This prompt makes one visible change to a fully clothed adult portrait: it reshapes the fabric-covered bust so the silhouette looks fuller, rounder, and more lifted. It does not create a 3D model. It asks an image editor to change a two-dimensional contour, fabric tension, and local shading while treating the person, pose, outfit, and scene as locked.
Start with a source image the model can edit cleanly
Use a clear, front-facing or slight three-quarter portrait with the complete head and upper body visible. Opaque fitted clothing makes the requested contour easier to see and review. Loose layers, crossed arms, cropped shoulders, hard side light, busy patterns, and hair covering the chest give the model less reliable evidence about the original pose and garment.
Use only an image of an adult that you own or have explicit permission to edit. Do not use a child, an age-ambiguous person, a scraped social-media photo, a public figure, a former partner, or another adult who did not agree to this specific edit.
Choose the strength before you generate
The prompt offers three bounded levels. Replace the bracketed variable with one option rather than adding several conflicting adjectives.
| Setting | Expected visual difference | Use it when |
|---|---|---|
| Subtle | A small change in support, curvature, and local shading | The source already has a balanced silhouette and needs only light refinement |
| Clear | An easily visible Before/After difference without becoming the only subject of the portrait | You want a dependable default for a normal fashion or profile image |
| Strong but natural | A substantially fuller and more lifted contour on the same slender frame | You want the difference to read immediately while preserving believable anatomy and full coverage |
The retained test used strong but natural. The After candidate is intentionally more obvious than a conservative beauty retouch, but it still keeps the high crew neckline, opaque fabric, slim waist, neutral stance, and non-sexual studio presentation.
Compare the retained Before and After
The two retained files are both 1024x1536 PNG images. In the author's visual review, the face, adult age cues, hairstyle, expression, arm position, waist, trousers, shirt color, high neckline, crop, lighting direction, and background remained consistent enough for a focused comparison. The bust contour became visibly fuller and higher, and the ribbed fabric gained plausible tension and shading around the edited area.
Run a four-part review before keeping the result
- Identity: compare facial proportions, hairline, expression, age cues, skin marks, and natural asymmetry with the original.
- Boundary: check the shoulders, waist, hips, arms, hands, neckline, and background for changes the prompt did not authorize.
- Fabric: follow seams, ribs, patterns, folds, and shadows across the edited area. Reject broken lines, mirrored texture, hard circular edges, or unexplained highlights.
- Proportion: view the complete portrait rather than judging only a chest crop. Reject an extreme, asymmetric, or anatomy-defying result.
If the change is too weak, rerun from the untouched source with the next named strength. Do not repeatedly edit the latest result: stacked generations make identity and garment drift harder to detect. If the change is too large, return to the source and choose clear or subtle.
Know when this prompt is the wrong tool
Do not use it for minors, age-ambiguous subjects, non-consensual edits, nudity, lingerie, transparent garments, sexualized presentation, humiliation, medical prediction, or cosmetic-surgery simulation. OpenAI's moderation reference treats text and image inputs as content that may be classified for potential harm; a prompt page cannot guarantee that any specific future input or output will be accepted.
This prompt also should not be used to claim that one body shape is objectively better. It is a controlled visual-editing recipe. The person shown in the retained example is fictional, and the result is evidence for one bounded test—not a promise that every photo, body type, product, account, or future model will behave the same way.
What the retained test supports
The test supports a narrow conclusion: on one synthetic, fully clothed adult portrait, the strong-but-natural direction produced a clear contour change without obvious identity, pose, neckline, outfit, or background replacement at normal viewing size. It does not prove exact non-target preservation, anatomy correctness, consent for another person's photo, destination-platform approval, or repeatability across tools. Treat every output as a candidate that must be compared with its untouched source.