On September 20, 2023, OpenAI announced the DALL·E-3 model through their blog post. DALL·E-3 is OpenAI’s most advanced Text-to-Image Generator artificial intelligence model announced so far.
Here is our comprehensive guide to understand DALL·E-3.
DALL-E-3 Model: OpenAI’s AI Image Generator From Text
The DALL·E-3 model is the next model that follows OpenAI’s DALL·E-2, which was announced on March 25, 2022. The name DALL-E is a blend of Salvador Dali, the famous artist, and Pixar’s WALL-E movie.
According to OpenAI’s own statement, the DALL·E-3 model can understand more nuances and details compared to its predecessors, making it easier to generate the image you have in mind.
With the DALL·E-3 model, you can send text input to generate image output. This means you can have it generate images with the specific features you desire. These AI-generated images can be used in various fields as per your needs.
ChatGPT is a language model specialized in taking text input and providing text output. It was not specialized in image generation. On the other hand, the DALL·E-3 model is an advanced artificial intelligence model trained by OpenAI specifically for image generation.
How to use DALL-E-3?
OpenAI has announced that DALL·E-3 is currently in the research preview stage and will be available for ChatGPT Plus and ChatGPT Enterprise subscribers in October 2023. Additionally, they mentioned that later this fall, there will be usage availability in the Labs interface, which provides an easy way to use OpenAI’s models.
You can use DALL·E for free through Labs, while ChatGPT Plus, Enterprise, and the API require payment for access to the DALL·E model.
Using DALL-E-3 through Labs
Labs is the easiest interface through which you can use this model.
You can use the DALL·E model through OpenAI’s Labs interface without the need for an API call. Here are the steps to follow:
1. Visit the address labs.openai.com.
2. Sign in to your OpenAI account. If you don’t have an OpenAI account, create a new one similar to creating a ChatGPT account.
3. On the page that opens, you will see a large input window. Input the desired characteristics of the image you want, just like you would input to ChatGPT.
4. For example, if you want a drawing of a car with a big orange on top, describe it. After expressing the characteristics of the image you want, click the Generate button on the right.
5. Once you click the Generate button, the DALL·E model will start creating an image with the features you described. During this process, you will first see a loading screen.
6. After DALL·E has finished its work, it will show you the generated images. Sometimes it may produce multiple images.
7. You can select and zoom in on the image you like. You can also perform Edit, Share, and Save operations on the image if you wish.
How to use DALL-E-3 via API?
You can integrate the DALL-E-3 model into your applications by making API calls. Using the DALL·E model via the API requires programming knowledge and can be more challenging compared to other methods.
Developers may choose this method to integrate the DALL·E-3 model into their own applications. Below, we’ve shown how you can make API calls to invoke DALL·E models in Python.
Sign up for OpenAI API: If you haven’t already, you’ll need to sign up for the OpenAI API and obtain your API key. You can do this on the OpenAI website.
Install the OpenAI Python Library: You can install the OpenAI Python library using pip:
pip install openaiImport the Library and Make API Calls: Once you have the library installed and your API key, you can start using DALL-E. Here’s an example of how to call DALL-E to generate an image:
import openai
# Set your API key
api_key = "YOUR_API_KEY_HERE"
# Initialize the OpenAI API client
openai.api_key = api_key
# Define a prompt for DALL-E
prompt = "Generate an image of a cat with wings."
# Call DALL-E to generate an image
response = openai.Image.create(
model="image-alpha-001", # Use the DALL-E model
prompt=prompt,
n=1, # Number of images to generate
)
# Get the URL of the generated image
image_url = response.data[0]["url"]
print("Generated Image URL:", image_url)
Make sure to replace "YOUR_API_KEY_HERE" with your actual OpenAI API key. You can customize the prompt to specify what kind of image you want DALL-E to generate.
Retrieve and Use the Generated Image: The response object contains the URL of the generated image. You can use this URL to display or download the image as needed.
DALL-E-3 in ChatGPT
OpenAI has stated that the DALL·E-3 model is natively built into ChatGPT. But it will be only used by GPT-4 model in ChatGPT. GPT-3.5 model will not be supported.
Also read: How to use GPT-4 model in ChatGPT?
When generating images with the DALL·E-3 model, effectively expressing your desired image is crucial. In this process, you can also benefit from ChatGPT as a prompt generator tool.
Explain your idea to ChatGPT and ask it to provide you with a high-quality and detailed prompt for DALL·E-3.
If the image generated by the DALL·E-3 model isn’t exactly what you want, don’t worry. You can continue to request adjustments to the image and iterate until you achieve the best result.
How to use DALL-E-3 in ChatGPT?
OpenAI has announced that DALL·3 will be available to ChatGPT Plus and Enterprise subscribers in early October. As soon as it is made available, we will share with you how to use this model through ChatGPT.
Therefore, ChatGPT doesn’t provide you image output using DALL-E-3 model right now.
DALL-E-3 vs DALL-E-2: A Comparison
The DALL·E-3 model can understand more nuances and details compared to DALL·E-2, allowing it to generate images that align better with your preferences and requests.
You can view how the DALL·E-2 and DALL·E-3 models generate images using the same prompt below.
Prompt: An expressive oil painting of a basketball player dunking, depicted as an explosion of a nebula.
The generated image through DALL-E-2 model:
The generated image through DALL-E-3 model:
Do I have copyright rights to the images I produced with DALL·E 3?
Yes, the images generated using DALL·E-3, just like in DALL·E-2, are entirely your own. You can sell, print, and merchandise them without needing any additional permissions.
Limitations of DALL·E-3
There are some constraints applied to DALL·E-3 because of safety concerns.
DALL·E-3 has been engineered to prevent the generation of violent, adult, or hateful content. Therefore, you can’t generate adult or + 18 content using DALL-E-3.
Additionally, it includes measures to decline requests that seek images of public figures by name. These efforts contribute to risk assessment and mitigation in areas such as propaganda and misinformation.
Moreover, DALL·E-3 is programmed to reject requests for images in the style of living artists. You can also exclude your images from future training of image generation models. To do that, fill out “Artist and Creative Content Owner Opt Out” form.