So, you’ve heard of DALL-E 2, or seen some amazing Midjourney art out there and though: I can do that. The good news is….yeah you can!
These AI models that can whip up images based on text descriptions, so the barrier for entry is low to start creating all kinds of images.
In this very basic guide, we’ll give you some basics to get started, and some examples to help inspire the kind of prompts that can give you some great results.
THE BASICS: CREATE AN ACCOUNT AND GET STARTED
First, it’s important to note that while they do similar things, DALL-E 2 and Midjourney are different programs, offered by different companies, and thus have different sign up methods and run differently. While the processes are different, the fundamentals in using them are similar.
Let’s get to the basics.
DALL-E 2
1 – First, head over to OpenAI’s DALL-E 2 page by clicking here.
2 – Click on “Sign Up” and register for free with your email, or connect your Google or Microsoft account.
3 – You’ll be redirected to the DALL-E 2 prompt page.
4 – From the prompt field, you can now start typing phrases and generating images, or upload an image to enhance and edit it with the tool. We’ll get more into prompts below.
MIDJOURNEY
1 – First, you’ll need a Discord account. Midjourney runs entirely on Discord, a chat platform (more on that here). So head over here and register for your account, or log in if you have one already.
2 – Next, head over to Midjourney.com and select “Join The Beta.” This will provide you an invite to the Midjourney Discord server.
3 – Accept the invite and you’ll be directed right into Discord. You’ll want to click the small sailboat icon – that’s the Midjourney server.
NEXT STEPS: CREATING IMAGES WITH PROMPTS
Both of these programs rely on prompts. Prompts are the written phrases you’ll enter to generate images.
For DALL-E 2, this will be straightforward. You’ll just enter the prompt into the prompt field and hit “Generate”.
It’s a bit more involved with Midjourney. Once you’re in Discord, you’ll need to find the Newcomer rooms section, and select one of the “newbie” rooms. There will generally be several available, named “newbies-116” or “newbies-59”. You can select any of these to begin. One thing to note: you only have a limited number of free image prompts with your trial (around 25) and after that, you’ll need to buy a full membership to continue creating. To prompt an image, type “/imagine” in your “newbie” Discord chat and enter your prompt.
SOME NOTES ON PROPER PROMPTS
Ok, now let’s talk about prompts. What makes a good one? Here are some tips to get started on the right foot:
Be detailed: The more info you give, the better the pic will turn out. Avoid vague words and give as much detail as you can. For example, instead of just saying “a cat,” say “a close-up of a blue-eyed Siamese cat with a fluffy coat.”
Set the scene: with your prompt, provide some context to help it understand what you want. For example, if you want a city pic, include info about the time of day, weather, and mood.
Use examples: Just like with text generation, examples can help these programs get a sense of what you want. Show it similar pics or describe what you want in text.
Get creative: Don’t hold back! DALL-E and Midjourney can do all sorts of wild stuff, so go ahead and get creative with your prompts. For example, try generating an image of a dragon or a surreal landscape. Descriptions aren’t limited to things – they can be action, lighting, an artistic style, camera angle. Truuly – the more creative you can get the better.
Here are a few examples of text descriptions that could work well as prompts:
- A futuristic, Blade Runner-type city at night with neon lights and flying cars over the skyscrapers. The mood is grim.
- A green, blue and gold lush jungle landscape with a hidden temple and sunlight filtering through the trees. This looks like a vintage photo from National Geographic magazine.
- A portrait of a person with a mohawk hairstyle and a leather jacket/ripped jeans outfit, Annie Liebowitz style portrait with studio lighting.
- A surreal pic of a giant robot on a mountain with a village at its feet
- A group of people playing soccer on a beach with clear water and a sunset
Remember, the more specific you are, the closer you’ll get to what you want. Keep in mind that the model is still developing, so the results might change. The images generated are based on the model’s understanding of the text description you give it.
Pingback: Generative AI delivers mouthwatering food photos for menus, UberEats, DoorDash – THE DEPT OF NEXT
Pingback: OpenAI Previews DALL-E 3 with Built-In ChatGPT Integration For Easy Image Prompts – THE DEPT OF NEXT