Charlie Loyd @vruba@everything.happens.horse

There’s a lot in the repo, but I’m happy to say the core code is small and efficient. You can pansharpen several megapixels/second on an ordinary CPU, and somewhere > 10 Mpx/s on a gaming rig.

It's all licensed CC BY-NC, like its training data: Maxar/Vantor’s Open Data Program (h/t @marcpfister). This is imagery for disaster response, and a goal of Potato is to publicize that data, and similar data, and to encourage work that makes it easier to use.

It’s got a lot of moving parts and I wanted to make it intelligible to several audiences, so the documentation is a bit much at times.

I’m not expecting most people to know what pansharpening is, or to have a pressing need to do it themselves. But it’s an interesting problem, and I hope it’s interesting to learn about. It’s taken a lot of inspiration from across domains (by movie colorists, for example).

I’ve been working on this occasionally for years. It’s been a weekend morning here, a notebook page on the bus there, week-long sprints between contracts, etc. Eager to learn what hugely embarrassing bugs I’ve left in it.

I’ve just published Potato, a new pansharpening package. It aims to render certain kinds of satellite imagery more clearly and accurately than what’s for sale and on satellite maps today: https://github.com/celoyd/potato/

It’s under 50k parameters, comes with a working checkpoint, runs on a home computer, and is specialized on WorldView-2/3 ARD imagery. It does a few things I haven’t noticed before in the literature: for example, using all the visible multispectral bands to make its visible colors.

@maxfenton I am rich in a type of digital asset known as FRIENDSHIP, Max. 🥰

It’s called the Getty Museum open data because it lets you get the wait no because you can freely be getting the no hang on okay because you can museum the gotten no I messed it up again but seriously folks you can “got” the artworks aw shit
https://mastodon.social/@creativecommons/115532158945263774

Technically this is a type of phase diagram.
https://pixelfed.social/p/Atanas/895754272486310141

Everyone’s going to be grateful for my explanation of satellite imagery data pathways if I make it look like a category theory diagram without actually using the notation correctly!

Decades from now, some relative of mine is going to be going over the old notebooks I willed to them, flip one open to the page where I doodled “Turgeneverending Story Musgrave of the Fireflies” and put them all in the nonrecyclable trash.

@tmcw Or more personally, things I use (and recommend) today that I didn’t 3 years ago: Obsidian, uv, tailscale, LSP, duckdb.

@tmcw C losing its social license for new low-level projects is probably by far the best and most consequential. Oceans of pain and waste saved, even at the cost of flamewars.

All ML applications aside, decent HLL GPU tooling is incredible. Doing with a couple lines of crisp python what used to take a page of CUDA is a big, big deal in all kinds of simulation, graphics, etc. (Admittedly you could date this as far back as maybe 7 years ago, so.)

Mastodon and Bluesky, warts and all.

Eager to try this. Good example of an open software project building consensus, doing the funding work, and properly fixing a UX trouble spot.
https://mastodon.social/@gdal/114478844610860445

Worryingly close to setting up a single-serving site that tells people to put spaces between numbers and their units. I’m surrounded by things like “a 5m gap” and “processing 15items/s” and I can’t take it much longer. Other than that I’m doing great.

Writing some tests and unfortunately they have turned up a bunch of bugs.