Photo styles with ChatGPT: What the ten image codes really do

Stefan Petri
published:
Reading time: 6 min
Images & text created with the help of AI
Photo styles with ChatGPT: What the ten image codes really do

A single word, and suddenly your photo looks like a night lightning shot or an instant photo? It doesn't work that automatically. Terms like /flashphoto or /polaroid are not official ChatGPT commands. However, they can serve as brief style hints for an image model – if you specify what should happen in the image.

For everyday work, this means: You can quickly test an image idea, develop a variant for social media, or sketch a visual style for a shoot. Photoshop and other image editing programs are not rendered unnecessary by this. They still give you more control over individual pixels, layers, and outputs. Image editing with AI is primarily an additional way to develop image ideas and try out changes.

What these codes actually are

A slash does not make a word a special function. /flashphoto is initially a text block for an image model – similar to “harsh direct flash, night reportage aesthetic.” Some models recognize English terms or detailed descriptions more reliably than self-invented abbreviations. If a code doesn’t deliver the desired result, express the effect in words instead.

It is also important to understand the difference between image generation and image editing. In a new generation, you describe the scene and style. If you upload an existing photo, you should also specify what should remain unchanged. For example: “Only change the lighting. The face, clothing, and image cropping should stay the same.” How reliably this works depends on the model and the respective editing mode.

Good prompts connect three things: the subject, the desired photographic effect, and important boundaries. “Portrait of a dancer, harsh direct flash, dark club, natural facial features” is more specific than “Make the photo cooler.”

A creative person is reviewing a nighttime portrait on a laptop
A style term works best in conjunction with a clear description of subject and effect

Ten photo styles as prompt components

The following terms are practical starting points, not switches with guaranteed results. Attach them to a description of your scene and adjust them as needed.

  • /flashphoto: Describe a harsh direct flash, deep shadows, and a dark background. This fits, for example, with a spontaneous concert portrait. Specify whether the skin should look natural or consciously overexposed.
  • /silhouette: Name a person against a bright sunset and request a dark, low-detail subject. This keeps the sky bright while the person appears as an outline. In a real photo, however, the model's facial features may change more than expected.
  • /long exposure: Specify where the light trails should appear, for example, behind a parked bicycle on a street. Add a fixed camera and an unmoving environment, so the whole image doesn’t look blurred.
  • /fisheye: Request a strong wide-angle distortion with bulging lines and a far-reaching image center. This is suitable, for instance, for a dynamic skateboarding motif. State whether the look should be subtle or deliberately exaggerated.
  • /motion blur: Define what stays sharp and what blurs. For example: “Runner sharp, city in the background horizontally blurred.” Motion blur alone does not specify whether the camera, subject, or background is moving.
  • /reflection: Determine the reflecting surface and the ratio of subject to reflection: for example, a face behind a slightly fogged window at night. Without these details, the reflection can appear anywhere.
  • /double shadow: Describe two clearly separated, colored shadows, like magenta and blue on a studio wall. The term can also create two light sources or an unwanted double contour – so clarify what you mean.
  • /lens flare: Request a visible light flare from a light source at the edge of the image, such as the low sun. If the flare should cover the face, state that; otherwise, the subject should remain readable.
  • /polaroid: Specify instant photo border, slightly faded colors, and fine film grain. If the image should not look like a product photo of a framed print, add: “The photo fills the entire image area; only the image style should remind of an instant film.”
  • /dramatic shadow: Describe the direction and shape of the shadow, such as narrow blinds cast over a wall and part of a face. This gives more control than just writing “dramatic.”

Studio portrait with two colored shadows
Light color, direction, and shadow shape can be specifically named in the prompt

How to formulate a useful image prompt

You don’t need to know all the technical details. What matters is translating the effect into verifiable image details. Instead of “like a movie,” you can, for instance, write “warm backlight, deep shadows, slightly desaturated colors.” Instead of “movement,” specify a sharp subject and a blurred background.

A simple structure helps in formulating:

  1. Subject: Who or what is visible, and what is the person doing?
  2. Image effect: Which of the ten styles fits? What light and color effects are included?
  3. Image composition: Close-up or wide shot, angle, and position of the subject.
  4. Boundaries: What should remain unchanged or definitely not appear in the image?

Example for an existing portrait: “Edit this photo as a night club portrait with harsh camera flash and dark background. Keep the face, hairstyle, clothing, and image cropping. Do not add any other people.” If the result looks too artificial, change only one point at a time: like “less harsh shadows” or “natural skin texture.” This helps you find out more easily which instruction makes the difference.

Where the limits are

An image model interprets your words instead of replicating a camera setting exactly. A prompt with “long exposure” can therefore create light trails but can also change wheels, faces, or street signs. With texts in the image, logos, hands, and fine details, you should check the result particularly carefully.

Even when editing an uploaded photo, it can happen that more is changed than intended. Compare the output with the original and check faces, products, brand features, and other important image parts. For a client project or a shop, such deviations are not just a matter of style: Check whether the result is technically correct and whether you are allowed to use the original image. Image rights and usage terms depend on the material used, the service, and your specific purpose.

An image prompt also does not replace a clean final retouch. If you need precise masks, consistent product colors, or print-ready files, a classic image editing program is often still the better step. Use the AI version as a draft or basis – not automatically as a finished result.

A short workflow for practice

First, choose a clear purpose: for example, a spontaneous portrait variant for a social media post. Upload the suitable image or describe your subject, name a style term, and add concrete details about light, background, and composition. Ask the model to keep important features. Then review the result and adjust one instruction at a time.

This is how the ten terms become useful: not as magic codes but as a short reminder of photographic techniques. They help you to name an effect. The clearer you describe the subject and desired change, the better you can assess whether the result fits your purpose.

Published on by Stefan Petri
Published on:
From Stefan Petri
Together with his brother Matthias, Stefan Petri runs the popular specialist forum PSD-Tutorials.de and the e-learning platform TutKit.com, which focuses on the training and further education of digital professional skills.

Comments

Register and comment.

Back to overview