AI video startup Haiper emerges from stealth, plans to build AGI with full perceptual abilities

As OpenAI continues to drive virality with the content material produced by Sora, its yet-to-launch AI video platform, rivals are going all in to push the bar with their respective choices. Just some days in the past, Pika Labs launched lip-sync to its product. And now, a completely new AI video startup “Haiper” has emerged from stealth with $13.8 million in seed funding from Octopus Ventures.

Based by former researchers at Google’s Deepmind Yishu Miao (CEO) and Ziyu Wang, London-based Haiper gives a platform that enables customers to generate high-quality movies from textual content prompts or animate present photographs. The platform has it’s personal visible basis mannequin beneath the hood and takes on present AI video instruments out there akin to Runway and Pika Labs. Nevertheless, from early exams, it nonetheless seems to lag behind what OpenAI has to supply with Sora.

Haiper plans to make use of the funding to scale its infrastructure and enhance its product, finally constructing in the direction of an AGI able to internalizing and reflecting human-like comprehension of the world.

What does Haiper provide with AI video platform?

Very similar to Runway and Pika, Haiper, in its present type, offers customers with a web platform, the place they get an easy interface to enter a textual content immediate of their selection to start out creating AI movies of something they may think about. The platform at present gives instruments to generate movies in each SD and HD high quality, though the size for HD content material stays restricted to 2 seconds whereas SD can go as much as 4 seconds. The lower-quality video instrument additionally comes with the choice to manage movement ranges.

Once we examined the product, the output for HD movies had been extra constant, in all probability because of the shorter size, whereas the lower-quality movies it produced had been typically blurred out with the topic altering form, dimension and shade, particularly in instances of upper ranges of movement. There’s additionally no choice to increase the generations, as seen in Runway, though the corporate claims it plans to launch the aptitude quickly.

Along with text-to-video options, the platform additionally offers instruments to let customers add and animate present photographs or repaint a video, altering its type, background shade, components, or topic with a textual content immediate.

Haiper claims that the platform and the proprietary visible basis mannequin underpinning it will probably cater to a variety of use instances, proper from particular person purposes akin to content material for social media to business-centric makes use of like producing content material for a studio. Nevertheless, the corporate has not shared something in regards to the commercialization plan and continues to supply the expertise free of charge. 

Plan to construct AGI with notion of the world

With this funding, Haiper plans to evolve its infrastructure and product with the last word purpose of constructing a man-made common intelligence (AGI) with full notion skills. The funding takes the full capital raised by the corporate to $19.2 million.

Within the subsequent few months, Haiper plans to iterate from consumer suggestions and launch a collection of enormous educated fashions that can improve the standard of AI video outputs, doubtlessly bridging the hole between rival choices obtainable out there.

As the corporate scales this work, it is going to look to reinforce the fashions’ understanding of the world, basically creating AGI that might replicate the emotional and bodily components of actuality – masking the tiniest visible facets, together with gentle, movement, texture and interactions between objects – for creating true-to-life content material.

“Our finish purpose is to construct an AGI with full perceptual skills, which has boundless potential to help with creativity. Our visible basis mannequin shall be a leap ahead in AI’s means to deeply perceive the physics of the world and replicate the essence of actuality within the movies it generates. Such developments lay the groundwork for AI that may perceive, embrace, and improve human storytelling,” Miao mentioned in an announcement.

With these next-gen perceptual capabilities, Haiper expects its expertise will even transcend content material creation and make an affect in different domains, together with robotics and transportation. This method to video AI makes it an attention-grabbing firm to be careful for within the red-hot AI area.