Prompt transformation
This article refers to the Microsoft Foundry (new) portal.
Scope
Prompt transformation (also called prompt revision) applies to Azure OpenAI image generation requests that support therevised_prompt field, including DALL-E 3 deployments. If your deployment doesn’t return revised_prompt, prompt transformation isn’t available for that model.
In API responses, this feature appears as the
revised_prompt field. This article uses “prompt transformation” and “prompt revision” to refer to the same behavior.Overview
Prompt transformation is a process included in the DALL-E 3 models that applies a safety and quality system message to your original prompt. It uses a large language model (LLM) to add a message before sending your prompt to the image generation model. This system message enriches your original prompt with the goal of generating more diverse and higher-quality images while maintaining intent. The processing flow works as follows:- Your input prompt is revised to improve quality and add safety guidance.
- Azure OpenAI content filtering evaluates the prompt and, when applicable, the generated output.
- If filtering blocks the request, the service returns a response indicating the filtered category and severity.
Prompt transformation example
The following table shows an example prompt and the difference in generated images with and without prompt transformation.The images shown are illustrative. Because image generation and prompt revision are non-deterministic, your outputs might differ.
| Example text prompt | Example generated image without prompt transformation | Example generated image with prompt transformation |
|---|---|---|
| ”Watercolor painting of the Seattle skyline” | ![]() | ![]() |
Why is prompt transformation needed?
Prompt transformation is essential for responsible and high-quality generations. Not only does prompt transformation improve the safety of your generated image, but it also enriches your prompt in a more descriptive manner, leading to higher quality and descriptive imagery. Default prompt transformation includes safety-oriented guidance intended to reduce requests for disallowed content, such as copyrighted or trademarked characters, public figures, and content in categories such as sexual, hate and unfairness, violence, and self-harm. For enforcement behavior details, see Content filtering.Use prompt transformation
Prompt transformation is applied by default to all Azure OpenAI DALL-E 3 requests. No extra setup is required to benefit from prompt transformation enhancements.Prompt transformation can’t be disabled or customized for DALL-E 3 deployments in Azure OpenAI. The revised prompt is always generated automatically.
Availability
- Prompt transformation applies to Azure OpenAI DALL-E 3 image generation requests.
- Availability can vary by region and by model deployment. For region availability, see Azure OpenAI models.
- The transformed prompt is returned via the API regardless of whether you use the Foundry portal or call the REST API or SDK directly.
View prompt transformations
Your revised or transformed prompt is visible in the API response object as shown here, in therevised_prompt field.

