Krea2 generates identical images even with new seeds: ComfyUI user stuck
orangeflyingmonkey_ · reddit · 2026-08-21
A ComfyUI user generating with Krea2 Raw + Turbo LoRA hits a wall: even after changing the seed, the output image stays identical. Their pipeline uses a Generate Text node to caption the input image, then generates from that description. Two common fixes failed—connecting a Seed Variance Enhancer node to the positive conditioning, and a Krea2T Enhancer Advanced node to the model output. The question: how to produce genuinely different variations from the same prompt in Krea2.
More from Multimodal
- Elf-Xuan 2.0 Robot Performs Live with Real-time Lip Sync and Gestures — rohanpaul_ai · 2026-08-21
- Unbound Loop Launches Creator Residency with Free Real-time Diffusion and 3D Generation — rms80 · 2026-08-21
- NovelAI releases Diffusion V5 image generation model — bdsqlsz · 2026-08-21
- The Marshmallow AI Benchmark: Testing vision models with counting — qu1etus · 2026-08-21
- Runway Teases 'Most Consequential' Releases in Company History — Uncanny_Harry · 2026-08-21
- Causal-rCM: Open-Source Recipe for Streaming Video Generation — chenhsuanlin · 2026-08-21