This article is also available in:
Français
After the emergence of tools like Midjourney, which allow the generation of images from text using AI, many companies joined the quest for the next step: text to video. In other words, creating animated videos, realistic or not, from text prompts. The major problem with these tools, until now, is often the lack of temporal stability. Artifacts will appear and disappear during the animation, giving an unnatural effect to the result.
Describe a video, Pika will create it for you
It is in this context that the team at Pika presents version 1.0 of its solution: a tool that generates quite stable and detailed videos. We will let you judge for yourselves:
Obviously, the end result is not perfect, but this is a huge leap forward compared to what was achievable a year ago.
It is therefore not surprising to learn that besides this official launch, Pika also announces having raised a total of 55 million dollars from different investors. Pika details that this fundraising was initiated with “pre-seed and seed rounds led by Nat Friedman & Daniel Gross, and our Series A round led by Lightspeed Venture Partners. Our investors also include Elad Gil, Adam D’Angelo (Founder and CEO of Quora), Zach, Andrej Karpathy, Clem Delangue (Co-Founder and CEO of Hugging Face and Partner at Factorial Capital), Craig Kallman (CEO of Atlantic Records), Alex Chung (Co-Founder of Giphy), Aravind Srinivas (CEO of Perplexity), Vipul Ved Prakash (CEO of Together), Mateusz Staniszewski (CEO of ElevenLabs), and Keith Peiris (CEO of Tome), as well as venture firms such as Homebrew, Conviction Partners, SV Angel, and Ben’s Bites, alongside many other esteemed industry leaders and AI experts.” In a nutshell, mainly inverstors from the AI and creative industries.
Target audience, Copyright issues, Business Model
Pika explains to Forbes that the company is “not trying to build a product for film production”, but rather “something more for everyday consumers”, for people “who are creators at heart, but not that professional”. As Pika explains, this is a way of differentiating themselves from other tools targeting a more profesional audience, such as animation studios.
This explains why Pika is currently free. A tiered subscription model might be launched in the future, with more features for paying users.
Obviously, when it comes to AI, the issue of copyright always comes up, whether it’s training data or results. Pika CEO Demi Guo explained to Forbes that the company is working on tools filtering copyrighted materials. However, “that’s still very exploratory” she explains. Commercial use of Pika thus seems to pose ethical and legal issues at the moment.
You can sign in to Pika for free, but please note that a waitlist is in place.
Every week a revolution?
Pika gives us a glimpse of a possible future where people could create entire videos from nothing but text. This announcement is yet another proof that AI is a constant-evolving field. It is very hard to predict where we’ll be one or five years from now, when there is a big announcement every week. We will, of course, keep following this field very closely.
