<h1 id="krea-2-identity-edit">Krea 2 Identity Edit</h1> #v1.2 — new thingsIn the model:<ul><li>Better likeness</li><li>Character sheets — use and creation</li><li>Head/face swap (+ eye/person replace) — thanks to stablellama</li><li>Outpainting</li><li>Inpainting</li><li>Try-on</li><li>Better person removal</li><li>Higher fidelity + a real high-res pass</li></ul>In the nodes:<ul><li><code>ref_boost</code> — reference-fidelity dial ← don't miss this one, try between 2 and 6</li><li>Fixed blurry images / <code>fit</code> geometry (also removes the old "match the aspect ratio" requirement)</li></ul>**Settings:** Turbo, 8–12 steps (8 = composition, 12 = face detail, ~10 balanced), CFG 1.0, LoRA strength 1.0. Removals/deletions: Two-ref edits: **scene = image 1, person = image 2.**Requires the [ComfyUI-Krea2Edit nodes](<a target="_blank" rel="ugc" href="https://github.com/lbouaraba/comfyui-krea2edit">https://github.com/lbouaraba/comfyui-krea2edit</a>) — see [CHANGELOG](<a target="_blank" rel="ugc" href="https://github.com/lbouaraba/comfyui-krea2edit/blob/main/CHANGELOG.md">https://github.com/lbouaraba/comfyui-krea2edit/blob/main/CHANGELOG.md</a>). Instruction-based, identity-preserving image editing for Krea 2 (12.9B single-stream MMDiT). Give it an image and a plain-language instruction; it edits while preserving what you didn't ask to change — including the person.An unofficial community fine-tune of <a target="_blank" rel="ugc" href="https://huggingface.co/krea/Krea-2-Raw">Krea 2 Raw</a>. Not an official Krea product; not affiliated with or endorsed by <a target="_blank" rel="ugc" href="http://Krea.ai">Krea.ai</a>, Inc.Requires the <a target="_blank" rel="ugc" href="https://github.com/lbouaraba/comfyui-krea2edit">ComfyUI-Krea2Edit node pack</a> — the LoRA is trained with dual conditioning (in-context VAE tokens + image-grounded Qwen3-VL encoding) that stock nodes don't provide. Two ready-made workflows ship with it.<h2 id="what-it-does">What it does</h2><ul><li>Person re-staging with likeness: "create a photo of this person at a night market" — same face, same outfit down to individual moles and marks, fully relit to the new scene. New camera angles and poses included.</li><li>Local edits: recolor, add/remove/replace objects, attribute and outfit changes, with near-pixel preservation of the rest of the frame.</li><li>Replace-with-reference: "replace the woman with a big orangutan" — the replace verb is trained, locality holds.</li><li>Full-image restyles: global style with preserved composition.</li><li>Two-input edits (experimental): scene + person as separate references. Outfits and placement work well; see limitations for faces.</li><li>Composes with your LoRAs: character/body/style LoRAs stack on top and steer the prior — something closed editors structurally can't offer.</li></ul><h2 id="recommended-settings">Recommended settings</h2>Task type Model Steps CFG Most edits (add, recolor, restyle, re-stage) Turbo 8 1.0 Removals / large deletions Raw 20 3.0<ul><li>Match the output aspect ratio to the source image. Training pairs are same-size; AR mismatch degrades preservation (edits may apply to only part of the frame).</li><li>Generate at ≤2MP. Above that, source content can bleed or subjects duplicate (training was 768/1024-class).</li><li><code>grounding_px</code> is a real dial (trained range 512–1536): lower values = stronger edit adherence and more uniform scene changes; higher values = stronger identity/likeness. 768 is a balanced default; try 1024+ for people.</li><li>At CFG &gt; 1, ground the negative too (empty prompt + same image).</li><li>LoRA strength 1.0.</li></ul><h2 id="known-limitations-(honest-list)">Known limitations (honest list)</h2><ul><li>Likeness is texture-faithful, proportion-conservative. Moles, skin character, hair, and lighting adapt beautifully; strongly distinctive facial geometry (unusual nose, eye spacing, face length) regresses toward typical proportions. People whose identity lives in texture and structure transfer best; geometry-defined faces read as a "close relative."</li><li>Two-person inputs keep outfits distinct but faces drift toward each other. Workaround that works today: chain single-ref inserts (place person A, then a second edit pass adding person B from their reference).</li><li>Removal works but is not yet reliable — always use the Raw/CFG 3 recipe; expect occasional re-renders instead of deletions.</li><li>Outfit swaps are hit-or-miss — changing what a person wears sometimes works cleanly and sometimes doesn't apply; reroll or rephrase.</li><li>Local edits aren't always perfectly local — add/remove/replace operations can sometimes alter other parts of the frame or shift the overall color grade. If preservation matters, compare against the source and reroll.</li><li>Highly unusual visual content (extravagant hairstyles, extreme body types) can drift toward the base prior — a subject LoRA stacked on top fixes this.</li></ul><h2 id="license">License</h2>The LoRA weights are a Derivative Model of Krea 2 and are distributed under the <a target="_blank" rel="ugc" href="https://huggingface.co/conradlocke/krea2-identity-edit/blob/main/LICENSE.pdf">Krea 2 Community License Agreement</a> (see also <code>NOTICE</code>). Key points for users: commercial use is permitted under the license's revenue threshold (§2.3, currently &lt;$1M/yr — above that, contact Krea for an enterprise license); deployments must implement reasonable content moderation (§4.2); AI disclosure obligations apply where required (§4.3). This repository modifies the Krea Model as permitted by §3; it is not endorsed by Krea.Research/portfolio release by a self-funded hobbyist.<h2 id="showcase">Showcase</h2>All reference people below are themselves AI-generated — no real likenesses. Prompts are embedded in each image.

Krea 2 Identity Edit

Krea 2 Identity Editv1.1 (recommended): <code>krea2_identity_edit_v1_1.safetensors</code> — substantially improved face likeness and image fidelity, much stronger edit locality (camera, pose, and untouched elements stay fixed far more reliably), better two-person identity separation, more reliable object remove/replace, better compound outfit-change compliance, corrected reference geometry handling. One honest regression: person-replacement ("replace the woman with an orangutan") is currently weaker than v1 — keep v1 for that use case until v1.2. No high-resolution adaptation pass yet: at high resolutions (especially two-person edits) identities can bleed together — prefer ~1–1.5MP and upscale. v1 remains available for workflow reproducibility.Instruction-based, identity-preserving image editing for Krea 2 (12.9B single-stream MMDiT). Give it an image and a plain-language instruction; it edits while preserving what you didn't ask to change — including the person.An unofficial community fine-tune of <a target="_blank" rel="ugc" href="https://huggingface.co/krea/Krea-2-Raw">Krea 2 Raw</a>. Not an official Krea product; not affiliated with or endorsed by <a target="_blank" rel="ugc" href="http://Krea.ai">Krea.ai</a>, Inc.Requires the <a target="_blank" rel="ugc" href="https://github.com/lbouaraba/comfyui-krea2edit">ComfyUI-Krea2Edit node pack</a> — the LoRA is trained with dual conditioning (in-context VAE tokens + image-grounded Qwen3-VL encoding) that stock nodes don't provide. Two ready-made workflows ship with it. What it does<ul><li>Person re-staging with likeness: "create a photo of this person at a night market" — same face, same outfit down to individual moles and marks, fully relit to the new scene. New camera angles and poses included.</li><li>Local edits: recolor, add/remove/replace objects, attribute and outfit changes, with near-pixel preservation of the rest of the frame.</li><li>Replace-with-reference: "replace the woman with a big orangutan" — the replace verb is trained, locality holds.</li><li>Full-image restyles: global style with preserved composition.</li><li>Two-input edits: scene + person as separate references — "create a photo of this man next to the tractor." Input order matters and is fixed: the scene is always image 1 (<code>source_latent</code>/<code>image</code>), the person is always image 2 (<code>source_latent_b</code>/<code>image_b</code>). Swapping them sharply degrades results (this matches the training layout).</li><li>Composes with your LoRAs: character/body/style LoRAs stack on top and steer the prior — something closed editors structurally can't offer.</li></ul> Recommended settingsTask type Model Steps CFG Most edits (add, recolor, restyle, re-stage) Turbo 8–12 1.0 Removals / large deletions Raw 20 3.0<ul><li>Match the output aspect ratio to the source image. Training pairs are same-size; AR mismatch degrades preservation (edits may apply to only part of the frame).</li><li>Generate at ≤2MP (source bleed / duplication above). For v1.1 two-person edits, prefer ~1–1.5MP — at higher resolutions the two identities may blend together; generate lower and upscale instead.</li><li>Step count is a mild dial too: fewer steps (8) favor composition adherence, more (12) favor face detail; ~10 is a good balance.</li><li><code>grounding_px</code> is a real dial. Lower values = stronger edit adherence and more uniform scene changes; higher = stronger identity/likeness. v1.1's trained range is 384–768 (768 default); 1024 often still works nicely. If you get duplicated/split compositions ("double pictures"), lower <code>grounding_px</code> — running far above the trained range is the most common cause. (v1's trained range was 512–1536.)</li><li>At CFG &gt; 1, ground the negative too (empty prompt + same image).</li><li>LoRA strength 1.0.</li></ul> Known limitations (honest list)<ul><li>Likeness is texture-faithful, proportion-conservative. Moles, skin character, hair, and lighting adapt beautifully; strongly distinctive facial geometry (unusual nose, eye spacing, face length) regresses toward typical proportions. People whose identity lives in texture and structure transfer best; geometry-defined faces read as a "close relative."</li><li>Two-person inputs keep outfits distinct but faces drift toward each other. Workaround that works today: chain single-ref inserts (place person A, then a second edit pass adding person B from their reference).</li><li>Removal works but is not yet reliable — always use the Raw/CFG 3 recipe; expect occasional re-renders instead of deletions.</li><li>Outfit swaps are hit-or-miss — changing what a person wears sometimes works cleanly and sometimes doesn't apply; reroll or rephrase.</li><li>Local edits aren't always perfectly local — add/remove/replace operations can sometimes alter other parts of the frame or shift the overall color grade (substantially improved in v1.1). If preservation matters, compare against the source and reroll.</li><li>Highly unusual visual content (extravagant hairstyles, extreme body types) can drift toward the base prior — a subject LoRA stacked on top fixes this.</li></ul> LicenseThe LoRA weights are a Derivative Model of Krea 2 and are distributed under the <a target="_blank" rel="ugc" href="https://huggingface.co/conradlocke/krea2-identity-edit/blob/main/LICENSE.pdf">Krea 2 Community License Agreement</a> (see also <code>NOTICE</code>). Key points for users: commercial use is permitted under the license's revenue threshold (§2.3, currently &lt;$1M/yr — above that, contact Krea for an enterprise license); deployments must implement reasonable content moderation (§4.2); AI disclosure obligations apply where required (§4.3). This repository modifies the Krea Model as permitted by §3; it is not endorsed by Krea.Research/portfolio release by a self-funded hobbyist.

Files with SHA256: 78b403a45024d51c24b618046ec1176c346416951f8f4c6c707d1b337ae1d40f