① Create a "Polar Bear" and a "Panda" wearing ice skates, along with the background location. Normal chat-based input is also acceptable. Whether you generate them all at once or separately, the process is the same. ② Perform annotation on the image. This is the most troublesome part, so I made it a tool. Based on the file name, I create a prompt that organizes "which part of which image to reference." (Pose / Background / Face / Clothing, etc.) ③ Add supplementary information such as composition and color scheme for the final generation.



{ "subject": { "description": "Young woman with silver-grey hair, tied in a messy, textured bun with loose, wispy strands framing the face. Facial features include striking green eyes, clear skin, and a defined jawline. Her mouth is c

{ "subject": { "identity": "Megan Fox", "appearance": { "facial_features": "Precise likeness of Megan Fox, signature almond-shaped hazel eyes, soft jawline, natural lip fullness, faint freckles visible on nose bridge",

Luffy and Goku having an epic fight scene on the sunny 路飞和悟空正在桑尼号上进行一场史诗般的战斗场景

Hyper-realistic premium fitting room/boutique mirror selfie photography, vertical aspect ratio approx. 9:16, a young adult East Asian woman standing in front of a large full-length mirror in a dark minimalist changing space for a selfie. Us


Extreme low-angle fisheye lens photograph of a young woman with long dark hair making rock sign hand gestures. She is wearing a Y2K street style outfit: a cropped denim vest over a white top, baggy olive-green cargo pants, a studded belt, c

{ "subject": { "identity": { "biometric_reference": "Ana de Armas", "ethnicity": "Cuban-Spanish", "age_representation": "mid-30s" }, "facial_features": { "craniofacial_structure": "heart-shaped facial m