Skip to main content
AI Image Editor logoAI Image Editor

Explore GPT Image 2 with a clear creative brief

An editorial still life with a precise brief

Object placement, material and light give this simple scene a coherent direction without depending on generated lettering.

Illustrative editorial photograph of a coral notebook, ceramic cup and silver pen on a clean desk

A quiet desktop composition

A coral notebook provides one clear color accent. A ceramic cup and a silver pen balance the scene, leaving open space for a page layout.

  • Composition: three objects with visible separation.
  • Materials: matte paper, glazed ceramic and silver metal.
  • Review: pen geometry, cup rim and shadow direction.

GPT Image 2 request

Use GPT Image 2 if available to create an editorial still-life photograph. Place a closed coral-red notebook, a small off-white ceramic cup and a slim silver pen on a pale desk. Soft daylight from the upper left, gentle shadows, natural paper and ceramic textures. Keep the arrangement simple with generous empty space. No text, logos or hands. Tell me which model is used.

Illustrative example; not a model benchmark.

Make a GPT Image 2 request in three steps

The page starts a conversation. Include the model name and your visual requirements in the request.

01Choose a new image or an edit

Sign in. For a new image, describe the subject and setting. For an edit, add one clear reference and identify the part you want changed.

02Name the model and constraints

Write “use GPT Image 2 if available,” followed by the creative brief. Separate details to preserve from the changes you want, especially for a reference edit.

03Continue in chat and inspect

Submit the request and review the model reported for the image task. Availability can affect selection. Inspect the result before asking for another generation, which uses additional credits.

Give the model a specific visual task

These are useful briefs to explore in the image workflow. The result still needs review against your intended use.

Text-to-image scene building

Describe objects, setting and viewpoint for a new image. A few well-defined elements make the scene easier to judge and revise.

Reference-guided changes

Change a background, atmosphere or surface while keeping named features. GPT Image 2 is configured here for one reference per edit.

Visual direction studies

Compare warm, tactile and cool, restrained treatments. Keep the subject constant to make the difference in art direction clear.

Focused follow-up edits

Reference a returned image for a revision. State what works and isolate the remaining issue, such as distracting props or strong shadows.

Build a request you can evaluate

A model name is only one part of a useful brief. Describe visible success criteria as well.

Request GPT Image 2 explicitly

This page does not lock an image model. Include its name in your request and check the model shown with the task.

Keep the reference unambiguous

Start a GPT Image 2 edit with one image. Explain any other attachments and check the model’s supported reference count.

Preserve details by naming them

Name the coral cover, bottle silhouette or window placement. A general request to “keep it identical” cannot guarantee unchanged details.

Separate image making from layout

Request empty space for a header and add text afterward. Review edges, reflections and small details before treating the image as finished.

Explore other ways to work with images

Use a broader workflow when the creative task matters more than a particular model.

AI image generator

Develop photos, illustrations and visual concepts from a text prompt.

Image to image

Explore how references guide style, lighting and scene transformations.

AI photo editor

Start with the specific change you want to make to an existing photo.

ChatGPT Image 2 questions

How this page relates to models, accounts and image tasks.

No. AI Image Editor is an independent service with its own image workflow and model options. This is not the ChatGPT website or an OpenAI account page.

Turn your idea into a specific image request

Name GPT Image 2, describe the scene and add a reference if you want an edit. Check the model used as you continue in chat.

Start a model requestExplore text-to-image