Krea 2 General Edit V2 underwhelms: great text-to-image, inconsistent instruction-based edits
Far_Estimate7276 · reddit · 2026-11-03
A ComfyUI user reports that Krea 2 delivers excellent text-to-image results but inconsistent quality when editing existing images with the new General Edit V2 LoRA.
They use the ComfyUI-Krea2Edit custom nodes, but the published example workflow predates the V2 LoRAs, and they've filed an issue asking whether it's still recommended. Open questions include whether Turbo or Raw edits better, optimal LoRA strength/steps/CFG/reference boost settings, whether lowering reference boost helps when the model refuses changes, and which edit types Krea 2 handles well — focusing on ordinary instruction-based edits: reskinning, object removal/replacement, background swaps, relighting, and local edits that preserve the rest of the image.
More from Multimodal
- ComfyUI Workflow Generates 1080p Video with MiniMax H3 on Only 6GB VRAM — cgpixel23 · 2026-11-03
- Midjourney Recipe: 5 Weighted --sref Styles for Cinematic Light and Motion — michaelrabone · 2026-11-03
- LTX 2.5 users struggle to lock static cameras even with the Cinemagraph LoRA — bytebitter33 · 2026-11-03
- Tiny latent diffusion model trained inside LLMs generates MNIST digits, 76% recognized — zeeshanp_ · 2026-11-03
- Realism LoKr for Krea 2 trained on 749 images, blind voting arena picks the best epoch — Wide_Director_8897 · 2026-11-03
- Stunning AI-generated dieselpunk elf warrior image showcases detailed mechanical design — azed_ai · 2026-11-03