Ultimate Guide to OpenAI DALL-E, with demo and examples

Harish GargHarish Garg
7 min read

DALL·E is a neural network(an Artificial Intelligence technology) that can generate images from text inputs in natural language.

DALL-E is created by the AI company OpenAI, the creators of GPT-3. In fact, DALL·E is also a version of GPT-3 that is custom trained on text-image pairs.

DALL-E 2 is version 2.0 of the AI model from OpenAI.

Table of Contents

Table of Contents

  • What does DALL·E do?
  • How to get access to DALL-E
  • Installing DALL-E
  • Using DALL-E
  • DALL-E alternatives
  • DALL-E cost
  • DALL-E Terms of usage
  • DALL-E API
  • DALL-E Product Ideas
  • DALL-E Resources

What does DALL·E do?

If you input a text phrase to DALL·E, it will generate an image that will be very similar to the concept described in the text phrase. These images are generated from scratch and do not previously exist before they are generated.

DALL·E can generate all kinds of images – anthropomorphized animals, combining multiple concepts, etc.

For example below image was created by DALL·E from the text phrase “A photo of an astronaut riding a horse”.

DALL·E
Image courtesy OpenAI

How to get access to DALL-E?

In order for you to start using DALL-E, you will need to apply for access. Follow these steps to get access to DALL-E:

  1. Open DALL-E 2 waitlist form in your browser
  2. Fill in the below information in the form:
    • First Name
    • Last Name
    • Email
    • Twitter profile
    • Instagram profile
    • Linkedin Profile
    • Choose your profession
      • Professional artist
      • Developer
      • Academic researcher
      • Journalist or online creator
      • Other
        Make sure to enter all information correctly. Your social media profiles are optional but entering it will help in getting access to DALL-E faster.
  3. Submit the form

This will add you to the waitlist for DALL-E. You will be notified by email when you have access to DALL-E.

Filling out the form will put you on a waitlist. Currently, there might be a limited number of spots available. You might be up for a little bit(maybe longer) oF wait before you can get your hands on DALL·E.

Do I need to install anything to use DALL·E?

No, you need not install anything to use DALL-E. The current version of DALL·E is available as a web application that can run from your browser. Once you have access, in order to start using DALL-E, follow these steps

  1. Open DALL-E Playground in your browser
  2. Sign in with your account where you have access to DALL-E
  3. Write a prompt for the image you want to generate and click on the “Generate” button
  4. Six variants of image for your prompt will be generated and displayed in the browser

Using DALL-E

These are a few different features of DALL-E.

  • Generate Images from a text prompt

  • Generate variations of an image

  • Edit and generate variations of the edited image

  • Generate variations of your own uploaded image

Writing great DALL-E prompts

Be as imaginative and creative as you can be. Write prompts that are absurd or implausible.

When writing a prompt, it’s better to have a concrete foundation, like a detailed prompt filled with detailed descriptions about certain things, the place they’re in, what they’re doing, how they look like, what they are,, etc.

The more detailed the description, the closer and better the image will be to the one you have in your mind. But, you don’t always have to be that detailed.

You can be vaguely vague and still get great images. For example, I tried a prompt like, “A tiger wearing a hoodie, digital art”, but I didn’t specify the age of the tiger (whether is was a cub or not), the color of the hoodie, where it was standing, etc. But I still got a great image like this-

In fact, even if you don’t have a specific, detailed image prompt of something, you can just use a vague description like “A tiger wearing a hoodie”, and get a great image.

So basically, if you have a specific, detailed way of how you want your image to look, you can use a detailed description.

But, if you don’t you can kinda just have fun with it and use a slightly vague description.

To get some interesting results, mix concepts that won’t make sense in the real world.

I don’t have DALL·E, what do I do?

Well, you and millions of others are waiting for their invite from OpenAI to start using DALL·E. There doesn’t seem to be a way right not to skip the queue. However, you can check out some DALL-E alternatives.

Here are a bunch of images I generated using the DALL-E mini

Is DALL·E free?

Every user gets 15 generations free per month. Unused credits do not roll over to next month.

Beyond 15 generations, the user needs to purchase credits in the increment of $15. $15 gives one 115 generations. Unused credits remain active for one year.

Also, people who have DALL-E access can generate up to 50 images per day. If you go above the limit, you will be blocked from using DALL-E for a day.

Things that can get you blocked from using DALL·E

There are certain kinds of images that can get you blocked from using DALL·E, if you try to generate them. Namely, if your DALL-E prompt has a word related to

  • Violence
  • Sexual content
  • Politics
  • Real people

You will get a warning couple of times before they suspend your account.

Who owns DALL-E-generated images?

The user who generates the image owns the image. Generated images can be used for any personal or commercial use.

Using DALL·E API to build your own image generator application

Currently, there is no official API from OpenAI for DALL·E.

There might be an API endpoint access in future versions for developers to use in building applications but right now there has not been any hints from OpenAI when that would be.

However, it’s inevitable, given how GPT-3 was rolled out that at some point in future versions of GPT-3, there will be an API endpoint for developers to use.

Till then you can one of the alternatives to building your own image generator applications.

Product Ideas and use cases for DALL·E

Here I list some product ideas and use cases for the OpenAI DALL·E image generation AI model. Feel free to use these ideas and use cases in your own projects.

  • Blog and article cover image generation

Generate cover and header images from blog posts and articles using DALL·E by using the title as the text prompt.

  • Instagram post generation

Use DALL·E to generate Instagram posts.

  • Music Cover Art

Generate cover art for music using DALL·E.

  • Animation and video generation

Use DALL·E to generate clips and artwork for animations and videos.

  • NFTs

Generate NFTs using DALL·E.

  • Teach art history

Use art created by DALL·E to teach art history to students.

  • Logos

Generate logos using DALL·E.

  • Stock images

Generate stock images using DALL·E.

  • Product images

Generate product images using DALL·E.

  • Print on Demand (PoD)

Generate print-on-demand images using DALL·E

  • Product packaging

Generate product packaging images using DALL·E.

  • Illustrations for books

Generate illustrations for books using DALL·E.

  • Posters

Generate posters using DALL·E.

DALL·E Resources

While you are waiting for your DALL·E invite, you can check out these resources:

0
Subscribe to my newsletter

Read articles from Harish Garg directly inside your inbox. Subscribe to the newsletter, and don't miss out.

#dall-e

Written by

Harish Garg
Harish Garg