Finally finished a concept I've been wanting to make for a while: Making a film using only AI tools from end-to-end.

Runway Gen-3 Alpha and their new(-ish) image-to-video made a big difference, but it was the release of Flux Dev to create the concept art that really convinced me to finally give it a go.

- ChatGPT to write the script and turn it into prompts for Flux Dev
- Flux Dev to generate the concept art
- Runway Gen-3 Alpha to turn concept art into video.
- UDIO to create an epic score (ChatGPT also helped me write the prompt for that)
- Topaz to upscale from 720p to 4K.

I still did the edit in Premiere Pro, so it wasn't 100% AI.

Tons of issues with the final output - the script had a decent narrative arc, but it's hard to preserve the consistency throughout, although the big context window of Flux does make it easier than other image generators. And because I have a day job, I very often picked the first or second take from both Flux and Runway, could definitely have been more selective.

But for 4-5 hours of working with the AI tools and 30 minutes of editing I'm pretty happy with the result.
Back to Top