Cinematic portrait prompt

The same look needs different wording per model — the fields are identical, only the formatting changes. Copy the version your tool actually reads:

A cinematic waist-up portrait of the uploaded person, soft window light, muted earth tones, shallow depth of field, shot on an 85mm lens.

the uploaded person, waist-up, centered, cinematic, muted earth tones, soft window light, 85mm, f/1.8 --ar 1:1 --style raw --v 6

the uploaded person, waist-up, centered, cinematic, muted earth tones, soft window light, 85mm, f/1.8, highly detailed
Negative prompt: blur, extra fingers

The cinematic portrait look is built on three levers: soft window light for a gentle, directional key, an 85mm f/1.8 lens for flattering compression and creamy bokeh, and a muted earth-tone grade that gives it the restrained, filmic mood. The prompt names all of them explicitly, which is why it reads as cinematic rather than just well-lit. Waist-up, centered framing keeps the subject grounded and the shallow depth of field separates them cleanly from the background. The negatives — blur, extra fingers — clean up the usual portrait failure modes. Tip: feed it a sharp, evenly lit source photo; the muted grade and soft light land best when the input isn't already heavily filtered or backlit.

Cinematic is a look you can build from three concrete dials, and the prompt names all three. Soft window light gives a single, gentle, directional key — the flattering, slightly moody quality of natural light through glass, not a flat on-camera flash. The 85mm f/1.8 lens does two jobs at once: 85mm compresses features for a natural, attractive face, and f/1.8 melts the background into creamy bokeh that separates the subject. The muted earth-tone grade — desaturated, warm-leaning, restrained — is what film colour science looks like, and it's the difference between a photo and a frame from a film. Waist-up, centered framing keeps proportions honest, and the negatives clean up blur and the classic extra-fingers failure. Skip any one of the three dials and the model defaults to a flat, evenly lit snapshot, which is exactly why they are spelled out rather than implied.

Use a sharp, evenly lit photo that isn't already heavily filtered or strongly backlit. The muted grade and soft key land best on a clean source where the face is clear and front-on. Avoid hard direct sun, deep shadows, or a busy background that competes with the shallow depth of field. Soft daylight and a simple backdrop give the model the most room to apply the cinematic look.

1. 1Feed it a sharp, evenly lit source that isn't already heavily filtered or backlit.
2. 2Copy or load the prompt and keep the 85mm, f/1.8 and muted earth-tone grade.
3. 3Generate with Midjourney or Flux (or an editing model for likeness) and check the shallow depth of field.

Expect a filmic, waist-up portrait: soft directional light, a creamy out-of-focus background, and a restrained, warm-muted grade — a frame from a film, not a snapshot. You should stay recognisable with natural proportions. The grade and bokeh shift between runs, so generate two or three and choose the most flattering; that beats over-stylising a single result past the point of looking real.

• Don't skip the lens and grade — without 85mm / f/1.8 / muted tones it looks like a flat snapshot.
• Keep the waist-up framing so proportions stay natural and flattering.
• For a true likeness, restyle your photo with an editing model rather than generating a new face.

• Flat snapshot lookkeep all three dials: 85mm, f/1.8 and the muted earth-tone grade.
• Distracting backgroundlean on the shallow depth of field and start from a simpler source.
• Face proportions feel offkeep the waist-up, 85mm framing rather than a wide selfie.

• Swap the muted earth tones for a cool teal-and-orange blockbuster grade or a moody low-key look.
• Move the window light to hard side-light for a more dramatic, high-contrast portrait.
• Change the lens to 35mm for an environmental portrait that shows more of the setting.

Which model nails the cinematic look?

Midjourney and Flux, which excel at filmic lighting and shallow depth of field. DALL·E follows the plain-language version reliably when you want a quick, faithful read of the prompt.

How do I keep the subject looking like themselves?

Start from a clear, front-lit photo and, for likeness, use an editing model that restyles your image rather than generating a new face. Keep the 85mm framing so proportions stay natural.

What is the common mistake?

Skipping the lens and grade. Without 85mm, f/1.8, and muted earth tones the output looks like a flat snapshot — those exact parameters are what give the portrait its cinematic feel.

How do I use this prompt?

Copy the prompt, open the tool, and click "Use this template" to load it — then upload a portrait photo and generate.

Which models does it work with?

It is written for the models listed above. Switch the format in the tool to get a version tuned for ChatGPT, Gemini, Midjourney, Flux or Stable Diffusion.

Related templates

Golden hour landscape promptNano Banana 3D figurine promptGender swap filter prompt