I Found the Perfect DALLE 3 Prompt

Introduction

I've been using DALL-E for quite a while. I started with DALL-E 2 back in April of 2022, and now inside of ChatGPT-4, we have access to DALL-E 3, which is a whole lot more useful. A couple of years ago, it was really hard to get anything good out of DALL-E. Now, you can get really useful images out of DALL-E inside of ChatGPT-4, but you do have to know how to prompt it the right way. Let me share with you a six-part prompt that's pretty simple, and you can follow this formula every time you want to generate an image inside DALL-E. Remember, you need access to ChatGPT's paid version to get access to DALL-E 3.

Be Descriptive

Firstly, you need to be very descriptive with DALL-E. The details really define how the image comes together. It picks up on those details in the text. So instead of saying "draw me a picture of a cat," it's much better to really flesh out your prompt, like "a fluffy cat with green eyes sitting on a windowsill at sunset." I'll discuss some extra pieces to add to DALL-E every time you give it a prompt. Being descriptive is the first thing to keep in mind.

Mention the Image Type

Next, you want to mention the image type. You can specifically say "create a photo, illustration, oil painting, sketch, or drawing." This will guide DALL-E to generate the specific type of image you want. For instance, if you say "watercolor painting of a serene lake view at dawn," you're going to get exactly that: a watercolor painting, not an illustration or sketch.

Set the Scene

Setting the scene has two parts: defining your character and describing the specific setting that character is in. For example, a "medieval knight standing guard outside of a stone castle covered in ivy." Here, "medieval knight" is your character, and "outside of a stone castle covered in ivy" is your setting.

Emphasize Key Features

If a certain part is critical, you should define it as a key feature. For instance, if you're making a logo, you might say "a logo with a bold golden eagle clutching a silver compass; the eagle is the primary focus." This way, DALL-E knows to emphasize the eagle in the logo.

Use Analogies and Comparisons

Sometimes referring to something similar that DALL-E or ChatGPT might know can help you get better results. For example, "a landscape reminiscent of Van Gogh's swirling style showcasing a bustling city square."

Specify Colors

If you have specific colors in mind, mention them in the prompt. For example, "dominated by shades of turquoise, gold, and coral pink" for a beach scene.

Experiment and Revise

Since this is all done inside ChatGPT, you can experiment and revise prompts. You don't need a perfect prompt initially because you can follow up with tweaks and revisions based on the initial output. This is really useful and sets it apart from tools like MidJourney, where you need to get the prompt right initially.

Edit Images

Recently, DALL-E has given users the ability to edit images right inside its interface. You can select part of an image, type in a new prompt, and DALL-E will add or subtract from that existing image.

The Six-Part Formula

Here’s the actual formula, which consists of six parts:

1. Image Type: Specify what type of image you want (photo, illustration, drawing, etc.)
2. Define the Subject: State your main subject (e.g., a fluffy cat).
3. Key Features: Highlight key features (e.g., green eyes).
4. Define the Setting: Mention the background and the environment (e.g., windowsill at sunset).
5. Color and Mood: Combine these two elements, if possible (e.g., sunset with warm hues).
6. Style: Optionally describe a specific artist style (e.g., reminiscent of Van Gogh).
7. Primary Focus: Emphasize what should be emphasized (e.g., the cat is the primary focus).

By following this formula, you’ll get better results that are closer to what you’re looking for. You can experiment and use this method in every prompt you craft.

Example Prompts

• Illustration of a Fluffy Cat:

  • Type: Illustration
  • Subject: Fluffy cat
  • Key Features: Green eyes
  • Setting: Windowsill at sunset with warm hues
  • Style: Reminiscent of Van Gogh
  • Primary Focus: The cat

• Photograph of a Sleek Modern Train:

  • Type: Photograph
  • Subject: Sleek modern train
  • Key Features: None specified
  • Setting: Snowy landscape at sunset with cool blue and white dominating the scene
  • Color and Mood: Capturing the essence of speed and technology
  • Primary Focus: The train

• Oil Painting of a Quiet Ancient Library:

  • Type: Oil painting
  • Subject: Quiet ancient library
  • Key Features: Single glowing lamp
  • Setting: Filled with towering books
  • Color and Mood: Warm, rich hues to evoke history and mystery
  • Primary Focus: The intricate details of the book and the shelves

Additionally, you can explore the DALL-E GPT by visiting the Explore GPTs tab in ChatGPT. This custom DALL-E GPT provides various styles and aspect ratios to help you tailor your prompts even more precisely.

If you want to dive further into DALL-E 3, check out the DALL-E 3 Mastery Course at Skill Leap, along with other creative and entrepreneurial tools.

Keywords

1. Descriptive
2. Image Type
3. Set the Scene
4. Key Features
5. Analogies and Comparisons
6. Specify Colors
7. Experiment and Revise
8. Edit Images
9. Six-Part Formula

FAQ

Q: Do I need a paid version of ChatGPT to access DALL-E 3? A: Yes, you need access to the paid version of ChatGPT to use DALL-E 3.

Q: Can I specify the style of the image in my prompt? A: Yes, you can specify the style, such as an oil painting, watercolor, sketch, etc.

Q: What should I do if my initial prompt doesn't yield the desired result? A: You can experiment with and revise your prompt within ChatGPT until you get closer to your desired result.

Q: Can I edit images directly within DALL-E? A: Yes, you can select parts of an image and type in a new prompt to add or subtract elements from an existing image.

Q: Is there a way to see different image types and styles before creating a prompt? A: Yes, you can use the custom DALL-E GPT in the Explore GPTs tab to see various styles and aspect ratios.