Descriptions:
Tired of AI image editors shifting your subjects or changing your aspect ratio? 🎨 Say hello to JoyAI Image Edit, the latest open-source model suite developed by JD.com!
In this tutorial, we dive deep into the two dedicated variants: Single (for surgical single-image editing like material swaps, object removal, and 360° camera rotation) and Plus (for multi-image reference composition). Most importantly, I will share an Optimized Padding & Cropping Workflow that completely solves the notorious “pixel misalignment” issue.
🔥 What you’ll master in this tutorial:
• 🧠 Single vs. Plus Architecture: Pairing the right Diffusion Model and Text Encoder with Wan 2.1 VAE.
• 📐 100% Pixel Alignment Hack: How to use proportional padding and exact cropping to lock original boundaries.
• 🖌️ Precision Single Editing: Object removal (with shadows & tile textures), material swaps (amber to translucent jade), and object repositioning.
• 🎥 360° View Generation: Rotating camera angles (Yaw, Pitch, Zoom) from a single reference photo.
• 👥 Multi-Image Assembly (Plus): Combining character identity, outfits, and scenes without “pasted-on” visual glitches.
#runninghub #JoyAIImageEdit #JoyAI #ComfyUI #AIImageEditing #PixelAlignment #Inpainting #Wan21VAE #GenerativeAI #AITutorial #ImageToImage #QwenImage #AIWorkflow #RunningHub #MachineLearning #AIGraphicDesign #PhotoEditingAI #AIModelTest
————————————————————————————————————————
resource
https://huggingface.co/jdopensource/JoyAI-Image-Edit-ComfyUI
https://huggingface.co/jdopensource/JoyAI-Image-Edit-Plus-ComfyUI
Runninghub Fan Benefit: Register to get 1,000 RHB
https://www.runninghub.ai/?utm_source=kol02-RH021
workflow and prompt engineering download：
https://github.com/amao2001/ganloss-latent-space/tree/main/workflow/2026-07-22%20joyai_image_edit
————————————————————————————————————————
