Overview

In late January 2024 I was approached by Stability AI (https://stability.ai) in regards to testing out the closed beta of their web-based Stable Video Diffusion (SVD) platform called Stable Video (www.stablevideo.com) It was a huge honour to be personally approached by one of the largest AI companies in the world who have brought so much to the AI space with its generative models.

The Process

A big part of the project was to trial & error and learn by doing in terms of testing out the SVD model and its capabilities within the Stable Video app context. Many features were still in Beta and/or Experimental to their nature, which was a big challenge in terms of the amount of trials needed to get the footage I was looking for. Nonetheless, it was really fun learning the specific quirks and Ins & Outs of the SVD model that I previously didn't have the chance to work with.

In terms of Art Direction and Cinematography, I went for the highly cinematic route, leveraging my portfolio of Midjourney V.6 Alpha Photographs in my AI portfolio- Aiming for a dynamic juxtaposition of Close up & Macroshots mixed with scenes with vehicles, aeroplanes and cities.

Stable Video & the SVD model produces 4 second 25 frame videos, so as always, I used both Topaz Video AI to upscale all footage to a crisp 4K Resolution with 60 Frames per second for that polished look.

The result

The final showreel was the result of 4 intense days of learning, generating, editing, tweaking, upscaling and assembling this piece of AI cinematography.