# 🔥 ONE MEMBERSHIP = FULL ACCESSYour membership includes :<ul><li>All models</li><li>Exclusive LoRAs</li><li>Exclusive workflows</li><li>Scripts &amp; tools</li><li>All future updates included</li></ul><a target="_blank" rel="ugc" href="https://ko-fi.com/enzinoxl/">👉 $25/month on Ko-Fi</a>⭐ Get every current model, every future release, and all exclusive content with a single membership.<a target="_blank" rel="ugc" href="https://ko-fi.com/enzinoxl/"><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/7dad7a25-d70b-45f2-a4cf-ed3bdfccdd2c/original=true/7dad7a25-d70b-45f2-a4cf-ed3bdfccdd2c.jpeg" /></a>📢 Join The CommunityWe're building a friendly space to chat, share creations, and get support.👉 <a target="_blank" rel="ugc" href="https://discord.gg/73pMQ5bRvQ">Click here to join the Discord</a><h2 id="krea2fusion-character-fusion-workflow-by-enzino">KREA2FUSION - Character Fusion Workflow by Enzino</h2>Swap the person in any photo with your own character LoRA, no prompt writing needed.Load a source image, pick your character LoRA, type its trigger word, hit Run. The workflow does the rest: it analyzes the image, rebuilds the prompt automatically, and regenerates the picture with your character in place of the original person, keeping the scene, outfit, pose and lighting intact.<h2 id="how-it-works">How it works</h2><ol><li>Auto Caption (Qwen3-VL) - The source image is captioned locally by Qwen3-VL 4B. The instruction is carefully designed to describe the scene, clothing, pose, lighting and background but never the face, hair or identity. This is the secret sauce: since the prompt says nothing about the original person's appearance, your character LoRA has full control over the identity.</li><li>Prompt assembly - Your LoRA trigger word (typed once in the <code>LORA Prompt</code> node) is joined with the auto-generated caption and fed into the sampler. A <code>Prompt Preview</code> node shows you the exact prompt used on every run.</li><li>Pass 1 - Identity swap - Img2img pass at denoise ~0.50: high enough for the LoRA to impose the new identity, low enough to preserve the composition.</li><li>Pixel-space upscale - The result is upscaled with 4xNomos2_hq_mosr (then scaled back to 2x). Real pixel upscaling instead of latent upscaling = much sharper output.</li><li>Pass 2 — Refine - A light second sampling pass (denoise ~0.30) cleans and sharpens the upscaled image.</li><li>Face Detail — A final FaceDetailer pass polishes the character's face.</li></ol><h2 id="requirements">Requirements</h2>Models<ul><li>Krea 2 checkpoint (tested with IntoRealism Krea 2)</li><li><code>qwen3vl_4b</code> text encoder + <code>wan_2.1_vae</code> (standard Krea 2 setup)</li><li><code>4xNomos2_hq_mosr</code> upscale model</li><li>A character LoRA trained for Krea 2</li></ul>Custom nodes (all available in ComfyUI Manager — use "Install Missing Custom Nodes")<ul><li>ComfyUI-QwenVL (by 1038lab) — the Qwen3-VL 4B caption model downloads automatically on first run</li><li>rgthree-comfy</li><li>ComfyUI Impact Pack (FaceDetailer)</li></ul><h2 id="first-run-setup">First-run setup</h2>Open the QwenVL — Auto Caption node and set it once:<ul><li><code>custom_prompt</code>: already filled — this is the "describe everything except the face" instruction. Don't remove it.</li><li><code>max_tokens</code>: 512</li><li><code>keep_model_loaded</code>: off (frees VRAM before sampling — recommended on 16 GB cards)</li><li><code>quantization</code>: 4-bit on 16 GB VRAM or less</li></ul>Then type your LoRA trigger word in the <code>LORA Prompt</code> node and load your LoRA in the Power Lora Loader.<h2 id="tuning-tips">Tuning tips</h2><ul><li>Character doesn't take over enough? Raise Pass 1 denoise toward 0.55–0.60.</li><li>Scene drifts too far from the source? Lower Pass 1 denoise toward 0.35–0.40 — the FaceDetailer will still lock in the identity.</li><li>Check the Prompt Preview after each run. If it looks empty or weird, the caption node isn't configured - revisit the first-run setup above.</li></ul>Runs comfortably on 16 GB VRAM.

Files with SHA256: 51fd1e9f13b9fd8fb27d2207eaae217296ea0aaff7477ec00b7f7842afff075e