Fallen On Broken Wings

This is another music video where all the images in the video are created or mish-mashed together by AI image generators based on the lyrics.

Fallen On Broken Wings
Lyrics and Music by Timothy Price

[Intro]
Grounded, fallen on broken wings
Angels are no more
Mortals fight for monkey’s rights
To knock on Heaven’s door

[Verse 1]
Homo sapiens that we are
Biped anthropoids
Fugacious errant avatars
Driving around in motorcars

[Chorus]
Oh Oh
how time has changed
Men in buns and skinny white girls
Meld into a single dress
Cultural Smores
Exchanges

[Verse 2]
Homo habilis was all
Handyman for an era
Gave rise to H. Erectus
Who was so effectus
To make us what we are

[Bridge]
Well you know
It amazes me
That we made it this far?
History does it again and again
D.C. al Coda, assures we’ll never win

[Chorus]
Oh oh
How time has changed
Men in buns and skinny white girls
Meld into a single dress
Cultural Smores
Exchanges

[Coda]
Grounded, fallen on broken wings
Angels are no more
Mortals fight for monkey’s rights
To knock on Heaven’s door

58 thoughts on “Fallen On Broken Wings

      • It is really weird how they mish mash images together. What’s more interesting is that different AI imaging generators will fetch the same images for some of the lines, which means different AIs are gathering images from the same sources.

      • They are supposed to learn from the content you feed them, but on images they are get stuck on the same images. If you get on the the AI image sites and see the examples of the questions asked, you see there is a real lack of art knowledge and creativity in the questions and results.

      • If you think my lyrics are garbage, I suppose so. I don’t think it’s quite the same old GIGO cliché. I think the issue is more that AI is not up to the level of interpreting abstract ideas and making the into images the way people do. Poetry and lyrics mess with grammar and structure. Considering people have trouble interpreting poetry and songs, AI is going to have issues as well.

      • I’m sorry I was unclear. I wasn’t referring to your lyrics, I was referring to “If you get on the the AI image sites and see the examples of the questions asked, you see there is a real lack of art knowledge and creativity in the questions and results.”

      • That is still not GIGO. They get out what they put in, I simply find most of it is stupid with no purpose. But then, you can say the same about my music and me trying to get AI to create images based on the lyrics. It certainly doesn’t seem to be serving much of a purpose. There doesn’t seem to be much entertainment value in the experiment.

  1. I did a fair amount of AI images in the fall and really enjoyed playing with it. I haven’t played with in 2023 and finding here what you are doing with it is fun. I enjoyed this video a lot.

    • Thanks, Susan. It’s an interesting exercise. However, I’m finding I’m not getting enough variation to keep in interesting.

      • That may also be why, after being totally immersed for a couple of months, I have had no desire to do more in 2023. 🤔

      • ChatGPT is excellent for writing scripts and anything with words. Dall-E 2, its image generator, leaves a lot to be desired.

      • I didn’t like working with Dall-E, especially for “gray-haired women” just living their lives. MidJourney was the one I liked working with best. I have not tried any of the Chat-bots.

      • The first two days I was on it in September I used up all my free allotment. In the next week I used up my $10/month allotment and went to $30/month through December. I used it enough I felt I got my money’s worth. Then I got burned out. 😱😂

      • I just don’t find I get good enough results for what I’m doing to pay a monthly fee. While MidJourney produces high quality images, they are still generally too fantasy and gamer-like for what I would ultimately like them to produce.

      • During a fall I learned a lot from an Italian artist (Patrizia Burra), who used Photoshop and additional techniques to create her images. I did those things also for the ones I liked. Another issue is that AI images cannot be copyrighted.

      • Interesting, because some of the AI generators are pulling copyrighted images off the web.

      • That is probably THE major problem with AI and probably the reason that they cannot be copyrighted in the US. I am doing phone photography for the most part now (can’t really use my real camera now). 🙂

      • I use my phone most of the time these days. I get out with the Bazooka every now and then.

    • Thanks, Janet. As I mentioned to Liz, the examples on the AI image sites show a general ignorance of art and a lack of creativity by the questions asked of the AI image generators and the resulting images. When given my lyrics, the AI Generators do not know how to deal with creating relevant images. I also found there is a lot of images generated with cultural biases and stereotypes from neutral questions. I found that more disturbing than the bizarre images.

  2. Well, the images at the beginning were beautiful and then got creepy. Like a happy movie to a creepy one lol but it was good song!

  3. Tim – we are entering a time of opportunity with AI, and I am definitely playing catch-up on this one. Do you have a particular app or AI site that you use?

    • Thanks, Rebecca. I used Draw Things, Dall-E 2, Fotor AI, Craiyon, and Night Cafe. I messed with MidJourney, which is the best, but it’s too expensive for experimentation. I run out of credits quickly on all the apps but Draw Things, therefore, I use Draw Things the most.

  4. Brilliant song and lyrics, Tim! Funny interpretation by AI. Definitely better than the last one however I’m still not convinced why ai should be involved in art. Au contraire. 😉

  5. I had this in my tabs and just realised I hadn’t watched it yet! The images generated were muy interesting, I have to say.
    Your lyrics are really well thought out. You’re good at this stuff, yanno?

  6. It will be interesting to see how this all plays out. Maybe a good litmus test to see if the engines are “learning” (in abstract only as it is simply relating) by running the same inputs through later in the year and see if the generated images are getting closer to your intent. It’s going to eventually take a lot entry jobs, just hope it doesn’t get manipulated against us.

    • I’ve already managed to piss off ChatGPT for being persistent in trying to make it give me real information. I asked Chat for references for research I’m doing to help Laurie. It just made up a bunch of relevant-sounding articles with authors, and dates with summaries of what the articles were about. None of the articles exist. Chat made up all of it. After many tries, I got one book that exists (I had already gotten the e-pub for the book), and two journals that exist, but the articles and authors who are supposed to be in those journals are not. Supposedly v4 of Chat is better, but I’m not paying no stinking $20/month to find out.

      • Yikes, that is piss-poor if it is making up references…I guess from their point of view it can never be wrong if they make up their sources.

    • I have another video with AI created videos. I had a mellinial and a gen z watch it. They liked the song but they thought the AI images were really creepy. I don’t think I will ever post it. Thanks, Resa.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.