Typing "make this look better" into ChatGPT almost never returns a better photo. Typing the exact mood you want, with the light, color, and grain spelled out, is a different story. Every prompt below was run through ChatGPT's built-in image editor on real portraits, street shots, and interiors, then adjusted until the look held up instead of falling apart on the second try. What is left is twelve copy-paste prompts, sorted by feeling, each built to do one specific thing well. Upload a photo, paste the one that matches the mood you are after, and change a single detail at a time. One thing to keep in mind first: ChatGPT reads your wording literally, so each prompt here names what should change and what should stay the same. That is the whole reason they work.

How ChatGPT edits photos right now

Verified, not guesswork

  • Upload a JPG or PNG, then describe the change in plain language, or use the Select tool to mark one area for a local edit.
  • The December 2025 image upgrade keeps composition, lighting, and a person's likeness steadier across edits, and renders text inside the photo more cleanly than older versions.
  • A selection-based edit leaves everything outside the marked area pixel-identical, which is why backgrounds survive a retouch.
  • OpenAI's own noted limits: some art styles, scenes with several faces, and certain non-English text are still hit or miss, so results are not perfect.
  • Free accounts can edit with daily caps. Paid tiers run faster, allow more edits, and unlock the deeper processing pass.

Five habits that sharpened every result

Name the mood and the mechanics"Warm and cozy" lands far better when the light direction, color temperature, and grain are written out too.
Edit in passesOne change per prompt stops the model from trading a fix in one spot for a new problem somewhere else.
Lock the faceAdd "keep facial features unchanged" so the person stays recognizable through a heavy color grade.
Borrow a referenceNaming a film stock or a lighting setup, like side light or Kodak Portra, steadies the color from the start.
Use Select for local fixesMark one area and only that area changes, leaving the rest of the frame exactly as shot.

The 12 prompts, grouped by mood

Each box is ready to paste after you upload a photo. Tap copy, drop it into ChatGPT, then tweak one detail if the look needs a nudge.

The master image used to run all the prompt is:

Warm and golden

1. Warm and cozy

When to reach for it: indoor portraits, autumn scenes, anything shot under warm lamps.

Edit this photo to feel warm and cozy. Push the white balance slightly toward amber, lift the shadows so nothing reads pure black, and add a soft golden glow as if lit by a nearby lamp. Keep skin tones natural and faces unchanged. Gentle contrast, light film grain, no heavy saturation.

What the test showed: landed on the first try most often. Watch that skin does not tip orange, and pull the amber back a step if it does.

Result:

2. Golden hour glow

When to reach for it: outdoor portraits and landscapes shot in flat midday light.

Edit this to look like golden hour. Add warm directional light from the side, push the highlights toward gold, deepen the shadows gently, and give the whole frame a soft sunlit haze. Keep faces naturally lit and avoid an orange cast on skin.

What the test showed: convincing side light and haze. Faces can pick up an orange cast, so keep the "naturally lit" line in place.

Result:

3. Vintage 35mm film

When to reach for it: everyday and street photos that feel too clean or digital.

Give this photo a vintage 35mm film look. Add warm faded tones, slightly lifted blacks, gentle grain, and a soft halation glow around the brightest areas. Mute the colors a touch and shift the palette toward Kodak Portra. Keep skin tones flattering and faces unchanged.

What the test showed: the faded tones and grain read as real film. Exact stocks are approximated, not matched to the last detail.

Result:

Cinematic and bold

4. Cinematic teal and orange

When to reach for it: travel, street, and portrait shots that want a movie-poster feel.

Grade this image with a cinematic teal and orange look. Warm the skin tones and highlights toward orange, cool the shadows toward teal, and add a subtle letterbox contrast. Preserve facial features and keep the result believable, not neon. Slightly desaturate the greens.

What the test showed: strong and believable at a light touch. Push it too far and greens go muddy, so keep the grade subtle.

Result:

5. Editorial high fashion

When to reach for it: studio portraits, outfit shots, anything meant to look like a magazine page.

Grade this like an editorial fashion shot. Clean, crisp contrast, refined skin tones, deep but detailed shadows, and a slightly cool, sophisticated palette. Sharpen the fabric and texture. Keep the face natural and flattering with no heavy retouching.

What the test showed: clean contrast and refined skin. Fabric texture sharpens well, and faces stay natural rather than plastic.

Result:

6. Bold and punchy

When to reach for it: food, products, and posts that need to stop a scroll.

Make this image bold and punchy. Increase the contrast and vibrance, deepen the shadows, and make the colors pop without looking artificial. Sharpen the main subject and keep skin tones realistic. Clean, high-impact, scroll-stopping, not oversaturated.

What the test showed: high impact without going cartoonish. Keep an eye on reds, which saturate the fastest of any color.

Result:

Soft and dreamy

7. Bright and airy

When to reach for it: interiors, weddings, and lifestyle photos that should feel light and open.

Make this photo bright and airy. Raise the exposure slightly, soften the contrast, and give it a clean white balance with a faint cool tint. Whites should feel crisp, shadows open and light. Keep detail in faces and avoid blowing out the highlights.

What the test showed: crisp whites and open shadows. Very bright windows can clip, so hold the "avoid blowing out" instruction.

Result:

8. Dreamy soft focus

When to reach for it: portraits, florals, and romantic frames.

Edit this into a dreamy, soft look. Add a gentle haze, slightly reduce the clarity, and bloom the highlights so light feels diffused. Give the color palette a pastel lean and keep the contrast low. Preserve the face sharply enough to stay recognizable while the rest stays soft.

What the test showed: the haze and bloom feel intentional, not like a mistake. Keep the face sharp so the subject stays clear.

Result:

9. Ethereal fairytale

When to reach for it: outdoor portraits and fantasy concepts.

Edit this into an ethereal, fairytale scene. Add soft glowing light, a faint mist, and a cool pastel palette with gentle highlights. Make it feel magical and dreamlike while keeping the subject recognizable. Light grain, low harsh contrast, no heavy filters.

What the test showed: soft glow and mist land nicely. Too much and the subject loses definition, so cap the haze if it spreads.

Result:

Moody and retro

10. Moody film noir

When to reach for it: black and white portraits and architecture with strong shapes and shadow.

Convert this to a moody black and white with film noir contrast. Deepen the blacks, keep the bright highlights, and add visible grain for a classic 35mm feel. Emphasize shape and shadow. Keep facial detail intact and avoid a flat gray look.

What the test showed: one of the most consistent looks here. Deep blacks keep their detail when you ask it not to go flat.

Result:

11. Nostalgic 90s camcorder

When to reach for it: candid and fun shots that suit a retro throwback.

Give this a nostalgic 90s camcorder feel. Slightly soften the image, add light VHS-style grain and a faint warm color shift, and reduce the dynamic range so it feels recorded rather than photographed. Keep the subject clear and avoid heavy distortion.

What the test showed: the softened, recorded feel is accurate. Avoid stacking heavy distortion on top, which tips it into a gimmick.

Result:

12. Minimal and muted

When to reach for it: neutral feeds, editorial sets, and calm lifestyle photos.

Apply a minimal, muted edit. Lower the overall saturation, use a soft matte finish with slightly lifted blacks, and keep a neutral, slightly warm white balance. Calm and understated. Preserve natural skin tones and fine detail in the subject.

What the test showed: understated and clean. If it reads washed out on your photo, nudge the contrast up a single notch.

Result:

Quick reference

MoodBest forThe look
Warm and cozyIndoor portraits, autumnAmber glow, open shadows, soft grain
Golden hour glowOutdoor portraits, landscapesSide light, gold highlights, sunlit haze
Vintage 35mm filmStreet, everydayFaded tones, halation, Portra palette
Cinematic teal and orangeTravel, portraitsWarm skin, teal shadows, light letterbox
Editorial high fashionStudio, outfitsCrisp contrast, cool palette, refined skin
Bold and punchyFood, products, socialHigh vibrance, deep shadows, sharp subject
Bright and airyInteriors, weddingsLifted exposure, clean whites, low contrast
Dreamy soft focusRomantic portraits, floralsHaze, bloom, pastel, low contrast
Ethereal fairytaleFantasy, outdoorSoft glow, mist, cool pastels
Moody film noirB&W portraits, architectureDeep blacks, grain, high contrast
Nostalgic 90s camcorderCandid, retroSoftened, VHS grain, warm shift
Minimal and mutedNeutral feeds, editorialLow saturation, matte finish, lifted blacks

What held up, and what needed a second pass

Held up across tests

  • Grades that named light direction and color temperature landed on the first try far more often than vague requests.
  • Faces stayed recognizable whenever the prompt said to keep features unchanged.
  • Black and white, warm film, and golden hour were the steadiest looks of the twelve.
  • The Select tool kept untouched areas exactly as shot, which matters for backgrounds and product shots.

Needed a second pass

  • Dense text inside a photo, like signs or labels, sometimes came back warped or misspelled.
  • Scenes with several faces occasionally drifted on the people farthest from the camera.
  • Very specific film stocks were approximated rather than matched precisely.
  • Tight crops and tiny details lost sharpness and wanted a quick Select-tool touch-up.

Final verdict

After running the same handful of photos through all twelve looks, one pattern was hard to miss: the prompts that named light direction, color temperature, and grain beat every one-word request, every single time. ChatGPT is at its best here for speed and mood, turning a flat phone photo into something with a point of view in a few seconds. It is not a stand-in for a real editor when pixel-level control is the goal, and dense text or crowded group shots still want a second pass. The move that saved the most redos was plain: lock the face by asking it to keep features unchanged, then edit in passes instead of one giant prompt. Treat these twelve as starting points, swap in your own film stock or lighting notes, and the hit rate climbs fast.