4 ChatGPT Shortcuts That Turn Any Product Photo Into a Technical Diagram

Upload a photo of a car. Type one word. Get back a cross-section render showing the battery pack, chassis and wiring sitting underneath the bodywork.

That is the promise behind a set of image prompts currently making the rounds on Instagram: /cutaway, /explodedview, /360view and /anatomy. Four keywords, four completely different ways of looking at the same object.

They work. But not for the reason most people assume, and understanding the difference is what separates a usable result from twenty minutes of frustration.

What these “shortcuts” actually are

Let’s clear this up first, because it changes how you use them.

/cutaway is not a built-in ChatGPT command. Neither are the other three. There is no registry inside ChatGPT that maps /anatomy to a dedicated image pipeline, and no hidden feature being unlocked. If you were expecting a secret menu, there isn’t one.

What is actually happening is more interesting. You are handing the model a single, unusually precise word and letting it infer everything else. “Cutaway” and “exploded view” are established terms of art in technical illustration, with more than a century of labelled examples behind them. The model has seen thousands of those examples. So when you attach an image and type the term, it already has enough to reconstruct your intent.

That is why these often land on the first attempt. It is also why they sometimes return three paragraphs of text instead of a picture. A one-word prompt is a coin flip on a bad day. The expanded prompts below remove the guesswork.

Use the shortcut when you are moving fast. Use the full prompt when the output has to be right.

1. /cutaway — see inside a sealed object

Slices away part of the outer shell to expose the internal structure, as though someone took a bandsaw to one side and pulled the panel off.

Best for: vehicles, appliances, machinery, buildings, consumer electronics — anything with an interesting interior hidden behind a housing.

The full prompt:

Generate a technical cutaway illustration of the [subject] in this image. Remove a section of the outer shell on one side to reveal the internal components in their correct relative positions. Keep the exterior styling, proportions and colour of the original intact. Clean studio background, three-quarter front view.

2. /explodedview — pull it apart into components

Separates an object into its constituent parts, suspended in space along a single axis in assembly order. This is the visual language of an IKEA manual or a patent filing.

Best for: watches, headphones, power tools, furniture, cameras — anything assembled from discrete parts that fit together in a defined sequence.

The full prompt:

Create an exploded-view diagram of the [subject] in this image. Separate each component along a vertical axis in assembly order, evenly spaced, with thin leader lines connecting each part to its position. Maintain the original materials and finish. White background, isometric perspective.

Worth flagging: this is the shortcut most likely to need a second pass. The model has to invent internals it cannot see in your photo, so a watch movement or a camera’s inner assembly will look convincing rather than correct.

3. /360view — every angle in one image

Produces a turnaround sheet: front, back, both sides, top, bottom and the three-quarter angles, arranged as a single contact sheet.

Best for: e-commerce listings, design reviews, 3D modelling reference, pitch decks — anywhere you would otherwise need to book a photo shoot.

The full prompt:

Generate a 360-degree turnaround sheet of the [subject] in this image. Arrange nine views in a 3×3 grid: front, left side, right side, back, back-left, back-right, top, bottom, and a centre panel. Label each panel with its angle. Keep lighting, scale and the white background consistent across every view.

4. /anatomy — label every part

Adds callout lines and short labels naming each layer or component. This is the one that produces those “Anatomy of a Big Mac” graphics — bun, patty, sauce, pickles, each with a hairline pointing at it.

Best for: food and beverage marketing, educational material, onboarding documentation, product explainers.

The full prompt:

Create a labelled anatomy diagram of the [subject] in this image. Separate the layers vertically so each one is clearly visible, then add callout lines with short text labels naming every component. Editorial infographic style, muted background, consistent sans-serif type.

Four things that make the difference

Start with a clean reference image. Single subject, even lighting, plain background, decent resolution. A cluttered photo forces the model to guess where your object ends and the background begins, and those guesses show up in the render.

Name the subject explicitly. Replace [subject] with a real description — “this 2024 electric SUV,” not “this.” Image recognition is good but not perfect, and a misidentification at step one cascades through everything downstream.

Specify the visual style. Blueprint, photorealistic, isometric line art, editorial infographic. If you leave it open the model picks for you, and its pick drifts between attempts, which makes a consistent set of images almost impossible.

Iterate inside the same conversation. “Move the battery pack lower and relabel it” will get you further than starting fresh. A new thread throws away the context that made your first attempt nearly right.