Hmm, so federated downvotes from Lemmy are public now on mbin, not just local downvotes and federated upvotes. Interesting. Does mbin-mbin downvote federation work? kbin doesn't federate downvotes to you. (I checked -- for science! -- but switched back to an upvote afterwards.)
artificial gestation
The word "matrix" literally means "womb" in its older sense.
Sorry if addressed in the link (I'm not willing to visit Twitter) -- but, like, actually McDonald's themed? Or are they just sponsoring a show (like P&G, etc. have done for ages)?
If the former, I guess there's some precedence with the KFC visual novel and Isekai Izakaya and such, but that still sounds pretty weird...
Edit: I went back and checked and it looks like McDonald's was also a sponsor on the show I remember P&G from (i.e. season 1 of Bleach), so there's precedence for them sponsoring Studio Pierrot's shows too -- I just don't usually pay that much attention to it, I guess.
Huh. So... it finally occurred to me to check this. Elfen Lied and Sora no Woto share the same director -- Mamoru Kanbe / 神戸守. I probably should've figured that out sooner given that Sora no Woto's OP also references Klimt's paintings heavily, but I had no idea.
It's not a GUI library, but Jupyter was pretty much made for the kind of mathematical/scientific exploratory programming you're interested in doing. It's not the right tool for making finished products, but is intended for creating lab notebooks that contain executable code snippets, formatted text, and visual output together. Given your background experience and the libraries you like, it seems like it'd be right up your alley.
I mentioned in a past comment a while back that I made a catalog of my anime. One of the observations I found while making it is that everything except for one movie had an entry on the English language Wikipedia already. That movie is Gundress from 1999. According to my personal journal, I watched this once back in 2014, apparently, but I remembered nothing about it, so I loaded it up recently and rewatched it.
The movie has that "sort of hard to follow if you don't already know the source material" kind of feel -- although I think this is the original work? I checked the Japanese Wikipedia entry about it after watching it. Sticking the article through a translator, there's a description of a seriously screwed up initial showing and mismanagement of production with the film being finished after it aired in theaters initially. The version I have is finished, of course; if half the movie wasn't colored in I'd definitely have remembered that!
The DVD menu prominently credits it as "Masamune Shirow's Gundress", but I'm not sure what his role in the production actually was. He's listed in the opening credits for 設定協力 which got translated to English as "Characters Designed by" -- but different people are credited with character and mech design in the end credits. A literal translation is something like "setting cooperation".
There's definitely a number of familiar elements with some buildings reminiscent of Dominion Tank Police, mech suits that reminded me of designs in GitS:SAC, as well as thermoptic camouflage, cable-based cyborg communication (jacked into the neck), cyberdiving, etc. coming up during the story.
Unusually, this anime features a Little Arabia enclave within the Japanese "Bayside City" the story is set in and one of the main characters is Muslim. I think this may be the only time I've seen Arabic script in anime -- although I don't know what it says.
I clipped some screenshots and stacked them up so you can see what it looks like, if you're curious: https://files.catbox.moe/qtsa0d.png (~8MB)
Gundress (1999)
Yep. It's Garden of Words. I just skimmed through my copy and this image is from about 18 minutes in.
It might be easier to just fire up Wireshark and look for relevant traffic when you trigger the action.
I was just thinking about the image resizing thing again when I saw your message notice pop up. Another option for preview is a web browser. A minimal HTML page with some JS to refresh the image would avoid the image resize on reload problem, and gives you some other interesting capabilities. Python ships with a kind of meh (slow and quirky), but probably sufficient HTTP server (python3 -m http.server) if you'd prefer to load the preview on a different computer on your LAN entirely (e.g. cellphone / tablet / ... ) for example.
A simple HTML file for this would be something like:
Regarding input from a gamepad -- I've had some similar ideas before but haven't really had much success using a gamepad artistically outside some limited things where I either wrote the entire program or was able to feed data into programs that accepted input over the network (e.g. via HTTP and which I wrote a custom adapter for). It's been a long time since I've tried anything in that space though, and it might be possible to do something interesting by trying to make the system see the combination of a gamepad stick as relative mouse motion and trigger as pen pressure. I'm not quite sure how to go about doing that, but I'll let you know if I find a way to do it.
The Wikipedia article for hqx points out that an implementation exists as a filter in ffmepg.
You can run a command line conversion of e.g. a PNG -> PNG using hqx upscaling like: ffmpeg -i input.png -filter_complex hqx=4 output.png
The =4 is for 4x upscaling. The implementation in my version of ffmpeg supports 2x, 3x, and 4x upscaling.
As a quick and dirty way to get semi-live preview, you can do the conversion with make and use watch make to try to rebuild the conversion periodically. (You can use the -n flag to increase the retry rate if the default is too long to wait.) make will exit quickly if the file hasn't changed. Save the image in your editor and keep an image viewer that supports auto-reload on change open to see "live" preview of the output. (e.g. eog can do it, although it won't preserve size of the image -- at least not in the copy I have, anyway; mine's a bit old though.)
Sample Makefile:
output.png : input.png Makefile
ffmpeg -y -i input.png -filter_complex hqx=4 output.png
Note the -y option to tell ffmpeg to overwrite the file; otherwise it will stop to ask you if you want to overwrite the file every time you save, and in case you're not familiar with Makefiles, you need a real tab (not spaces) on the line with the command to run.
ffmpeg also appears to support xbr (with =n option as well) and super2xsai if you want to experiment with those too.
I'm not sure if this will actually do what you want artistically, but the existing implementations in ffmpeg makes it easy to experiment with.
Is there a way (on kbin.social) to edit the body text of a thread that links to an image? I can see how to change the title, tags, badges, and NSFW and OC flags, plus remove the associated preview but not how to edit the body text. Am I just missing it somewhere, or is it really not there? (I'd like to fix some typos I noticed in a recent post I made...)
Mew is a character who shows up in the episode "Dream of an Oasis" from the 2003 anime Wolf's Rain. During her introduction, there's a panning shot showing her animal form in her reflection. I thought it'd be interesting to try to reassemble those frames and combine it with the cute pose she makes later in in the scene to form a composite. It ended up being a rather of a lot of work -- involving not just using an image editor but writing a custom program to solve one of the trickier alignments by manually selecting corresponding points and numerically estimating the transformation -- but I had fun with it. This image is the final result.
Have you tried Resonance? It's a mystery adventure game set in modern times where you play as four different characters whose stories interconnect. It's been a while since I played it (a decade or so?) but I remember that it had an interesting game mechanic that let you use memories like items in various interactions, as well as a number of puzzles that I rather liked the design of.