Krea 2 Identity Edit LoRA v1.2

About

Krea 2 Identity Edit LoRA v1.2 is an instruction-based LoRA for identity-preserving image editing on Krea 2. Upload a reference image, write a plain-language edit, and Sogni runs the workflow on the Supernet while preserving faces, outfits, product details, and scene elements you did not ask to change.

The v1.2 release is the recommended Krea 2 Identity Edit LoRA build on Hugging Face and Civitai. It improves face likeness on restaged subjects, adds a 1024 high-resolution pass, and supports reference-sheet workflows, head, face, eye, and person swaps, inpainting, outpainting, and virtual try-on.

Sogni exposes both bases: standard Krea 2 Identity Edit v1.2 for cleaner, more consistent edits, and Dark Beast Krea 2 Identity Edit v1.2 for creators who want the Dark Beast look in an identity-edit workflow.

Variants

Krea 2 Identity Edit LoRA v1.2 ships in 2 builds — same family, different speed/quality trade-offs. Pick the variant that fits, or start with the default.

Prompting tips

Krea 2 Identity Edit v1.2 works best when the instruction is direct and the reference roles are unambiguous.

• Start with the reference — This is an image editor, not a blank text-to-image model. Add one source image before describing the change.
• Use action verbs — Ask for a concrete edit: restage the person, replace the object, change the outfit, remove an element, extend the frame, or apply a visual style.
• Order two-image edits — For scene-plus-person workflows, put the scene image first and the person or identity reference second.
• Tune preservation — Grounding controls how much visual structure comes from the reference. Reference Boost around 4 is the strong-likeness default; lower it for looser edits.
• Choose the base — Use Krea 2 for cleaner identity edits. Switch to Dark Beast Krea 2 when you want a more stylized, less predictable Dark Beast result.

The app exposes Grounding, Reference Boost, and Reference B Boost only for the two Krea 2 Identity Edit v1.2 model IDs.

What v1.2 adds

v1.2 is the version users are looking for: it is the recommended Hugging Face LoRA build (krea2_identity_edit_v1_2.safetensors) and the version Sogni has deployed. Compared with earlier builds, the author documents stronger likeness in restaged people, a 1024 high-resolution pass, FIT reference geometry for mismatched source/output aspect ratios, and the ref_boost fidelity dial exposed in Sogni as Reference Boost.

The LoRA is an unofficial community fine-tune of Krea 2, not an official Krea product. Source pages: conradlocke/krea2-identity-edit on Hugging Face and Krea 2 Identity Edit on Civitai.

Pricing

Use pay-as-you-go Spark packs for each render (1 Spark = $0.005), or choose a flat-rate Sogni plan for credit-free fair-use generation in the app.

API

One Sogni API key reaches every model on the Supernet — call Krea 2 Identity Edit LoRA v1.2 with the exact model id for the variant you want.

import { SogniClient } from '@sogni-ai/sogni-client';

const client = await SogniClient.createInstance({
  appId: crypto.randomUUID(),
  apiKey: process.env.SOGNI_API_KEY,
  network: 'fast',
});

const project = await client.projects.create({
  type: 'image',
  modelId: 'krea2_identity_edit_v1_2',
  positivePrompt: 'a slothicorn surfing a wave of liquid paint, studio lighting',
  numberOfMedia: 1,
});

const [url] = await project.waitForCompletion();
console.log(url); // result link — download within 24h

# krea2_identity_edit_v1_2 is addressable by id via the JavaScript SDK (see the JavaScript tab).
# The REST workflow API routes this family through its flagship alias instead:

curl https://api.sogni.ai/v1/creative-agent/workflows \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $SOGNI_API_KEY" \
  -d '{
    "input": {
      "title": "First render",
      "steps": [{
        "id": "step1",
        "toolName": "generate_image",
        "arguments": { "prompt": "a slothicorn surfing a wave of liquid paint, studio lighting" }
      }]
    },
    "confirm_cost": true
  }'

Example uses krea2_identity_edit_v1_2. Use any exact model id from the Variants table above. Full reference at docs.sogni.ai.