Only GPT Image could render the bow-stringing scene after Flux, Anima, and Krea2 failed
Asphyxiem · reddit · 2026-07-25
The poster tried to recreate Odysseus pulling a bowstring across Flux, Anima, and Krea2, but none of those models could reliably produce the pose. The only model that worked was GPT Image.
- The user shared a long prompt built around a cinematic tavern scene with a bow-stringing character.
- They asked others to test open-source models and report prompts/results, suggesting this may be a prompting or capability gap.
- The attached image shows the generated scene that the poster was trying to reproduce.
More from Fun
- A poster imagines AI turning the future into a nerdy reality show — sull · 2026-07-25
- A portrait of Fable 5 turns the chatbot into a keyboard-tapping companion — repligate · 2026-07-25
- A quote says healing is limited by the level of safety you actually have — relic_radiation · 2026-07-25
- Meme mocks companies that praise open models but hide even 5 lines of code — teortaxesTex · 2026-07-25
- AI Bubble Swag Turns the Bubble Debate Into Merch — lennysan · 2026-07-25
- Claude Code users are already begging never to see another task chip — inductionheads · 2026-07-25