Mastering Art Fundamentals: The Key to Commanding DALL-E
DALL-E’s Mouse From “I need a picture of a mouse please.”
Boulder Schools are amongst the finest. How many of our students both use computers and also study art in school? Without that education, it’s like being given a race car without first knowing how to ride a bicycle or drive a car.
Just as pocket calculators replaced slide rules and as computers appeared in our lives again changing the rules and tools of education, we now have AI. Artificial Intelligence. I’m going to address CHATgpt and DALLE only, since that what I’ve played with. DALL-E, the image generator linked with CHATgpt, acts as if it has its own imagination. Here I asked for a cartoon illustrating the progress from slide rules to calculators to computers, style of RC Crumb (Fabulous Furry Freak Brothers’ artist).
What does that have to do with old school?
I have long lamented the losses of art classes in school. Here’s why I see that is such a current concern and a loss to those not paying attention. If I say to DALL-E “please paint a landscape in the style of Van Gogh” here’s the reply. 
“Paint a Landscape In The Style Of Van Gogh”
I love the results. I had to know what I wanted, and to more to the point, had to know what “style of Van Gogh means. DALL-E AI will within limits (it cannot spell well yet) produce what I ask. The more I know about art and artists, the more I can work with DALL-E to get the picture that I want. For example, let’s try “a mouse sitting at a table writing Shakespeare plays in the style of Vermeer.”
You’ll recall that Vermeer painted during the Golden Age of the Dutch Masters along with Rembrandt. Vermeer is known for his lighting, his astonishing detail, and for the small number of paintings that exist. The window source of the room light is “Vermeerish.” So are the colors and the style. But I first needed to know what a Vermeer might look like. If I had said in the style of R.C. Crumb, the results would be a tad different. 
“Please paint a mouse at a desk with a quill pen in hand, writing a Shakespeare play in the style of RC Crumb”
What about limitations? For example, if I ask WALL-E to draw a picture in the style of Roy Lichtenstein it tells me that my request cannot be fulfilled. “I’m unable to generate the image in the style of Roy Lichtenstein due to content policy restrictions. However, I can help create an image inspired by the general pop art style if that would work for you. Let me know if you’d like to proceed with that or if there’s something else you have in mind.”
OK, no problem. I’ll take the hint and ask for the same thing but instead of the name of an artist who is too recent, and who presents IP, intellectual property, or copyright problems, I’ll do what DALL-E suggests. “Yes, proceed with that.” 
Pop Art Style Without Naming Roy Lichtenstein
The software can provide the intelligence portion quite adequately. The “art” portion, the pictures, depend upon the instructions you give the software for image generation. While DALL-E does well with simple general instruction like “paint a mouse,” who do you want in control? Is DALL-E a tool or is it a beginning AND endgame?
CHAT tells me that the image generation aspect of this AI is DALL-E, and adds: Yes, “DALL-E” is spelled correctly when referring to OpenAI’s image generation model. The name is a play on the famous artist Salvador Dalí and the Pixar movie character WALL-E
If I’m still not getting what I want, I can “regenerate” with the same instructions, or I can try something quite different. Fan of Goya? Let’s see what DALL-E makes of that. 
The options I have are limited by what I know to ask for. The more we know about art and art history, the more “paint brushes and tubes of paint” we have at our disposal. What will DALL-E do with a request to use the style of Ansel Adams after telling me it violates the rules (copyright rules I’m sure)? 
OK, pretty cool, but Ansel Adams generally photographed outdoor scenes, usually of the West. Obvious good taste! So let’s try “that same style” for Devil’s Tower. This took me several modifications. 
The Van Gogh effort captured his style. The Ansel Adams attempt, after several tries, came up with a great picture, though not as close in style as the Van Gogh.
My point is that the more of the language of images you know, the more classes, books, museums and the like you’ve exposed yourself to, the more interesting and varied your results will be and the more you’ll be able to translate an image in your mind to the results you get. While the results are generally unpredictable, the more broad and informed your “art vocabulary” is, the more control you have, and the better the results.
Lenny Lensworth Frieling
Shared Knowledge Is Power!
![](https://b128718.smushcdn.com/128718/wp-content/uploads/2018/07/2-2.png?lossy=1&strip=1&webp=1")
