This is great work and the description text is quite insightful. Is it possible to get to know better about how it was developed (conceptually and technically)?
The idea behind this was to test old and new prompt structures in RunwayML Gen-3 which is their latest model. The speed of movement and fidelity of image has been greatly improved with the new model, and it also looks as if the training data has been expanded. Interrogating any of these media creating AI with even a basic prompt like "city street" can provide a fascinating insight into what a generic "city street" is conceptualized of within the latent space inside the model.
This is great work and the description text is quite insightful. Is it possible to get to know better about how it was developed (conceptually and technically)?
The idea behind this was to test old and new prompt structures in RunwayML Gen-3 which is their latest model. The speed of movement and fidelity of image has been greatly improved with the new model, and it also looks as if the training data has been expanded.
Interrogating any of these media creating AI with even a basic prompt like "city street" can provide a fascinating insight into what a generic "city street" is conceptualized of within the latent space inside the model.