This workflow is designed for Flux2-Klein9B seamless head replacement and high-resolution upscale finishing. Its main purpose is to help creators replace the head from a base image with the head from a reference image while preserving the original background, camera perspective, body posture, lighting, exposure, contrast, and overall photographic structure. It is built for controlled and authorized portrait editing, not random face mixing or uncontrolled regeneration.
The workflow uses Flux2-Klein9B as the main generation backbone, with qwen_3_8b_fp8mixed as the Flux2 text encoder and flux2-vae as the VAE. It also includes a specialized head-swap LoRA route, using a Flux-Klein head LoRA to strengthen the model’s ability to perform realistic head replacement. The prompt is written specifically for seamless head swap: Image 1 is treated as the base image, while Image 2 provides the replacement head identity. The instruction asks the model to preserve the environment, background, camera distance, body proportions, neck alignment, lighting direction, shadow contact, and sharpness consistency.
The core strength of this workflow is identity transfer with proportion control. Many head-swap workflows can place a new face into an image, but they often fail at scale, neck blending, jawline connection, skin tone matching, or lighting continuity. This graph is designed to reduce those problems by combining ReferenceLatent conditioning, VAE encoding, FluxGuidance, LanPaint_KSampler, SetLatentNoiseMask, and image-comparison nodes. The result is a more controlled editing process where the target body and scene remain stable while the head identity is transferred from the reference image.
The workflow also includes inpainting-style control. The LanPaint_KSampler route is used to focus the edit on the head replacement area rather than redrawing the entire image. This helps keep clothing, background, body pose, and overall composition closer to the base image. At the same time, the prompt emphasizes matching head tilt, rotation, gaze direction, micro-expression, lip position, face-to-body ratio, neck thickness, and shoulder alignment. These details are important for making the edit feel natural instead of pasted.
The upscale and finishing section makes this version more complete than a simple head-swap test. The workflow includes image scaling, latent refinement, VAE decoding, and comparison output, making it suitable for generating a clean final image rather than only a rough preview. It is useful for portrait retouching, authorized character replacement, fashion concept previews, AI model-shot editing, social media cover production, Civitai showcase images, and RunningHub workflow publishing.
In short, this is a full Flux2-Klein9B head replacement and upscale workflow: reference identity in, base image structure preserved, inpainting control applied, and final image refined for higher-quality output. If you want to see how the head-swap prompt is written, how the reference images are connected, and how the upscale finishing stage improves the final result, watch the full tutorial from the YouTube link above.
⚙️ Try the Workflow Online
👉 Workflow: https://www.runninghub.ai/post/2026299604819517442?inviteCode=rh-v1111
Open the link above to run the workflow directly online and view the generation results in real time.
If the results meet your expectations, you can also deploy it locally for further customization.
🎁 Fan Benefits: Register now to get 1000 points, plus 100 daily login points — enjoy 4090-level performance and 48 GB of powerful compute!
📺 Bilibili Updates (Mainland China & Asia-Pacific)
If you are in Mainland China or the Asia-Pacific region, you can watch the video below for workflow demos and a detailed creative breakdown.
📺 Bilibili Video: https://www.bilibili.com/video/BV1WHfSBtE4u/
I will continue updating model resources on Quark Drive:
👉 https://pan.quark.cn/s/20c6f6f8d87b
These resources are mainly prepared for local users, making creation and learning more convenient.
⚙️ 在线体验工作流
👉 工作流： https://www.runninghub.ai/post/2026299604819517442?inviteCode=rh-v1111
打开上方链接即可直接运行该工作流，实时查看生成效果。
如果觉得效果理想，你也可以在本地进行自定义部署。
🎁 粉丝福利： 注册即送 1000 积分，每日登录 100 积分，畅玩 4090 体验 48 G 超级性能！
📺 Bilibili 更新（中国大陆及南亚太地区）
如果你在中国大陆或南亚太地区，可以通过下方视频查看该工作流的实测效果与构思讲解。
📺 B站视频： https://www.bilibili.com/video/BV1WHfSBtE4u/
我会在 夸克网盘 持续更新模型资源：
👉 https://pan.quark.cn/s/20c6f6f8d87b
这些资源主要面向本地用户，方便进行创作与学习。