On November 19, 2020, I published the article “What can AI do today?“
Since then, I have been looking for a free AI model for creating art from text for a long time.
First, I used Craiyon, which used to be called DALLE mini. It is an AI model that can make pictures based on any text.
The main problem with Craiyon is that this model generates mostly blurry images, and clear people’s faces can only be created in portrait mode.
So I was looking for a quality AI-Generated Art alternative to Craiyon and found it.
I learned about DALLE-2, an artificial intelligence system that creates realistic images and photographs from natural language descriptions. Unfortunately, I have never managed to get access to DALLE-2.
What is the best free AI art generator?
In August 2022, a group of enthusiasts, including Patrick Esser from Runway and Robin Rombach from the Machine Vision & Learning research group at LMU Munich created Stable Diffusion.
With the help of the text-to-image model known as Stable Diffusion, billions of users will be able to create beautiful AI-Generated Art in a few seconds. It is a breakthrough in terms of speed and quality.
The developers of Stable Diffusion made it possible for everyone to use their model in the public domain.
For more control and rapid generation you can try our DreamStudio beta here: http://beta.dreamstudio.ai.
AI-Generated Art: From Text to Images
Let’s check how Stable Diffusion is implemented on the Dreamstudio AI website.
Please note that you can choose from several methods for creating AI-generated art. I prefer the k_euler method, but you can choose another from the list.
It is important to choose the right Cfg Scale parameter. I usually set the value in the range from 10 to 13.
Cfg Scale adjusts the degree to which the image will match the prompt. Higher numbers increase the proximity of the image to the prompt.
The first text prompt is “The constellation Andromeda, the stars in it are like shining diamonds and in the foreground the face of a beautiful white girl, attention to detail, hyper-realistic, 4K”.
What does it look like? It certainly is beautiful. To achieve this quality is quite difficult. It will take several attempts to generate the image.
Let’s create an image of a beautiful girl in the jungle.
Please note that the words “girl”, “naked” are prohibited for prompt text. Instead, you can use the words “woman” and “bikini.” When the algorithm generates too explicit images, it applies almost 100% blur to them.
Sometimes the model creates a beautiful story but distorts the face or adds extra arms or legs. Because of this, the number of attempts to create quality AI-generated art content increases.
In the last image, you see a beautiful plot and only minor and subtle flaws are a little upset. Perhaps I need to take a closer look at how the model Stable Diffusion works. If you are creating a portrait, add to the prompt “DSLR, 50mm lens f/2.8, shutter speed 1/60, soft light source, realistic, attention to detail, 4K“.
I really hope that this DreamStudio Prompt Builder will help you create real masterpieces.
You can also create murals. Here is an example of the settings.
Prompt: !dream of cats, dogs, women, men, birds, HD Mural by David Schnurr, Colorful, Staled glass, Attention to detail, 4K.
Sampler: plms, Cfg Scale: 9.
Another version of the mural created by this program with these settings.
I really hope you agree with me that Dreamstudio AI is the best free AI art generator.
Yes, I know it is super cool, but you will want to upscale your photo. Try Ojoy, it is awesome and completely free.