Pika Labs new generative AI video tool unveiled — and it looks like a big deal::The new Pika 1.0 tool comes after a $55 million funding round for the generative AI company and is a big step up in AI video production.
It doesn't really do all the work of an artist though. It generates pictures, but consider that a camera also generates pictures of things, and yet photography is considered an art form these days, and one's results from doing that can vary quite a bit between someone who understands both artistic principles and how their tools function, versus someone who does not. Having an image generator does not also entail knowing what to ask the generator for, or how to make any adjustments to it's output if it gives you something that is close to what you envision but not quite there. If anything, I personally suspect a more mature version of the technology will get integrated into art tools in some way rather than looking like it currently does, because a text prompt is a somewhat vague and inexact way to describe an image. If you ask it for a spaceship, for example, it'll give you some sort of spaceship, and if you ask it for a specific spaceship from pop culture it may likely give you that, but if you're imagining a specific design for a spaceship, with specific details, that does not already exist in existing art, it would be very hard to completely describe that just through text, versus if you could start sketching out and have it sort of act as a kind of graphical autocomplete that you can steer in given directions.
It doesn't really do all the work of an artist though. It generates pictures, but consider that a camera also generates pictures of things, and yet photography is considered an art form these days, and one's results from doing that can vary quite a bit between someone who understands both artistic principles and how their tools function, versus someone who does not. Having an image generator does not also entail knowing what to ask the generator for, or how to make any adjustments to it's output if it gives you something that is close to what you envision but not quite there. If anything, I personally suspect a more mature version of the technology will get integrated into art tools in some way rather than looking like it currently does, because a text prompt is a somewhat vague and inexact way to describe an image. If you ask it for a spaceship, for example, it'll give you some sort of spaceship, and if you ask it for a specific spaceship from pop culture it may likely give you that, but if you're imagining a specific design for a spaceship, with specific details, that does not already exist in existing art, it would be very hard to completely describe that just through text, versus if you could start sketching out and have it sort of act as a kind of graphical autocomplete that you can steer in given directions.