By Sol Gonzalez, on 28 November 2022
The phrase, “a picture is worth a thousand words," is especially true in marketing.
Marketers are constantly creating images for all kinds of campaigns, so an AI capable of helping with this would be great news. But, are today's AI generated images up to the task?
To find out, we at Cyberclick have tested or analyzed 10 AI image generation tools, and here are the results. We entered the same keywords in all of them: "a sloth dancing in a field of sunflowers, realistic, 4K." Read on to see the results!
What Are AI Generated Images
It's been years since we discovered that artificial intelligence can create photo-quality images of human faces that don't really exist. But now it has gone a step further and can generate artistic images automatically from scratch.
There are multiple solutions on the market for AI generated images, and the operation is very simple: you just have to enter a series of keywords and the program will generate one or more images that will meet your requirements.
Best Tools for Generating Images with AI
Dreamstudio
Dreamstudio uses an open source image generation model called Stable Diffusion, which was created by the Stability IA developer community. One of their top priorities is to create AI tools that are ethical, accessible, and truly useful.
Dreamstudio is open source and free of charge. In our little experiment, it generated the most satisfying results! The sloth was clearly visible and looked quite friendly.
Dalle-2
Dalle-2 is one of the first AI image generation tools to go viral. It uses GPT-3 technology that is capable of interpreting descriptive text and generating images through machine learning. It also allows you to edit existing images and generate variations of them.
When you create your Dalle-2 account, you get 50 free credits for the first month. The following months, the free credits are reduced to 15. If you want to continue using the tool after consuming them, you can buy additional credits.
At Cyberclick, we liked the quality of the results, although it is true that the generated images did not meet the requirement of being "realistic".
Craiyon
Craiyon is the completely free version of the Dalle-2 AI, which you can use without consuming credits. After entering as descriptive a text as possible (always in English), the AI generates nine different images based on your keywords.
In our experiment, the images generated were somewhat peculiar, as none of the sloths were anatomically correct.
Google Image
This Google AI uses linguistic models to understand the text you entered and to design images accordingly. The images shown as an example are of high quality and have a good level of detail, so it promises to surpass other systems such as Dalle-2.
However, for the moment this is an internal Google project and is not available to the public, as they have stated that it can lead to the creation of images containing "stereotypes and harmful representations". We hope they manage to overcome these obstacles and make this tool available for the public to enjoy.
NeuroGen
This AI image generation tool offers 100 free images. After that, the price is $10 per 1,000 images generated.
The images it generates are private in principle, but can be made public, deleted, or saved in PNG format.
Photosonic AI
This tool is available without registration and free of charge. When we entered our keywords, it offered us a lot of variations. For each one, we could hover the mouse over it to see the prompts that led to it, so this can also serve as a database of AI-generated images.
After testing different image generating AIs, this is one of our favorites, as it is able to generate varied and high quality images in a matter of seconds.
Nightcafe
This AI was created in 2019 and takes its name from Van Gogh's painting, "The Night Cafe". It has now reached more than 5 million users and you can create up to 5 free images.
One of its most interesting functionalities, compared to other tools, is that it allows you to play with different AI generated images or image generation modes within the same app. Also, as an additional service, you can print the images you have generated.
When we tested this tool in our experiment, the results were quite good!
Midjourney
Midjorney is an independent research lab composed of 11 members that seek to "break new ground and expand human creativity". It has a limit of free images (about 25) and, after that, you must select one of their rates.
As for how it works, you have to create a Discord account, enter a channel, and use a command to generate the first 4 images. Then you can create variations of these images or enlarge them.
It has a very particular style, which is rather artistic and even dreamlike. Mind you, our sloths were pretty scary.
Lexica
More than a tool to create AI generated images, Lexica is a database that allows you to search images and see what keywords they have been generated with.
It can be a very useful resource for marketing campaigns, as you can take ideas, copy them and modify them. It helps you better understand how keywords work in this context and improve your own.
Canva
This well-known tool has a beta text-to-image functionality that uses the same open source technology as Dreamstudio. To use it, all you have to do is describe the image you want. Also, as it is still in the testing phase, you are invited to leave your comments on the generated image in order to help Canva improve the tool.