Neural networks are changing everything we know about creativity, design and motion design. We’re happy to be at the forefront of this process: JetStyle’s designers experiment with all the new AI tools that emerge at the market.
We asked our designer Alex Perminov to share insights about the future of video production and AI. Throughout 2023 we experienced the evolution of Midjourney and other graphic neural networks. They were changing according to the audience’s feedback, and Alex expects that video AI generators will follow the same pattern:
"When AI video generators first entered the market, they caused huge buzz. However, the reality didn’t match the expectations, so we had to pause and experiment. Currently, the primary task for those working with AI video tools is to explore their range of capabilities and leverage them for maximum benefit.
AI video generators are effective when you aim to stylize or animate an image; at the same time, they excel only at processing relatively simple prompts. You specify your prompt a lot to get the desired result.
AI video tools face challenges when asked to create photorealistic motion art. Generating videos based on a text prompt from scratch is also not their strongest suit. In my opinion, we should not expect significant improvements in this area, at least in the coming year.
The evolution of AI video tools is likely to follow the law of supply and demand, similar to what has happened with graphic neural networks. For example, people had concerns about AI-generated body parts, and Midjourney fixed the bug with its next release. Stable Diffusion offered separate tools that also solved the problem.
AI video editors have similar issues today, and it presents a potential area for their development. There are two possible pathways: neural networks may undergo training with larger datasets to better visualize human movement, or developers may release add-ons and plugins specifically targeting these issues. We’ll see how it goes and will definitely experiment with all the new features."
A few examples of how AI generators solve their creative tasks: