GPT-6 places harbor objects better than Fable 5.1, Code4Scene tester says
The tester says both models used the same prompt and asset pack, but Fable 5.1 repeatedly left objects floating above the ground. GPT-6’s placement was better, though still imperfect, in their account.
TLDR
A Code4Scene tester reports comparing GPT-6 and Fable 5.1 on building a harbor with cranes, containers and vehicles. With the same prompt and asset pack, they say GPT-6 got more placement details right—though it remained imperfect—while Fable repeatedly left objects floating above the ground. The tester says the benchmark tests spatial details: containers should stack without gaps, wheels should touch the road, and trucks should have room to pass.
Combined views
4.3K
4 Sources, first seen 19d ago