With DALL·E 3 now available directly in ChatGPT for Plus subscribers, it's become a major competitor to Midjourney for it's ease of use and image generation capabilities.
One of the key differentiators of DALL·E 3 that makes it so easy is that, given it's native integration in ChatGPT, it actually converts your initial prompt into a much more detailed prompt for the final output.
For example, if you enter a simple prompt like "create an image of a futuristic AI robot", this prompt get's augmented by ChatGPT's language model into the following:
You can see the full prompt that was used to create the image by clicking the information icon in the top right:
While DALL·E 3 is undoubtedly the easiest AI image generator to use when it comes to prompting, there are still a few key things to keep in mind in order to get the best possible results.
In this guide, we'll look at prompt structure in DALL·E 3, as well as 20+ prompts to try for the following categories:
- In-image text prompts
- Comic book
- Artistic syles reimagined
If you want to compare these DALL·E 3 results with Midjourney V6, check out our article on Midjourney prompts below:
Let's get started.
Prompt Structure in DALL·E 3
As mentioned, DALL·E 3 and ChatGPT do a lot of the heavy lifting when it comes to prompting, although you still want to provide as much structure & detail to your prompts as necessary.
While this certainly doesn't need to be followed strictly, here's an overview of how to think about writing prompts for DALL·E 3:
- Start with a Core Idea: Start with a clear, central theme. This could be anything from a serene landscape to an abstract concept or a narrative
- Expand with Details: Add layers to your idea. Describe the subject, setting, mood, and style. Whether it's a bustling city, a tranquil forest, or a fantasy realm, your words set the stage.
- Vary Linguistic Style: Switch between brief, direct phrases for simplicity and longer, descriptive sentences for depth. This flexibility lets you guide DALL·E 3 from minimalistic concepts to intricate scenes.
- Add Artistic Elements: Define aspects like lighting, perspective, and artistic style. Whether you desire a sunset's warm glow or a bird's-eye view of a city, these details enrich the final image.
- Iterate and Refine: Review the generated images, identify areas of improvement, and adjust your prompts accordingly.
That's just a high-level overview, now let's start looking at more specific prompts.
Generating In-Image Text with DALL·E 3
Although OpenAI actually states in their docs that it's not trained to produce text, it does still work sometimes:
DALL·E is not currently designed to produce text, but to generate realistic and artistic images based on your keywords or phrases. Right now, it does not have a specific understanding of writing, labels or any other common text and often produces distorted or unintelligible results.
That said, DALL·E 3 does still work at generating text, you just need to keep it to one or two words. Also, if it fails on the first try give it a few more attempts or prompt variations before moving on.
Here are a few examples of in-image text with DALL·E 3:
Generate a blue sky with white clouds and the word "Hello" in skywriting.
Depict a clear night sky filled with stars, with "Wish" subtly formed by the arrangement of the stars.
Generate an image of a computer chip, with the phrase 'Hello World' integrated into the circuitry design, symbolizing the intersection of technology and programming.
"Quantum Leap" displayed on a digital screen surrounded by quantum computing visualizations
Not bad, but as mentioned, these take a few tries to get right.
Next, let's move onto prompts that incorporate various photographic styles.
High-resolution photo of a city skyline reflected in a serene river, capturing the interplay of urban architecture and natural elements.
Black and white photo depicting the dynamic energy and candid moments of urban street life.
Macro Dew Drops
Close-up photograph of morning dew drops on a vibrant green leaf, highlighting the intricate patterns and reflections.
Long Night Sky
Long exposure image capturing the mesmerizing trails of stars above a tranquil mountain landscape.
Now, let's look at a few fashion-related prompts.
A runway show displaying futuristic fashion, blending high-tech elements with traditional haute couture.
A high-fashion concept showcasing avant-garde designs, with models in abstract, sculptural outfits, presented in a minimalist, modern art gallery setting.
A glamorous fashion setting reminiscent of 1950s Hollywood, with models in elegant gowns and classic tuxedos, against the backdrop of an old cinema.
A dynamic urban fashion scene highlighting contemporary streetwear trends, with models posing in a graffiti-filled alley, showcasing bold colors and edgy designs.
Here are a few examples of generating scenes & story panels for comic book.
Sequential comic panels depicting: 1) A spaceship captain navigating through an asteroid field. 2) The crew engaging in a laser battle with space pirates. 3) A dramatic escape to a distant, uncharted planet.
Comic story panels showing: 1) A lone survivor scavenging in a desolate cityscape. 2) A tense encounter with mutant creatures. 3) The discovery of a hidden utopian society amidst ruins.
In this two-panel fantasy comic: 1) A young mage in an ancient, rune-carved chamber, unleashing a powerful spell. 2) The final stand on a crumbling castle tower against a dark sorcerer, with a magical storm swirling overhead.
Four-panel spy comic: 1) A tense briefing in a dimly lit underground headquarters. 2) Disguised infiltration at a high-society gala. 3) A high-tech gadgetry encounter in a sleek, modern laboratory. 4) An adrenaline-fueled escape on a motorbike through narrow city streets at night.
Next, here are a few examples of generating landscapes & character concepts for gaming:
Create a vibrant, side-scrolling landscape reminiscent of Super Mario, with whimsical mushroom-like structures and a colorful, obstacle-filled course.
Fantasy Forest Landscape
Design an image reminiscent of a fantasy RPG, featuring an enchanted forest landscape, a medieval village, and characters equipped for an epic quest.
Stealth Operative Character
Design a stealthy operative character for a modern action game, equipped with high-tech gadgets and a sleek, tactical outfit.
Create an image in the style of a 1980s platform game, featuring a side-scrolling landscape with pixelated art, vibrant colors, and classic elements like ladders, coins, and quirky enemies.
Artistic Styles Re-Imagined
Finally, let's look at at a few artistic styles re-imagined by DALLE:
Create an image in the style of Van Gogh's 'Starry Night,' reimagined with a futuristic city skyline, where the swirling sky blends with neon lights and holographic elements.
Climate Change Scream
Reinterpret Munch's 'The Scream' to reflect the theme of climate change, with the figure standing before a scene of environmental decay.
Creation of AI
Visualize Michelangelo's 'Creation of Adam' with a robotic hand reaching out to a human hand, symbolizing AI creation.
Cyberpunk Mona Lisa
Reimagine Da Vinci's 'Mona Lisa' in a cyberpunk setting, with Mona Lisa as a humanoid robot against a backdrop of a futuristic city.
Summary: DALL·E 3 Prompts
As discussed, DALL·E 3 is one of the easiest AI image generators to use when it comes to prompting given that ChatGPT uses it's language model to improve your prompts automatically.
That said, in my opinion Midjourney V6 is still quite a bit more impressive in terms of image quality, although DALL·E 3 is a close second in the AI art race.
With the speed of improvement from both Midjourney and OpenAI, however, that may very well change soon.