AI Art With Leonardo (4 Students Max CAMP)

AI will revolutionize every aspect of human knowledge, including the arts .  Powerful AI Art generation models are capable of taking text and making images that match the text description.  Whether you are already an artist or someone who can't communicate visually at all, AI Art generation has a lot to offer you.

This field is developing very quickly.  I have offered classes in AI Art before and have learned what works and what doesn't.   This is the latest version and it supersedes my other classes on the topic.  The four 50 minute sessions will be structured as follows.

About the class:
     1.  Introduction to AI txt2img models:   These are the models that generate art and we will look at a few of them.  I will demonstrate working with Midjourney, Lexica, DALLE-3 and Dreamstudio, before likely settling on Leonardo.  The student has the choice as to which AI generator to use.  I recommend Leonardo as the features are excellent and students receive free credits daily and indefinitely.  Lexica and Dreamstudio give a free trial so we will take advantage of  those.  However, after that the cost is around  $8-10/month.  DALLE-3 is availabe through Bing Chat but the processing has been slow.    Midjourney creates beautiful images but  does not offer a free trial.  I will be taking students' text prompt ideas and demonstrating them on Midjourney on my end.   
     2. After choosing an AI model, Leonardo most likely, we will dive into prompt generation.  Students will receive a spreadsheet with art terms related to art movements, artists' names, color schemes, camera angles, lighting conditions, all generated using ChatGPT.  We will create some ChatGPT files for art subjects that interest the student, e.g. exotic flowers, detailed architecture, etc.  Finally we will  use g-prompter, a model that uses ChatGPT to 'improve' our prompts.  We will be learning all of this while making beautiful images and building up a gallery.  We will use the concept of a seed in order to create similar images using different terms in the prompt.
     3. Now it is time to dive into the image generation pipeline.  Writing a prompt and creating beautiful images is just the first part.  Now we need to use some advanced tools to make the image truly stellar.  img2img is such a tool.  It allows us to create an image based on text AND another image.  How this image influences the output is not set in stone so it will require experimentation.  We will then start learning about ControlNet, a tool which gives us great control over how one image is used to create another.
     4.  After we finish using ControlNet, the remaining class will devoted to the practice of inpainting and outpainting.  In/Out painting are very similar concepts.  They basically refer to using AI to change only part of an image.  In the case of outpainting, that would be the area outside the image.  For example, imagine a photograph of a car on a street.  Now place that photo inside a larger black rectangle.  Outpainting asks the AI to fill out the black area with, say, more street, more  buildings, etc. so that the new additions still work with the old image.  When doing inpainting, we will be taking a brush and selecting different areas of the canvas to be changed by AI generation.   

About my teaching style:
I am a strong believer is the Socratic method and I use it in all my classes, but especially in one-on-one classes.  I know of no other way to probe a student's understanding other than asking questions or observing their work.  I like doing both.  

I believe building a strong foundation is essential to any human endeavor.  Explaining things from first principles is essential and I always encourage students to ask 'why?'  Though this class is not as technical as my programming classes, it is a great an opportunity to explore the basics of AI models and image processing.  We will be thinking like scientists as we run experiments and see how our images change when holding everything constant except one parameter.  We will learn about art history, art movements, and have a lot of fun expressing our creativity visually.  And we will be having a lot of fun bringing in our hobbies, interests, and culture into  our images.

About student interaction:
We will be interacting for most of the class.  I give students some time to explore and make images on their own unless they would rather work together for the entire class.  I often ask students to screen share while they are working so I can give guidance.

Students don't need any special background.  They simply need to have an interest in making interesting images.  Once they learn these skills they will be able to make images not only for their entertainment but also for school or even a business.