Two small new prints from this afternoon, this time not Kallitype but contact printing on silver chloride FB paper, developed with Adox MCC. The increase in depth and detail is just wonderful and the feel of the final results an instant throwback to my early darkroom adventures ~40 years ago... 😍
The pictures are of two of my favorite places in recent years. Larger prints will be available for purchase in September.
ceux qui se servent d' #IA au travail en dressent un profil nuancé. 75 % en tirent des gains de productivité, Š titre individuel, mais seuls 13 % estiment que leur entreprise en profite également. Le temps passé au contact des modèles est par ailleurs consacré aux deux tiers Š baby-sitter l’IA, contre 36 % Š s’en servir réellement.
I can image a developer parallel to the first, too: the human still using all their skills and experience, but with the machine catching mistakes, providing context and validation and vigilance that is •orthogonal to• testing and type checking and code crafting and — the big one! — actually •thinking• about the problem.
That’s a regime I imagine developers would feel a lot better about. And I know there are people out there pursuing it! But they’re not the ones dominating the conversation.
4/
📊 Results: text context produced 50 successful games at ~9 minutes average. Image context produced only 23 successful games at ~20 minutes average. Damage numbers were comparable or slightly better with images, but the runtime gap was drastic.
💸 Token usage: 306 million tokens for the text runs vs 213 million for the image runs — with roughly half as many completed games.
Apple announces new parental controls, letting kids ask to browse specific websites or contact specific people, adding automatic image filtering, and more (Daniel Cooper/Engadget)
https://www.engadget.com/2189728/apple-expands-parental-c…
Meta launches Muse Image in Meta AI, Instagram, and WhatsApp, and previews Muse Video, the first media generation models from its Superintelligence Labs (Meta)
https://ai.meta.com/blog/introducing-muse-image-muse-video-msl/
Partys, Sperren, Demo - ESC geht ins Finale: Ganz Wien im Ausnahmezustand #News #Nachrichten
I also saw that uncredited, context-free photo in my timeline.
This entire thread is a good explanation of the many ways that post was bad form:
https://mas.to/@meganL/116888106293249758
Give credit, give context, don’t do inspo porn.
Klingt nice. Aber fast die Hälfte der Commits stammt von #Claude.
#vibecoding
https://bsd.network/@lcheylus/11656610
🧠 The reasoning: text tokens are discrete, drawn from a fixed vocabulary of ~50,000 entries, each mapping to a fixed point in embedding space.
Image tokens are continuous vectors of a thousand floating point numbers, so they can pack far more information into the same space.
🎮 The test: a tower-defense game where the AI plays in JSON mode without graphics, run for hours with two context variants — the same game state as a plain text string, and as an image rendered via PixelPipe.…