Martin Ruskov @mapto@social.cleverlibre.org

A recent paper proposes a systematic approach for prompt refining for #text2image generators. Yet, they work only on style. No consideration of scene composition, which some generators now support. #AIart

https://creative.ai/@arxiv_hc/110748438780295466

@kierkegaank @parismarx hahah.. well, correct about windows. Again, subjective experiences, but to me ME never existed and I was happy with NT, simply because it was a poorly disguised replica of OS/2 and for unexplainable reasons I liked OS/2. Not that I was able to use it for anything since I didn't even have a copy of Lotus Notes. It feels so far away though.

@kierkegaank @parismarx let's agree to disagree as to whether Windows 95, 2003, Vista and 8 work, shall we. Then, when it comes to office, I'll limit myself to challenging its stability, in all possible meanings of the word. But again, it's fine to not agree due to our different experiences

@kierkegaank @parismarx "the man who ruined windows and office"... I love it, genius. I'd have never thought it is possible to ruin these even more

@hey @GossiTheDog average and random are very different things. I'm happy that average works for you

@hey @GossiTheDog it happened to propose something that matched your intention. How often your intentions match GitHub's average? It's not a quantitative question

"Friederike Otto, a climate scientist at Imperial College London, cautioned against thinking that we've reached a "new normal," since that implies a semblance of stability."

"We’re nowhere near a normal, we’re in a phase of accelerated warming, because we are still increasing greenhouse gas emissions. The result is the weather we see. Whenever we stop burning fossil fuels we can begin to figure out what 'normal' means again."

https://www.axios.com/2023/07/17/heat-waves-extreme-weather-globe-climate

@axios excuse me, if it is not humans - at large - to take the decisions, who else could it be? You're certainly not implying that there exists any artificially made entity capable of making design decisions whatsoever? Remember that what for marketing purposes is called Artificial "Intelligence" is - in fact - stochastic parrots. This was true for systems like DeepBlue, AlphaGo, Watson, and is true for any Large Language Model. These are systems that are incapable of conceiving anything that they haven't previously seen. In fact, their creators, quite ignorantly and not based on any research whatsoever, believe that humans have the same limitation.

@lucifargundam @garyackerman "means" as in " averages", right? As in stochastic parrots, right? ;)

@garyackerman I bet they tried asking ChatGPT "what's a template"

@johncarlosbaez that's the most I've seen so far on the topic, but the topic is developing as we speak https://www.reddit.com/r/Mastodon/comments/z7hchs/how_do_i_migrate_a_mastodon_archive_targz_to/

@johncarlosbaez @gmusser I wouldn't look for an unique selling point in mastodon in general, but in particular servers. I am certainly more willing to experiment than average users, but I am happy with my journey so far. Initially, I didn't know which server to join (luckily all mastodon.social type of names I tried were not accepting new subscriptions) so I went for one by the name: main criterion was that it's short. Turned out mine was the one instance that allows you to follow the local threads of other instances, and now I follow about a dozen specialist servers.

See this for the feature in question https://github.com/mastodon/mastodon/discussions/19819

@internic @johncarlosbaez @gmusser @pluralistic @catvalente @kdund actually moving servers is not that difficult after all:
https://screenrant.com/mastodon-change-servers-migrate-account-how/
...and the usability of this and other experiences is bound to improve with scale