Monalia's World

Observations on a New Life in Spain

Skip to: Content | Sidebar | Footer

Jumping into the AI multiverse (using Midjourney illustrations)

26 March, 2023 (09:49) | Living in Europe | By: admin

I know it’s silly, but I hate reading peoples opinions about the impact of AI on culture and civilization. To me they all simply state the obvious. But I guess its to be expected. 

I remember going to the annual music conference in LA in  1983, the year MIDI was introduced. (for non musicians who are not familiar with MIDI, it was introduced in 1983 at NAMM convention (National Association of Music Merchants) it modernized music in an unique and useful way by creating a “universal synthesizer interface.” The word MIDI was being received with typical mix of reaction, some who embraced but many who resisted, or maybe simply didn’t understand. But it has proven to be one of the most brilliant advances in music. The whole music world reacted with all kinds of crazy opinions about this essentially simple technology which modernized the music industry overnight. After that it was samplers, sequencers, then auto tune…At the outset no one understood they are all  *tools* which when used with moderation and no fear, can result in a truly original sounds. To be clear, for awhile the “Cher Effect” (autotune) was used ad nauseam, making pop music more generic than ever. But that’s another discussion. And that’s not to mention the over saturation of both sampled sounds and autotune in the Hip Hop genre.
For this image I use the following prompt, no qualifiers to make it weird:
(prompt) /imagine a powerpoint presentation with a conservative moderator and colorful, casually dressed musician on a platform in front playing a Roland synthesizer.

As my limited experience with AI unfolds (I only started using AI this month), I immediately got great results on my first try. The only AI I’ve used in this post is Midjourney, though my next post will be a multi-media CHAT GPT4 scene I created to fun effect.

I am a novice to using (the apparently controversial) CHAT GPT4, as well as midjourney4 but I already understand it’s purpose, use and in short time already hit linguistic limitations, thus establishing the given parameters for creative collaboration. Again, its a tool.

What many people don’t know is that they’ve been using AI for years just by learning to use succinct and specific commands in the words they put into their search engines; this will get both the results they want plus some amusing related surprises sometimes.

In 2023 we have been given a new tool (AI). Any artist or wordsmith with clear ideas can cajole the results they want, with surprisingly delightful results.

With Midjourney, in my experience the key to unique images is to give it a prompt with succinct visual detail and specific adjectives to get those specific results. The important thing is to remember the “arguments” and have a link or a notepad with the exact words to use for desired effects so you can copy and paste them (the arguments) at the end of your prompt to ask for specific exaggerated effects.

I’ve found there are more limitations in Midjourney using your own photo as a source image, but once you get a feel for it you can put it to good use.

selfie with my dog, Zuma, taken in last month in Albi, France
Here’s a selfie I took of myself and my dog Zuma in France last month. I used the image as source material, asking Midjourney to take the woman (me) and dog and to make my dog into a robot assistant, and to put us at Ponte Vecchio in Florence, knowing it has a similar background to the original. I had to try several “arguments” to get my desired results.

This is the prompt I came up with to get the above mage:

hyper-realistic add a mechanical steam punk robot to this photo of a woman in front of Ponte Vecchio in Florence, Italy cinematic lighting ultra photoreal –ar 16:9 –v

And here’s a variation by slightly changing the prompt omitting the word dog.

This image was just a test, using the prompt:

hyper-realistic add a mechanical steam punk robot to this photo of a woman in front of Ponte Vecchio in Florence, Italy with cinematic lighting ultra photoreal –ar 16:9 –v 5

The above are experiments with “arguments”, or commands. I forgot to add the word “dog” in the above prompt, but I like the fun, unexpected result of the collaboration.

In Chat GPT4 I have found you have to feed it the exact right, succinct and visual words as possible to work from. My very first attempt bore surprising results when I came upon a persistent AI bias which tries to make all scenes or stories end with a happy ending or one of the people becoming a hero. Avoid using the word naked or suicide or anything with ambiguous meaning,These little walls make me laugh, and for a moment it becomes the familiar old game of “tricking the program into doing what you want”, which forces a writer show rather than tell the story they are collaborating on with AI. This is the part where once you like the skeleton you two have come up with, it becomes a tool where the writer simply goes in and rewrites the AI results in their own language, adding, deleting, replacing passages and you can give it whatever sinister or ambiguous ending you choose. It’s kind of fun.
I’ll be experimenting with mixed media AI in the next few blog posts, why not?

But, going back to Midjourney, here’s my first experiment “imagining” a female chamber orchestra conductor – using various Arguments. (arguments are html code instructions you can find on any Midjourney tutorial. You just need to keep them handy until you learn the logic and lingo).

I leave you with these fun images of a female conductor in different styles and ethnic origin heh heh.

May be a cartoon
These were all my first attempts at trying out different “arguments” and also different adjectives. I share these to give you an idea of how fun Midjourney4 is. In hindsight I should have specified only one baton, heh heh heh.