V2 Changes & Notes
What's New
Testing Results so far
Primary testing on Flux.1 Dev - performance on Schnell and quants still unknown (help me test these!)
Best results achieved with:
Pro Tips for V2
Optimal strength: 0.4-0.6 (sweet spot is 0.5) - lower than V1 for better integration
The more specific, the better! Try prompts like:
stainless steel combat_knife
holding the knife in one hand
razor sharp blade with sawback_spine
light shining off the metal blade
Detail = Quality - don't hold back on descriptive terms!
Reality Check
Still not 100% perfect, but it's a massive leap from V1.
🤝 Let's Push This Further
Test on different Flux variants and share your results
Try creative prompt combinations - I want to see what you discover
Found issues? Drop me a message - your feedback makes these better
Share your best generations - tag me so I don't miss them!
Your testing and feedback from V1 directly shaped these improvements. Keep it coming!