Cart Log in
IMAGES

Commands for ChatGPT

The visual codes that steer ChatGPT when it generates an image. Each one with its example and the exact prompt to reproduce it.

What they are

They are prefixes written at the start of a prompt to steer the kind of representation you want. They are not a fixed official menu: their effect depends on the model, the context and how detailed the prompt is. To see the shortcuts enabled on your account, type / in a new chat.

How to use them

Put one or more codes at the start, then add the subject, the must-have elements, the style, the format and the constraints. To edit an image, open it and state the change precisely.

/[code] /[secondary code] + subject + what it must show + style + language/labels + format + constraints

Recommended combinations

Two codes together are far sharper than one. On the left, the reference image; on the right, the result with the codes applied.

/xray /infographic

Anatomy of an SEO strategy

Reference · A flat infographic of the same strategy, with no layers or visible interior.
With the codes applied
Prompt for this image
/xray /infographic anatomy of an SEO strategy: keyword research, content, internal linking, technical SEO and measurement. Clean style, English labels, vertical 4:5 format.
/blueprint /explodedview

A web server in exploded view

Reference · The same server assembled, as a conventional product photo.
With the codes applied
Prompt for this image
/blueprint /explodedview a web server: rack, CPU, RAM, storage, network and cooling system. Blue technical plan, parts ordered and annotated.
/beforeafter /ui

Agency homepage, before and after

Reference · The cluttered homepage alone, with no comparison.
With the codes applied
Prompt for this image
/beforeafter /ui an agency homepage: left, cluttered and slow design; right, modern, fast and conversion-driven. 16:9 format.
/handwritten /flowchart

How an AI agent works

Reference · The same diagram in a clean corporate style, with no hand-drawn stroke.
With the codes applied
Prompt for this image
/handwritten /flowchart explain how an AI agent works: user input, data retrieval, reasoning, tool calls and response. Notebook style with arrows and sticky notes.
/xray /explodedview

Ultrathin laptop from the inside

Reference · The closed laptop on a desk, as a normal product photo.
With the codes applied
Prompt for this image
/xray /explodedview an ultrathin laptop. Show screen, battery, SSD, motherboard, fans, speakers and ports. Photorealistic technical illustration, dark background, parts separated and clearly identifiable. No brands or logos. 16:9 format.

Useful modifiers

Add them at the end of the prompt to adjust framing, light, aesthetics and limits.

Format
16:9 format square 1:1 vertical 4:5 story 9:16
Composition
close-up top-down shot isometric view rule of thirds negative space on the left
Light
soft window light cinematic light golden hour studio light
Aesthetic
photorealistic editorial clay-style 3D watercolour vintage engraving vector pixel art
Consistency
crisp detail realistic materials consistent shadows
Constraints
no text no logos no watermarks no people plain background

Good practice and limits

  • Be specific: state what must appear, what must be kept and what must not change.
  • If you need text in an infographic, write the literal wording, ask for legible text and cut the number of blocks.
  • For local edits, select the area of the image and describe only the change you need.
  • A code like /xray does not produce a real X-ray or inspection: it generates a visual interpretation. Never use it for medical, technical or safety diagnostics.
  • Always check text, proportions, labels and brand details before publishing the image.

Reusable template

Copy this structure and fill in the brackets. It works with any code on the list.

/[code] /[secondary code] Create [image type] for [purpose].
Main subject: [what must appear and what it is doing].
Setting: [place, era, atmosphere].
Style: [desired aesthetic].
Composition: [framing, position, space for text].
Lighting and colour: [light type and palette].
Format: [1:1, 16:9, 4:5 or 9:16].
Constraints: [what to leave out or what to preserve].
CLAUDE SKILLS

What if yours should be a Skill?

Turn your image workflow into a custom Skill. No code, in minutes.

Create your own Skill ↗
Vellius
Vellius · Digital asset management

Your digital assets, with security first

Custody, risk control and AI strategy bots in a single dashboard. Backtests, calibrated parameters and real-time execution, with the whole process in plain sight and available on mobile too.

  • Single dashboard
  • AI strategy bots
  • Risk control
  • Mobile ready
Join now