Same Old Track

This project was completed in around 3 days and was a push to try and tell an emotional story of father and son relationships.

I wanted to create something completely with AI to really challenge the narrative that AI content is 'slop'. This topic is one that is close to me and effects all men across the globe. I wrote these original lyrics a couple months ago but my like of musical talent (and terrible singing voice) meant i knew could never really bring my full creative vision to life.

After playing around with AI tools over the last few months i knew this was the perfect opportunity to really bring together something powerful. I started by inputting the lyrics into 'makebestmusic.com' which i believe is a wrapper for SUNO music engine. These lyrics came alongside a long, detailed prompt and the song came to life.

After this process i created a storyboard, linking each individual 3-4 seconds to a visual idea. My process was literally to listen to the song on repeat for an extended amount of time, each time visualising what came to my mind as the lyrics were played.

The next step was to create the 2 characters for the story. I made these using my own prompts within CapCut using Seedream 4.0. This involved a close up as well as full body shot and then creating the younger versions of the characters for the flashback scenes. These character reference images were then used alongside Gemini / ChatGPT assisted prompts to create the individual scenes. Each image took around 2-3 attempts to get exactly where i wanted.

Once all static images for scenes were created, i placed them in a timeline over the song within CapCut and decided roughly how long each image would last. the scene without lip syncing were then upscaled using GigaPixel and brought into Flow to apply VEO3. A mix of JSON style prompting and normal descriptive prompting using ChatGPT to optimise was used.

For the Lip Syncing effects, i used Dreamina.AI's Avatar model. Inputting the upscaled picture alongside the cut audio section. Dreamina was the ideal choice as it allowed for prompting of gestures / small actions alongside the lip syncing.

Kling.AI frame to video tool was also used for this project, mainly for the animation of scenes featuring the charecter as a young boy as VEO3 guidelines do not allow it.

I learnt a lot from this project and was very happy with the end result.

Full version is uploaded to YouTube (please play in highest resolution)

Built With

• capcut
• dreamina.ai
• kling.ai
• veo3