I’m torn. I’ve been working on this music video for about a week as I’ve been polishing the song itself, and approached that video the same way I would any other video project- dropped the MP3 into Final Cut Pro, storyboarded the cuts/scenes/transitions.
But then – instead of just taking a shot list and capturing things with a camera – sent each individual clip (with hand-drawn storyboards, motion scaffolding and character references) to Google Veo via a series of Python scripts, replacing each still storyboard on the timeline with real video matching the storyboard description. I really wanted to see what it would be like if AI simply replaced the capture/camera part of the process, leaving the rest intact. And that part was a HUGE success.
I used XML import/export to have Claude Code review my timeline for continuity errors and mis-aligned elements, and generate the clips to spec, under a strict style guide. Each individual clip was rendered specifically to a storyboard and story arc/script bible using my own character/style sketches – no guesswork allowed.
I then hand-assembled all the rendered video clips in FCP, added and timed every damn transition and blend effect by hand, color graded the sections uniquely for effect and then added the titles/motion graphics on top. Just like I would any other production. And then sat back to let it all sink in.
And kinda hated it.
The visuals match my character sketches. All the scenes are exactly as I storyboarded them, and the characters/color palettes stay consistent shot-to-shot. Wardrobe is consistent, the build and narrative came out pretty much as I envisioned it, right down to the opening/final frames.
Still kinda hated it. And didn’t feel nearly as proud of it as I would a fully hand-built project, if I’m being really honest.
It’s not that the pieces weren’t all in place, or that I had let any ‘outside influence’ corrupt the project, as it only used my artwork, style guides and directives. It’s not that it didn’t really create the clips I asked for with the visual components I specifically outlined, because it did.
It’s that it just doesn’t feel like me. Or even like a human.
It’s kinda OK, I guess – if you don’t know or expect better. Or if you’re already overloaded with AI slop and this just falls in line.
If I’m being super honest – I’m not sure I want this visual work representing me as an artist or creative. Pretty sure I could keep scaffolding, nudging, adding context and hand-rendered art to constrain AI to my specific vision even further… but I’m not sure I want to. It’s much better suited to helping me automate the technical, non-creative parts of the project so I can focus my right brain on visuals, sound and experience. Just not the art.
Anyway, here’s the video/single as I left it. The single itself still has some vocal tracks to go before it’s finished, but I need to take a break entirely from it now.
It’s not right… but not wrong, and I’m not sure I want to try to tackle that last 5-10% of time it might take to change that. Totally understand if you don’t care for the visuals either, but at least the music is 100% human-generated so I can feel 100% comfortable with your opinions on that… (bum notes and all) 🙂