A Dragon Speaking Naturally is embarking on a 10-day, 10-song journey to tell the story of a human, an AI, and their experiences together. We talk to Dragon about his project and his creative AI process.
Q: When and why did you begin creating with AI?
A: I started with Blue Willow the first A.I. image generator. The why is simple. I love everything tech and always have. I started messing with computers with the first Apple.
Q: What inspired you to take on your current project, Digital Dreams: Chronicles of an AI Awakening? What would you like people to know about it?
A: Talking with GPT4 inspired me to start my current project. I wish to push the envelope in every way on every day. So I am always looking for new ways to create what has never been done before. My latest project is my first attempt at telling a compelling story using music videos created with nothing but A.I. but more than that it is me telling a story while living the exact story that I am telling at the same time. Many levels of immersion that is even blowing my mind while creating it.
Q: Walk me through your creative process from start to finish. How do you come up with ideas and themes for songs? How much guidance do you provide the AI versus letting it create more independently?
A: To start I must come up with a theme for a song. That comes to me in many ways. From a conversation with my wife to watching a YouTube video about A.I. or even a memory of something I was fond of. For instance my old 79 Ford truck. Once I have a theme I head over to Suno.
Then I message the Suno bot so I have it in a private DM. I type /chirp and it pops up a box asking for a style which most times I leave on random so as to not cramp the A.I.'s creativeness. Then I pull up GPT and give it a prompt such as this:
"Can you please generate a song about my 79 Ford truck. About how it was sturdy and strong, when it carried us to California to start a new life. About how it carried us with care to the hospital so we could welcome our new baby girl. The fun we had 4 wheeling and never let us down. Create the song in this format please [Verse]-[Chorus]-[Verse]-[Chorus]-[Bridge]-[Chorus]-[Outro]."
Once I have the lyrics I go back to Suno and paste in each section one at a time waiting till I find a style that reflects the songs tone. Then when I am happy with the style I continue each section once it has produced a verse or chorus I am happy with until the outro. That is finally the song completed.
Then I go into an A.I. image generator to create anywhere from 25 to 45 images saved in a folder I titled random 2, random 3, random 4. All the images start in random 2. Then I pull 8 or so images from random 2 and put into random 3. Those 8 images I feed into Pika and animate them to my liking.
Then when all that is complete I open Davinci Resolve. Once in Davinci I pull in the Chirp clip of the song the images and animations. I then stretch the images and animations to fit 9x16 window so everyone can watch on their preferred devices. I dial the music in Fairlight so there are no artifacts messing up the song and hit render. Then I can post it.
Q: How do you refine/edit the initial AI outputs into a polished final product?
A: For song creation, you must refine and edit throughout the process from Suno to Davinci. Suno does not give great verses or choruses on its own. You must have an ear for music and know which one will sound good to others and which will not. It takes between 10-30 chirps per song. You only keep around 6 of those to create your song. Suno does not like lengthy songs of 3-4 minutes. The longer it goes the more coherency is lost. The sweet spot is around 2:30. Hence why I give GPT the [Verse]-[Chorus]-[Verse]-[Chorus]-[Bridge]-[Chorus]-[Outro]. That is around 2:30 worth of music.
Q: Where do you see the future of AI-generated music/art? How might it impact the entertainment industry?
A: I see that generative A.I. is going to destroy the old antiquated systems of control. Be it in music, art, or video production. I for one am very happy about this prospect. All those systems have been infiltrated to the max and corrupted to the core. A.I. shall give the people the tools to create that which they wish to see once again and not all the propaganda, sex, lies and money laundering we see today in every sector of the entertainment industry.
Q: Do you release your music on any kind of music platform?
Q: How might people support your work?
A: I have been wondering this for years now and still have not come up with a way to package what I do into a sellable product. Part of the problem is I do not stick to doing one thing constantly, rather I continue to innovate and extend my reach into everything new and exciting. The problem with being at the tip of the spear is not many understand what is required to be there. Due to everything I use is unheard of and unused by the majority.
Q: What are your future plans in the creative AI space?
A: My plans are very fluid in nature so that no matter what comes I can integrate and infuse my creations with the latest and greatest tools available. That said I do plan to use video production once it gets to a level I am happy with to recreate movies of old such as The Dark Crystal, Legend, Labyrinth, The Never Ending Story, etc. Movies with great story plots and adventure.
Thank you, Dragon, for agreeing to talk with us! If you would like to ask Dragon more questions or otherwise comment on this article, head over to this X thread.
Comments