GPT Image

GPT Image Overview

The GPT Image Configuration uses Open AI’s powerful image generation API to modify photos using simple text prompts. Survey data can also be used within the prompt to dynamically control the prompt itself.

*Works with Multiple Guests

In this Guide

• Examples
• GPT Image Settings
• Other Settings
• Using Survey Data

Examples

GPT Image Settings

Prompt

The Prompt is the text you enter to describe the image or image modification you want to create. The Image Prompt can contain Survey Tokens to create dynamic prompts based on data collected at the Photo Booth.

For example, to turn the image in to a Claymation style image, you could try the prompt:

Redo this image in the Claymation style, handmade clay textures, visible fingerprints and smudges, slightly uneven proportions, stop-motion aesthetic, soft lighting, expressive sculpted faces

Dimensions

These are the dimensions of the final processed image that will be returned to Breeze. When adding to your print layout, make sure the image placeholder is the same ratio, to avoid cropping. The uploaded image does not need to match the output’s size and ratio.

*1024×1024px is the smallest size (by pixels) and is therefore the fastest size to process.

Quality

Lowering the quality will make the processing time faster, but, may also effect the interpretation of the prompt giving a worse result, or a lesser likeness to the original image. By default the quality is set to High.

Background Transparency

This works similarly to the Background Removal option, except the background removal is processed by the GPT Image configuration itself, and can only be applied to the final processed image.

*the remove.bg option in some circumstances will give better results. Please test accordingly.

Target Image

The Target Image is an optional image that can be uploaded to be used as a style reference. This is useful for when you are trying to achieve a consistent result. In your Prompt, you can say something like:

make the uploaded image look like the second image

Mask Image

The Image Mask allows you to specify what parts of the uploaded image should be edited. You need to use an image that is the same size as the uploaded image, and then use solid black to specify the areas you want edited, and transparent the areas you do not.

!Open AI have acknowledged that this function does not currently work as reliably as intended. But will be fixed in a future update.

Source Image

The Source Image is for uploading your own test image. Once you have uploaded the image, press the Test button.

Other Settings

Additional processing can be run on the uploaded or processed image. See below to learn more about each option.

• remove,bg Apply background removal to uploaded or processed images.
• NSFW Scan the uploaded or processed image for inappropriate content.

Using Survey Data

The Image Prompt can include Survey Data collected at the photo booth. This allows you to create dynamic prompts where the output can be fine tuned to the user.

Example

A Sports Trading Card Photo Booth, where you have two survey questions:

• Pick your Team Color {survey1, radio, color}
• Enter your Shirt Number: {survey1_text_shirtNumber}

Prompt

the person in the photo is a professional basketball player.
The color and style of the uniform is mainly {survey1_radio_color} with white features.
The number {survey1_text_shirtNumber} is on the shirt.