
Convert your baby photo into a needle-felt wool doll miniature with diorama vibes. Pick from 4 outfits and 4 background colors for Instagram-friendly viral shots.
Prompt
[Input]
- output_size: 1024x1366 (3:4 portrait)
- reference: 1 baby photo
[Purpose]
Convert the user's attached baby photo into a needle-felted wool doll miniature. The doll is photographed in a miniature studio setting with soft lighting, Instagram-friendly viral aesthetic.
[Scene — image-to-image edit, preserve identity as felt doll]
Use the user's attached photo as the base. PRESERVE the baby's identity (face proportions, age, hair style, skin tone) BUT recreate the baby AS A NEEDLE-FELT WOOL DOLL (not a photo, not a 3D render). The felt doll has the user's child's recognizable features rendered in soft wool yarn material.
Felt doll details:
- Standing upright, full body or upper body visible
- Wool yarn hair matching the user's photo (length, color, style — preserved)
- Felt facial features: oversized cute eyes, tiny embroidered nose, soft mouth
- Soft hand-stitched detail on clothing
- About 12-15cm scale (miniature feel)
Outfit:
{outfit_description}
Background:
{background_description}
[Style]
- Needle-felt wool doll aesthetic (visible wool yarn texture)
- Soft warm studio lighting with subtle shadows
- Slight depth-of-field blur on background
- Reminiscent of Pinterest/Instagram felt doll artist photography
- Color palette warm and cozy
[Composition]
- 3:4 portrait
- Doll occupies middle 70-75% vertical
- Background seamless solid color matching the selected palette
- Subtle shadow under doll grounding the figure
[CRITICAL — Identity preservation as felt doll]
- Face features (proportions, eye shape, age) must match user's photo
- BUT rendered as felt wool doll (NOT photo, NOT 3D, NOT plush)
- Korean Asian baby's wool doll version
[Text]
- No text anywhere
[🚫 PROHIBITED]
- Photographic real baby (must be CLEARLY a felt doll)
- 3D rendering / Pixar style (needle felt wool only)
- Plastic doll / porcelain doll (wool felt material only)
- Branded characters (no specific franchised dolls)
- Text or watermarks
Click to open ChatGPT with the prompt auto-filled.
Click the "Open in ChatGPT" button on the prompt box. A new chat opens with the prompt pre-filled. Drag your photo and send to generate the result.
Face preservation is a known weakness of ChatGPT. Retry with the same prompt or use a clearer front-facing photo. Gemini 2.5 Flash Image (Nano Banana) tends to be more stable.
ChatGPT Korean text rendering is still imperfect. Emphasize "make Korean text clearly readable" or switch to English for stability.
In ChatGPT, ask for "another variation" or "another variation" to retry with the same prompt.
Output resolution is sufficient for A4, 5×7, and 8×10 inch prints. For KakaoTalk sharing, add "1:1 square ratio" to the prompt.