RunningHub workflow page: https://www.runninghub.ai/zh-cn/post/2079889918284877826?inviteCode=rh-v1111
Krea 2 Identity Edit v1.2 - Single-Image Mask Editor
Purpose: the masked variant of the Krea 2 Identity Edit v1.2 graph. You supply one image plus a mask, and the instruction is applied only where the mask allows, which is how you edit one object (a product, a piece of clothing, a detail) without repainting the rest of the frame.
Graph (as saved)
22 nodes / 27 links. Groups: Mask.
- UNETLoader:
krea2_turbo_fp8_scaled.safetensors; LoRA:krea2_identity_edit_v1_2_fp16.safetensorsat strength 1. - CLIPLoader:
qwen3vl_4b_fp8_scaled.safetensors, typekrea2; VAE:qwen_image_HDR_vae_fp32_comfy.safetensors. - KSampler: 10 steps, CFG 1,
euler,simple, denoise 1. - Krea2EditModelPatch: ref_boost 3, single-image mode, fit.
- GrowMask: expand 6, taper on - the saved mask is grown before it gates the edit.
- Krea2EditGroundedEncode: the example instruction adds thin black cat-eye sunglasses and explicitly asks to preserve face, hairstyle, flower accessory, makeup, clothing, background and lighting (encode size 768).
- SaveImage prefix
ComfyUI/Kera2Edit.
How to use
Load the image and its mask, then write the change with an active verb ("add", "replace", "change", "make"). The grow setting is worth tuning per subject: too tight and the edit shows a seam, too wide and unrelated areas start changing. The example instruction is a good template for identity-preserving object edits because it names what must not change.
Limits and not-verified items
No inference was run for this listing: no speed, VRAM, mask-alignment or quality claim is made. The saved mask, image and instruction are the author's demo content, not evidence of an output. Mask behaviour depends on the exact node implementation and image; verify it on your own material. Weights, LoRA and input media are not bundled; check their licenses separately. The cover image is a promotional still, not a frame produced by this graph.
Tutorial: YouTube · Bilibili. Workflows are supplied through AI-KSK; verify workflow, node, model, LoRA, source-image and cover rights separately. No blanket commercial or redistribution license is asserted.
Description
RunningHub workflow page: https://www.runninghub.ai/zh-cn/post/2079889918284877826?inviteCode=rh-v1111
Krea 2 Identity Edit v1.2 - Single-Image Mask EditorPurpose: the masked variant of the Krea 2 Identity Edit v1.2 graph. You supply one image plus a mask, and the instruction is applied only where the mask allows, which is how you edit one object (a product, a piece of clothing, a detail) without repainting the rest of the frame.
Graph (as saved)22 nodes / 27 links. Groups: Mask.
- UNETLoader:
krea2_turbo_fp8_scaled.safetensors; LoRA:krea2_identity_edit_v1_2_fp16.safetensorsat strength 1. - CLIPLoader:
qwen3vl_4b_fp8_scaled.safetensors, typekrea2; VAE:qwen_image_HDR_vae_fp32_comfy.safetensors. - KSampler: 10 steps, CFG 1,
euler,simple, denoise 1. - Krea2EditModelPatch: ref_boost 3, single-image mode, fit.
- GrowMask: expand 6, taper on - the saved mask is grown before it gates the edit.
- Krea2EditGroundedEncode: the example instruction adds thin black cat-eye sunglasses and explicitly asks to preserve face, hairstyle, flower accessory, makeup, clothing, background and lighting (encode size 768).
- SaveImage prefix
ComfyUI/Kera2Edit.
Load the image and its mask, then write the change with an active verb ("add", "replace", "change", "make"). The grow setting is worth tuning per subject: too tight and the edit shows a seam, too wide and unrelated areas start changing. The example instruction is a good template for identity-preserving object edits because it names what must not change.
Limits and not-verified itemsNo inference was run for this listing: no speed, VRAM, mask-alignment or quality claim is made. The saved mask, image and instruction are the author's demo content, not evidence of an output. Mask behaviour depends on the exact node implementation and image; verify it on your own material. Weights, LoRA and input media are not bundled; check their licenses separately. The cover image is a promotional still, not a frame produced by this graph.
Tutorial: YouTube · Bilibili. Workflows are supplied through AI-KSK; verify workflow, node, model, LoRA, source-image and cover rights separately. No blanket commercial or redistribution license is asserted.