I Found the Perfect DALLE 3 Prompt
Check out our complete Dalle 3 course, along with 20 other AI courses: bit.ly/skillleap
After months of using DALLE 3, I found a simple 6-part prompt that gets me much better results.
Be Descriptive: DALL·E thrives on detailed descriptions. Instead of saying "a cat," you might say "a photograph of a fluffy orange cat with green eyes sitting on a windowsill at sunset."
Mention the Image Type: Specify if you want a photo, illustration, drawing, oil painting, sketch, etc. For instance, "a watercolor painting of a serene lakeside view at dawn."
Set the Scene: Provide context. If you want a character in a specific setting, mention both. For instance, "a medieval knight standing guard outside a stone castle covered in ivy."
Emphasize Key Features: If certain features are crucial, reiterate or emphasize them. “Create A logo with a bold, golden eagle clutching a silver compass. The eagle is the primary focus."
Use Analogies or Comparisons: Sometimes, referencing something familiar can help. "A landscape reminiscent of Van Gogh's swirling style, showcasing a bustling city square."
Color Descriptions: If you have a specific color scheme in mind, mention it. "A tranquil beach scene dominated by shades of turquoise, gold, and coral pink."
Experiment and revise: If the initial result isn't what you want, tweak your description. Sometimes a minor change in phrasing can yield a significantly different result.
Here is the formula: Type, Subject, Setting, Mood/Color, Style, Emphasis.
Type: Specify the image type (e.g., "photo," "illustration").
Subject: Describe the main subject (e.g., "fluffy orange cat").
Features: Highlight key features (e.g., "with green eyes").
Setting: Mention the background or environment (e.g., "on a windowsill").
Mood/Color: Convey the mood or color scheme (e.g., "at sunset with warm hues").
Style: If needed, describe a specific artistic style (e.g., "reminiscent of Van Gogh").
Emphasis: Reiterate any crucial element for emphasis.
Example using the formula:
"Illustration of a fluffy orange cat with green eyes on a windowsill at sunset with warm hues, reminiscent of Van Gogh. The cat is the primary focus."
Other examples:
Photograph of a sleek, modern train gliding through a snowy landscape at dusk, with cool blues and whites dominating the scene, capturing the essence of speed and technology. The train is the primary focus.
Digital artwork of a futuristic cityscape viewed from above at night, with neon lights and holographic billboards creating a vibrant palette, inspired by cyberpunk aesthetics. The city's architecture is the primary focus.
Oil painting of a quiet, ancient library filled with towering bookshelves and a single, glowing lamp, using warm, rich hues to evoke a sense of history and mystery. The intricate details of the books and shelves are the primary focus.
Watercolor scene of a hidden garden in full bloom after a spring rain, with droplets glistening on petals and leaves, rendered in a wash of delicate, translucent colors. The interplay of light and nature is the primary focus.
Charcoal sketch of a rugged coastline battered by waves under a stormy sky, capturing the dramatic contrast between the dark, churning sea and the stark cliffs. The raw power of nature is the primary focus.
Пікірлер: 18
Check out our complete Dalle 3 course, along with 20 other AI courses: bit.ly/skillleap
Something else I would like to share in terms of DALLE-3 making better images is the viewing angle of the image ex front view, rear view, birds eye view, worms eye view etc. I have found when generating images of a car, it can get the viewing angle wrong as an example. If you have a specific dominant subject in an environment that is a piece of advice that I can give! Have a good day :)
These prompt work well within Ideogram as well you might want to try that as well. Good video.
Thanks very much. I actually tried your Dall-E prompt samples in MidJourney and it works great 👍
Great video
Great formula sir 🎉
Given how descriptive you have to be with Dalle 3. I am bad at such descriptions. A method I would use is to go to the GPT store and go to a store that gives you automated MidJourney prompts off of rough descriptions of what you want. Given that the prompts they give are extremely descriptive, you can use the prompts they give you and enter those into Dalle 3. And normally you get 4-5 prompts from your rough description
@SkillLeapAI
Ай бұрын
that's a really good idea
Very good video. I wonder if you want your file to look like an particular artist can you provide a URL or a file? I suppose you can name the painting instead.
I can't believe I didn't know about the Dalle Gpt before!
Hi :) Thank you for the video: i took the insights from it and created the following prompt: i believe you will like it and will be glad for the feedback :) PROMPT → !PromptElaboration: To create a highly detailed and effective prompt for DALL-E 3, please provide the following information: **# Image Description** **## Type** Specify the desired image type (e.g., photograph, illustration, digital artwork, oil painting, watercolor, charcoal sketch). **## Subject** Describe the main subject of the image in detail, including any key features or characteristics that should be emphasized. **## Setting** Provide the background or environment in which the subject is placed. Be specific and descriptive to help set the scene. **## Mood/Color** Convey the desired mood or atmosphere of the image, and mention any specific color schemes or palettes that should be used to evoke that mood. **## Style** If you have a particular artistic style in mind, describe it here. You can reference specific artists, art movements, or aesthetics to help guide the image generation process. **## Emphasis** Reiterate the most crucial elements of the image that should be the primary focus. This will help ensure that the generated image aligns with your vision. **# Prompt Elaboration** Based on the provided image description, I will generate an elaborate and detailed prompt for DALL-E 3 using the following formula: [Type] of [Subject] with [Features] in [Setting], [Mood/Color], [Style]. [Emphasis]. The generated prompt will be designed to provide DALL-E 3 with a clear and comprehensive understanding of the desired image, incorporating all the key elements you specified. **# Prompt Refinement** After generating the initial prompt, I will review it and consider the following aspects for further refinement: - Clarity: Ensure that the prompt is clear, concise, and easy to understand. - Specificity: Verify that the prompt provides enough detail to guide DALL-E 3 towards generating the desired image. - Coherence: Check that all elements of the prompt work together harmoniously to create a cohesive and compelling image description. - Emphasis: Confirm that the primary focus of the image is adequately emphasized in the prompt. If any areas for improvement are identified, I will revise the prompt accordingly and present the refined version for your review. **# Prompt Variations** To provide you with additional options and creative possibilities, I will generate three alternative prompts based on your original image description. These variations will explore different combinations of elements, styles, or emphases while still adhering to the core aspects of your vision. **# Final Review** Please review the generated prompts and provide your feedback: - If you are satisfied with one of the prompts, indicate which one you prefer, and I will consider it the final version. - If you have any suggestions for further refinement or would like to explore additional variations, please let me know, and I will iterate on the prompts accordingly. Once you have approved the final prompt, you can use it to generate images with DALL-E 3, confident that it incorporates all the necessary elements to achieve your desired results. ``` And here is an example of use: chat.openai.com/share/1d15278b-0758-406b-9af5-5404db58feab
Thank you
i keep tring but this is what i keep getting I can't directly generate images, but I can help you describe what you'd like
I got a lot of good stuff out of Dall E 2. It was difficult though. What everyone came here for: 4:00 prompt formula
Instead of Square, Horizontal or vertical you can finetune this very simple by using the actual ratios: [Square] 1:1 [Landscape] 3:2, 5:4, 16:9, 16:10 [Portrait] 2:3, 4:5, [Ultrawide] 21:9 Just use the Numbers not the words and it'll work.
Do you have suggestions on how to make DALLE3 images more realistic and less AI-surreal, especially with people and settings? DALLE3 really sucks now compared to others. People look like airbrushed wax models.
@SkillLeapAI
Ай бұрын
I never use Dalle 3 for images with people. It just can't do a good job yet. I use Midjourney for people.