Boats flying in space, pigs fighting with swords, pop culture characters surfing… Do you know How to create images from text with artificial intelligence?
In recent months, the art world has been shocked by the advancement of programs that allow images to be created through artificial intelligence. In fact, many artists around the world began to question whether this type of image can be considered art or not. Whether you agree with the argument or not, it’s undeniable that technology is impressive, and I’m sure you also want to learn how to use it, don’t you?
But after all, how can artificial intelligence create the most unique works of art using just a text command?
How do AIs that create images from text work?
Table of Contents
How to create images from text with artificial intelligence
All available tools for creating images through text use Machine Learning to perform mathematical calculations without the need for human intervention. With this, AI can generate incredible images through any text written by the user, without requiring contact with external programs or any type of programming on the user’s part.
The program team DALL-E are pioneers in this revolutionary way of making art, and the explanation is surprisingly simple. The GPT-3 neural network from OpenAI was originally used to create texts , scripts, translations and even posts, but it didn’t take long for engineers at OpenAI to discover that the network could also be used to create images based on commands of texts written by the user. In this way, some basic images of chairs, animals and signs were created by AI, starting a new era of art that would leave everyone on the internet in awe.
It is important to note that GPT-3 is not a genius that simply generates images out of thin air, but it was trained using a database with several images that already exist online to edit, morph, create, adjust and combine into a new image, no matter how crazy she is. Now, other companies (including Google) have also presented their image-to-text alternatives, but how can you be using them?
What are prompts and how to create them
Prompts are text commands that you send to artificial intelligence. As artificial intelligence needs instructions from humans to perform a task, you will find that, to create images, you will need to send a detailed description of what you want to receive. And this applies to practically all artificial intelligence: creating text, images, video, music and much more.
Creating a command prompt is simple: just imagine what you want to receive and write it as specifically as possible for the AI to understand. For example: if you want to receive a photo of a dog, remember to detail the color of its hair, its size and even the possible accessory that will be in the photo you want the artificial intelligence to create. The same goes for the setting.
- The Hidden Risk of Artificial intelligence and Big Data
- Data Science and Artificial intelligence- Clear Your Doubt
Remember that everything created by a generative AI is based on what already exists on the internet, but it is worth being very detailed to avoid errors. See this example from Bing Chat, which has Dall-E 3: See the image created by Microsoft artificial intelligence:
Top Image Creation Tools from Text
Several other tools with the same purpose have been created, and it is clear that each one has a different objective in creating these images. Check out the main tools for creating images from text:
Dall-E 3
Currently in its third generation, it is necessary to have a paid OpenAI account to use this artificial intelligence directly in ChatGPT, but you can also access the image creation tool from Bing or via Microsoft Designer, which have this integrated AI and only require logging into your Microsoft.
After that, just go to the text bar in the center of the page and type whatever your imagination allows. Remember to use specific terms to help the AI understand what you want, such as “High Definition, Anime, Realistic” among others.
Midjourney
The Midjourney is responsible for creating surreal images and is much more accurate than the Dall-E . To be able to use it, you will need to have an account and the Discord app installed.
After installing Discord and logging into your account, you must:
- Enter the official site;
- Click on the “Join the Beta” button;
- So, connect your Discord account with the website
- When connecting, you will have entered the website’s official channel within Discord. In the app, search for one of the channels in the tab “Newcomer rooms“;
- When entering a room, simply type the command “/Imagine” and continue with the text of your choice;
- Within a few seconds, Discord will display 4 images based on what the user wrote.
You will also have the opportunity to edit some image parameters to make it the way you want, just carry out some commands that appear on the screen.
Crayon
Craiyon is based on OpenAI technology, and It’s very simple to use. Unfortunately, the parameters of this tool do not allow it to provide such realistic images, but it is possible to generate some very unusual creations.
To generate images on Craiyon, just go to the official website and enter as many keywords as your imagination allows. The website design is very simple and straightforward, so don’t worry about getting lost in lots of information and technical jargon. As with the previous tools, try using terms to help the AI understand what you want to see.
Leonardo.AI
Created by developers JJ Fiasson, Peter Runham and Jachin Bhasme, this artificial intelligence is recommended for those who need to create draft videos and even of games. It works in a similar way to the other artificial intelligence we mentioned in this text: just type a text command and wait for the photos to be created.
To be able to create images in this tool, you need:
- Access the Leonardo.AI website;
- If this is your first use, you will need to create an account: you can choose to register your Google, Microsoft, Apple account or make a traditional registration via email and password.
- Then simply create a username and password. The next screen will only require you to indicate your interest in using artificial intelligence.
- Finally, on the home screen, just type the commands to receive an image never seen on the internet.
The free plan offers 150 credits every 4 hours. For each image, 5 credits are required. If you need more, you can subscribe to one of the plans, which range from US$ 10 to US$ 48. The site can also be accessed via the app, but for now, it is only available to those with Apple devices.< /span>
StarryAI
StarryAI is a little different from what we’ve seen so far. The platform (which is also available on Android and iOS cell phones) allows that you create images through three different AI. Altair produces images similar to the “dreams” we have, that is, he creates abstract images. Orion produces more realistic images and Argo creates images with a more artistic feel.
To create an image, you will need:
- Create an account using your email Google or Apple;
- Select one of the three available AI;
- Type in the text box what you want to see represented by the AI;
- Select a style, ranging from “Pop Art” to “3D rendering”;
- Click “Create” and wait a few minutes.
At the end of creation, the application will send a notification saying that your new image is ready. Access or install the application:
Dream by WOMBO
Dream was created by Wombo and is very similar to StarryAI. Like the previous tool, Dream also offers versions for the web and cell phones Android and iOS , with the main difference being that images are generated in a few seconds. It also works on Discord, through a server.
To generate your artwork, simply click on the “Create” button and type in what your mind wants. No account is required for its use, but you can publish the image through your social networks. Access the website or install the application on your cell phone:
Adobe Firefly
One of the great innovations of 2023, Adobe Firefly can be accessed for free and also creates images completely and free of charge. Up to two photos can be created for free per day and, whoever uses the free plan, receives a photo with a watermark, but you can crop it in an editing application. And using Adobe Firefly is very simple:
- Go to the Adobe Firefly website;
- Log in or create an account through your Gmail, Facebook < a i=4>or Apple ID. You can also use the same account that subscribes to Adobe or Adobe Express services;
- Then, just type the text commands (preferably in English) and wait for the images to be created.
Stable Diffusion
Appearing on the internet in mid-August 2022, Stable Diffusion stands out for being completely free and, in addition to its online version, can also be installed on computers with a CPU with at least 8 GB of VRAM. If you prefer to do everything on your cell phone, the application Dreamer uses the image creation base of this artificial intelligence.
It is not necessary to pay anything to use Stabble Difusion and its use is very simple, in addition to not requiring any registration. The online platform is also very intuitive and can be used for free. Here’s how:
- Access Stable Diffusion;
- Click Free AI Image Generator;
- Type the text command (preferably in English) about the photo you want to receive and click Generate.
- Now, just wait for the photo to be created and the time may vary depending on the queue on the website. To save the photo, simply open it in a new tab and download it. See this photo created using the command: “a photo of a man running in the rain, wearing a suit and carrying a bag“:
Go to the website or install the application that uses Stable Diffusion:
Microsoft Designer
Launched in 2023 as a complete image editing option reminiscent of Canva, Microsoft Designeralso has a tool that can create images using text commands. Here, DALL-E 3 has been integrated into the service and there is no additional payment required. The step-by-step guide to use is also simple:
- Go to the Microsoft Designer website or open the app on your smartphone;
- Click Get Started for Free and sign in with your Microsoft account ;
- Access the option Image Generator;
- Type the text command (preferably in English, but Microsoft Designer also understands Portuguese) and click Generate.
- To download the photo, simply click on one of the results and then press the purple button written Download.
See ways to access Microsoft Designer:
- Site Microsoft Designer;
- Microsoft Designer App for Android;
- Abeta access for Microsoft Designer for iOS.
Bing Chat (Microsoft Copilot)
Finally, the tool that resembles ChatGPT can also generate images using a text command. The good news is that it has DALL-E 3, without the need for additional payment and everything takes less than 1 minute to do. The step by step consists of:
- Access the Bing website;
- Sign in with your Microsoft; account
- Now, just send the text command in Portuguese and wait for the photos to be created;
- When you click on the photo, you will be directed to a new page and just click on Download for it to be saved on your computer.
Bing Chat can be accessed by:
These were the main ways of creating digital art using texts that are available today! So, which one are you going to use? Tell us in the comments!
