A simple test instruction that every image gen model fails to follow, but will happily confirm it did when challeneged, is to generate an image of specific dimensions.
"generate a 128 pixel wide by 128 pixel high image of..."
Maybe its unfair for me to expect this due to how llms work, but then again maybe its unfair to call an llm "intelligent".
Ike
npub1vt…hg33x
2025-11-20 16:54:54 UTC
in reply to nevent1q…styv
Author Public Key
npub1vtfjqm3qq0cc6mt28vgsw5anlhpmvlarsw0kxzy9p0h33uktkd0sghg33xSeen on
wss://relay.momostr.pinkPublished at
2025-11-20 16:54:54 UTCEvent JSON
{
"id": "a2e714cd358138a4ab1c2e5793a218ae1fc63c71c2229c5d197383d113c29256",
"pubkey": "62d3206e2003f18d6d6a3b110753b3fdc3b67fa3839f6308850bef18f2cbb35f",
"created_at": 1763657694,
"kind": 1,
"tags": [
[
"e",
"428a0516d2e64064967d75348b4d27d339d4f06b740a06a853d2831c02af671b",
"",
"root",
"4315a187e024818492e61938093ba683dae66624d202cd43738de5b8ba198c0f"
],
[
"p",
"4315a187e024818492e61938093ba683dae66624d202cd43738de5b8ba198c0f"
],
[
"proxy",
"https://pkm.social/@ike/115583070695387035",
"web"
],
[
"proxy",
"https://pkm.social/users/ike/statuses/115583070695387035",
"activitypub"
],
[
"L",
"pink.momostr"
],
[
"l",
"pink.momostr.activitypub:https://pkm.social/users/ike/statuses/115583070695387035",
"pink.momostr"
],
[
"-"
]
],
"content": "A simple test instruction that every image gen model fails to follow, but will happily confirm it did when challeneged, is to generate an image of specific dimensions.\n\n\"generate a 128 pixel wide by 128 pixel high image of...\"\n\nMaybe its unfair for me to expect this due to how llms work, but then again maybe its unfair to call an llm \"intelligent\".",
"sig": "472238ade8cb695dcd096a99348134f1a5dadae0d3c954460f783a4f79b3cc4257677839cd366df032dade20a007bb76c0d913f006bf5e564facad833f45cc17"
}