Inspiration
Born from a Reddit challenge by one of my peers in the AI space Yickity / xrm4. A friend of mine noticed how most of my lyrics up to that point kinda dealt with lots of self-doubt and me battling with imposter syndrome. So he challenged me to write with confidence, no doubt and most of all: every line with at least double-syllable rhymes. I understood what he tried to tell me and I think it gave me the necessary push to try something new. The Theme i was looking for was one of authorship. Instead of battleing with myself over the fact that I use AI , i wanted to express that these tools help but the voice behind it is still human. The refrain says it plainly: All me, these thoughts are mine. Visual cue set by my Aidan Yagu identity: hooded silhouette, single vertical line, this time with some cyan and magenta light, fog and a female Figure both representing the female part of the AI vocals and also morphing into me showing that even that voice is again "ALL ME".
What it does
Delivers a tight performance One Take- video that feels like one continuous move from first bar to last. Puts the writing through the person front and center: dense multisyllable rap that stays conversational and clear. But again the most important fact is that it is my AI Persona that inhabits it all, it's a one take from female, to male to mystery (fog) ....and through and through it remains me controlling it.
How we built it
Lyrics: written by me to the challenge specification. Dense rap, layered, conversational. Vocal and music: Suno outputs shaped by my punctuation and phrasing method. Video: Sora1 sequences designed to stitch into a near one-shot. Prompts reused strict anchors for silhouette, wardrobe, prop, and palette. I did everything in Soras Storyboard feature , stitching endless amounts of takes together to get the best transitions and then moved from one gen to the next. Edit and grade: DaVinci Resolve free version
Challenges we ran into
Keeping the one-shot illusion with clips that vary in motion. Solved with repeated micro-cuts and fully utilizing the storyboard function in sora and the lessons learned from my Video "Ghostwriter in the Shell" Control character drift, keep the figures in frame, and move it close when the switch occurs. Make it seamless and use typical AI morphing as a special effect for a change.
Accomplishments that we're proud of
A cohesive near one-take performance piece from early Sora outputs. If you ever used Sora1 , you know how challenging this tool is. So I'm especially proud on what i accomplished. On the musical side: Dense multisyllable writing that stays readable and true to my voice. A style I refined from this point onwards.
What we learned
Sora1 is haaaaaard, but not impossible, and when your means are limited you get really creative. In fact Sora1 in some cases can still challenge modern Models in some areas. It especially comes in handy if you want to be getting really creative and make it hallucinate on purpose. Picture quality can still be top notch even on lower resolutions.
What's next for All Me
Same as with Ghostwriter, the reception was small due to my lack of reach but not off the table given more exposure. If folks start liking the video and the track i may do more background features. But for now it's a closed project.
Built With
- flux-schnell
- leonardo-ai
- sora
- suno


Log in or sign up for Devpost to join the conversation.