Everlane ‘Playback’
A.I. Spec Ad for the clothing company Everlane.
Tools used:
ChatGPT, Midjourney, Runway Frames, Kling, Runway, and Topaz
Scroll down to see how it was made
SeanGPT
After researching Everlane’s Target Audience, I submitted a prompt describing a persona, inspiration images of Simon Stahlenhag’s work, and Everlane’s URL, to a custom marketing GPT.
MidJourney
Once I discussed the campaign in depth with the custom transformer, we made a Shot list for a 30 second ad. But this duration quadrupled because I was having too much fun with the tech.
To style the images, I uploaded Everlane’s brand photos to Midjourney’s ‘Moodboard’ feature. From there I generated style guide images that were used as ‘weights’ for every subsequent image.
Storyboards
Using the Everlane style weight images, I generated Storyboards for each shot / scene.
Casting
I prompted for hundreds of generations in search of an Anchor Reference Image, using a gray, studio backdrop, and an ‘even lighting’ description.
The Deck
From here, I created the campaign slide deck with the help of the same custom transformer.
-
References + Frames
I combined Midjourney and my Frames Character Spread using reference slots and prompts to place the character in the scenes described in the shot list.
-
5 & 10 Second Generations
I generated videos using Runway’s latest Gen-4 Model. After performing the voice over myself, I used Runway’s Speech Generator to create the voices of each character.
-
Physics & Action
I used Kling 2.1 for scenes with multiple people, complex walking shots, and visual effects.
From ‘This’ to ‘That’
4K Upscaling
I used a desktop upscaler application called Topaz AI to clean the generations from Runway and Kling.
NLE & Sound Design
Finishing with After Effects, Epidemic Sound, and believe it or not, FCPX.
Thank you for engaging! Please don’t hesitate to reach out with any questions. I’d be glad to answer them.