making a claim to such radical competence that they can somehow keep an open source project of considerable scope together -- one which until now was coded by generally at least half a dozen committed authors at any given time. The darktable project was originated by some quite bright folks. Others have cycled in/out over the years. Generally without too much drama. It's too bad to be giving so much attention to divisive claims.
> So which is more useful, one that doesn't even know there is a new album coming out, or one that knows what was its release date as of just a couple of months ago?
To semi-misquote Lewis Carroll: Which is better, a stopped clock or a clock which loses a minute a day? Carroll posits the former, as it is precisely correct twice a day. The trick, of course, is knowing for sure when those two times per day will be.
Of course one could argue that the hardcore Linux (er, GNU/Linux) document creator will use Emacs with AUCTeX (or canny uses of org-mode), with rendering via XeTeX/LaTeX/LuaTeX... Or markdown piped through pandoc, for those who want to take it easy.
LibreOffice, it's a slippery slope... Next thing we'll be using the mouse and ditching the tiled window manager.
The OP is burying the lede! The exciting news [0] is
> We are adjusting our engineering priorities for RHEL for Workstations and focusing on gaps in Wayland, building out HDR support, building out what’s needed for
color-sensitive work, and a host of other refinements required by Workstation users.
This is a long-standing and important effort [1] to make Wayland more plausible for image/video-editing.
Steve Roberts had a gig (maybe during the Winnebiko period) writing a column about his travels for CompuServe Magazine. In case you don't remember it, CompuServe was a walled-garden pre-AOL dialup internet site which included forums, games, and news. In this early-in-the-virtual era, subscribers to CompuServe also received each month, via snail mail, a glossy real-world periodical.
I sent Steve a fan email, and to my wonder, received a reply back. He struck me as a lovely guy, happy to correspond about the music and his general state of mind.
My memory is that an attraction of his column was that each was written from a different place, and described the adventures getting there, and being there. But then he got to the Florida Keys. And hung out at the Florida Keys for another column. And then hung out there for another column or so -- it was a pretty great place. And eventually, CompuServe dropped the column.
So nice to read the article on Steve Roberts, and fill out a bit these memories. Perhaps somewhere out there, someone has also written a memoir about CompuServe in its pre-Internet 1980s glory?
> I think the general trend is that actual useful applications are emerging from enormous models trained and owned by billion dollar companies only.
One way to think about it: Today's LLMs require incredible outlays of capital and processor power (and crews of folks with doctorates), such as billion dollar companies can provide. But how is that different from what Intel brought to commodity CPUs in the '90s/'00s, or what Nvidia brought to GPUs in the '00s/'10s? Or even what Cisco and folks brought to networks?
Though we may never design an artisanal CPU/GPU/router, we get to work with them every day to make things, and to communicate. These LLMs can be that for us at this moment. Let's go out and enjoy them, and see what we can make within their (vast) domain-specific capabilities.
Also worth checking out Alvy Ray Smith's 2021 book, "A Biography of the Pixel". It is (from what I've read so far), much more in the history/theory of technology field, examining the math/technology behind resolving continuous to the discrete (waves to pixels, vectors to rasters, etc.), and the associated cultural/business/entertainment shifts.
Not to be petty, but I see that to all appearances Smith is not, having given "Catmull, Edwin" about half a page of index entries. (The cover identifies Smith as "Cofounder of Pixar. And that there are index entries for "Pixar, origin and cofounding of, cofounded by Ed Catmull and Alvy Ray Smith" as well as for "Pixar, origin and cofounding of, not cofounded by Steve Jobs".)
An anecdote from someone in IT at a major university: The registrar has two employees whose sole job is to write SQL queries. These are to answer basic questions such as, "How many undergraduates are currently enrolled." And it turns out that this is a nearly impossible question to which to give a definitive answer.
I'm curious about this as well, but wouldn't be pointing fingers at the Wayland developers. See, for example, one of the most recent comments in the discussion linked to in the parent post, by Nilvus, one of the (wise and knowledgeable) darktable developers:
> The only thing is that color management for Wayland is progressing and far from being completely done. Even if work is quite slow, things seems to go in good direction. For correct and complete Wayland color management, we just have to wait again.
I can't find any reference to color management in the Blender meta-issue at https://developer.blender.org/T76428. For X11, I believe applications would have to manually determine the color profile of the display holding the current window, then query colord or and X atom to determine the profile. The application would then manually do the colorspace conversion. Does querying colord and making an in-application conversion works for Wayland until Wayland becomes colorspace-aware? Or if there are more wrinkles?
Which reads like a bad translation of a bad translation. Like the the old joke about the AI program which was supposed to translate "The spirit is willing but the flesh is weak" from English to Russian to English, and after the roundtrip came up with "The vodka is good but the meat is rotten."
Quite true. Best case, we're seeing DJ Spooky style culture jamming/remixing. But more likely it is as you write.
On the other hand, the market for stock photography was already decimated by the internet. Where previously skilled photographers would create libraries of images to exemplify various terms and sell these as stock, in the last decade or so, an art director with the aid of a search engine could rapidly produce similar results.
Indeed. And the artist/creative who can guide DALL-E or GPT-3 will be as exciting as the composer who can make a room dance with music from a synthesizer...or the artist who can make compelling images with a camera.
Indeed, there's an argument that the culture around anything (including software) is the main thing. And if the culture thrives, then that becomes the stable backup (or rather stable living vessel). Isn't there a Linus Torvalds quote that he has the world's largest distributed backup system for his code? But the point isn't that he bamboozled people into keeping copies of the Linux source everywhere -- it's that he started and oversees a living and helpful project.
I had a teacher who was known as a master printer. Students would ask, "How can I make a truly archival print?" His answer: "Show me something you make worth keeping, and I'll tell you."
My high school had a computer lab with a networked LaserWriter (this might have been 1987). It outclassed any screen in the room, so far as display resolution. And with its 68000 processor and PostScript interpreter, it was the fastest and most capable graphics machine I'd ever seen. Indeed, a 300dpi laser printer driven by PostScript was a wonder.
I ordered the Addison Wesley documentation. I figured out that sending a text file to the printer which started with "%!PS" would make the rest be processed as PostScript. Suddenly I could code vector graphics programs and send them to the LaserWriter and get beautiful prints. I was reading Douglas Hofstadter's Metamagical Themas, and my memory is that there were some great ideas for graphics in that.
The good news: I got a chance to program in a well thought-out interpreted language (a wonder, coming from the world of BASIC, Fortran, C or even Pascal). I got to produce wonderful and intricate graphics output. Reading the IEEE article makes me grateful that the Adobe designers "put functionality over speed".
The bad news: The most interesting output could still take minutes (or more) to run. Of course, this was confusing (or annoying) if someone else in the lab wanted, say, to print their term paper then run off to class. I'd like to think that I learned to run my print jobs at odd hours, and in the worst case, to power cycle the printer and lose my print, to make the device available to the room.
- It appears to be an out-of-band pre-processing stage (run the image through denoise to produce an intermediary TIFF), unlike most other parts of the program.
- All AI features are gated behind compile-time flags which default to off.
[0] https://github.com/darktable-org/darktable/pull/20523