I have a particular fondness for the film 28 Days Later and I liked the idea of presenting an apocalyptic war scenario in a different period of history. The creation process, once the script is finished, is based on generating stills for each shot and then generating the animations with veo3. When generating stills, I first create the base in MIDJOURNEY to achieve better photography and texture. From there, I work in Freepik with Nano Banana and Seedram to maintain the consistency of the characters and different angles in each shot. The biggest challenge lies in the camera movement of each shot, since the character's movements and gestures are not usually easy to achieve and require many iterations. The greatest learning I've gained lies in how to combine different image generation tools to achieve texture, detail, and consistency. Models like Nano Banana tend to simplify details while achieving consistency. With Seedram, the problem lies in the overly bright photography, which detracts from the harsh reality I wanted to achieve in this short film. My goal is to create creations with multiple characters on screen at once and for consistency to be good.
Built With
- chatgpt
- midjourney
- nanobanana
- seedram
- veo3
- wan2.2
- wan2.5
Log in or sign up for Devpost to join the conversation.