The original STAR inference uses total_noise_levels=900, preserving input structure during SDEdit. We had 1000 which starts from near-pure noise, destroying the input. Also always append the quality prompt to user text instead of using it only as a fallback. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
23 KiB
Executable File
23 KiB
Executable File