No more "invisible makeup" — 5-feature face analysis + real brand SKUs
Photo tips (not enforced)
Natural light, bare or sheer makeup, no glasses or large jewelry, face squarely to camera, 3 angles for triangulation. Filtered/made-up still works — results carry a ⚠️ flag.
Welcome to ChatIMG Image Assistant
Upload an image or describe the image you want to generate
Click or drag and drop an image
Supports multiple images — AI will blend them together
8 vibes × 8 locations × 5 outfit slots, same subject photo reused. Asakusa, Paris cafe, NY SoHo — without leaving home.
Eiffel / Great Wall / Mt Fuji / Santorini / Sahara / Iceland Aurora — 12 icons at one-second teleport.
8 gowns × 8 venues × 5 makeups × 5 aesthetics, lock the subject and remix freely.
6 ages × 6 eras × 5 visual styles. Preserves features, ages everything else — instant share-bait.
4 cultures / 10 dynasties × 6 scenes × 5 beauty looks, lock subject and remix.
5 semantic slots: subject / top / bottom / accessory / jewelry, plus free-text scene + style. Multi-image fusion.
Optimised for East Asian / K-beauty / J-beauty skin
Generate a full makeup guide from the diagnosis: - 4 sub-zones: brow (with brow tint HEX) / eyeshadow 4-color quad / blush 3 colors / lip 5 colors - Label brand SKU: 3CE / RMK / Dior / Perfect Diary / Etude House - Sample swatches with eyes-closed test patches - Top half-face BEFORE/AFTER (bare vs full) - Bottom strip: 4 looks (Daily / Office / Date / Wedding) as step-by-step
XHS verbatim: 5-feature analysis (face/eye/brow/lip shape + vibe) → 3 makeup tips + 10-color palette + 3-tier suitability comparison
From the uploaded portrait, create a premium makeup style analysis guide. Preserve original facial features and identity. Perform 5-feature analysis: face shape, eye shape, brow shape, lip shape, overall vibe. Based on these, recommend the suitable makeup style. Output 3 key makeup tips + a 10-color palette + side-by-side comparison of 3 makeup looks (Suitable / Normal / Unsuitable). Layout: clean editorial, like a pro stylist report. Visual-led, labels only, no paragraphs. High resolution, social-share optimized.