How to create an image with AI: 5 top AI generative apps

What is an AI image generator?

Generative AI is a type of artificial intelligence that involves training MLL (machine learning models) to generate new, original content based on a delivered prompt. A prompt can be anything from text and images to music and video, and even new chemical compounds for use in drug development. In this way, generative AI has the potential to revolutionize a wide range of industries and applications.

So basically an AI image generator is a tool to generate images from prompts (different images, text or even video). And yes, you can even create videos with AI image generators, because video is just a sequence of images :) I personally use Deforum Stable Diffusion for generating amazing AI video content. There are different Stable Diffusion models - just choose the ones which best fits your idea / art style. And play with them - it’s not like one fits all.

But I want to create images from text prompts - which one should I use? Well, you have to choose for yourself, but I made a small review of all the best text to image generative AI tools for you. For every of them I use the same prompt to have a better comparison. So, let’s dive in!

PROMPT USED

During a sunset a cute ragdoll cat sitting on a keyboard in a meadow full of wildflowers, children book art style, 8k, high resolution

Leonardo.ai

Leonardo.ai is a tool that uses AI to create stunning game assets, such as items, environments, helmets, buildings, and concept art. It enables users to rapidly ideate, train their own AI models, and create unique production-ready assets with an artist-friendly interface.

JJ (founder) wrote that “ Our initial experimentation proved to be a somewhat frustrating and costly endeavour. The goal of reducing the friction between ideation and creation was hindered by a lack of control, consistency of output, and underdeveloped workflows.”

In the really user-friendly interface you can select one of the models (Stable Diffusion or specially trained by Leonardo) or upload your own. It even has an option to help you train your own model. Also amazing is that you can choose the output resolution, you can outpain in it and upscale! Right now every user has 250 tokens per day and can upscale 25 images per day. And what is the result?

Stable Diffusion

Stable Diffusion is a latent text-to-image diffusion model capable of generating stylized and photo-realistic images. It is pre-trained on a subset of the LAION-5B dataset and the model can be run at home on a consumer grade graphics card so everyone can create stunning art within seconds. It’s free and you can use it online or on your computer. It has many amazing features, like controlnet which allows you to have absolute control over the output. As it is completely created by the community, you can find almost a tool for everything you need. Oh, and the output with model 2.1!

Dall-e 2

DALL·E is a 12-billion parameter version of GPT-3 trained to generate images from text descriptions, using a dataset of text–image pairs. We’ve found that it has a diverse set of capabilities, including creating anthropomorphized versions of animals and objects, combining unrelated concepts in plausible ways, rendering text, and applying transformations to existing images.

It has a user-friendly interface, built-in outpainting option as well as editing option.

Mid Journey v4

Quoting Mid Journey founder David Holtz from his Discord announcement from November 2022: “Mid Journey V4 is an all-new code base and all-new AI architecture, this is our first model trained on the new Midjourney AI supercluster and has been running for over 9 months.”

We all have seen amazing AI artworks created with it and you still can use it, despite Mid Journey v5 being live. To use it you either use it publicly on discord (limited free access or with one of the paid plans) or have access to stealth mode through PRO paid option.

Mid Journey v5

Mid Journey v5 was released on 15th of March 2023. It is currently available only to paid subscribers. Trained model is a huge improvement compared to the v4 one. It is better with hands and tooth generating. It also already generates upscaled pictures. Also you’d better communicate with it as a whole sentence rather than just random words, as it performs better with such prompts. There are many more features which I will cover in the future, but let’s see how it dealt with our task:

Should I use generative AI art?

I mean AI tools are designed to improve your work, save you time or make the world a better place, so why won’t you benefit from it? You can use text to image generators to visualize your concept, brainstorm your idea or make a first draft of an idea. You can use it to create really outstanding visual works. Of course, people from the creative industry have an advantage but people without a creative / artistic background can benefit from it. As it helps us to save time looking for a perfect TV show, why wouldn't we use it to create a better image to send to our loved ones and make their day better? In the end it’s a gesture that matters.

And also If you are a creative person and you’re facing a problem which you think AI can solve, but don’t have technical knowledge to execute it - don’t worry. There are people who will help you create an AI based app to solve it. And we encourage you to work with people from lablab.ai’s amazing community of builders, creators and innovators.

Just join one of our AI Hackathons, pitch your idea to other participants, form a team and create a working prototype of a tool you will use. And millions of people with the same problem you faced.

Innovate with AI. Shape reality with AI. Make a better place with AI.