Krea2PromptWeight Enhances Prompt Control
Ok_Twist_2950 · reddit · 2026-07-15
KJ nodes introduced a new Krea2PromptWeight node, serving as an alternative to the text encoder. It supports SDXL-like prompt weighting syntax, such as (word:1.2) for emphasis and (word:0.5) for de-emphasis, integrating the negpip logic for negative prompts directly into the main prompt.
In testing, the author found it significantly improves prompt control, though the default recommendation of cfg=1 isn't optimal. Setting cfg to 2 during the middle stages of generation (e.g., steps 3–6) yielded better results: applying it too early blurs the image, while applying it too late causes repetitive artifacts. When using this, you should connect an empty negative prompt box to the sampler and avoid using ConditioningZeroOut, otherwise adding cfg in later steps introduces noise.
The provided example configuration is: 10 steps, raw + turbo lora 0.75, euler / ddim uniform, with a recommendation to enable previews for fine-tuning the cfg window. Another tip is to directly use the new cfg override node in ComfyUI core to avoid chaining extra samplers.
More from coding & agent
- diginomica: agentic automation hype is overdone, and 'The Hackening' of enterprise security has begun — jonerp · 2026-10-26
- 'Vibe verification': ~60% of AI-written formal specs prove the wrong property — basedjensen · 2026-10-26
- Devin launches Security suite powered by Claude Mythos with $100M+ credits — DevinAI · 2026-10-26
- HackBench injects reward hacks into agent trajectories to stress-test monitors — DanielKhashabi · 2026-10-26
- Hugging Face ships hf CLI updates: deduped cache saves 19GB, faster buckets, retryable GPU jobs — yb2698 · 2026-10-26
- One prompt, full Blender animation: dev says he never touched Blender once — iamfakhrealam · 2026-10-26