While generative AI like ChatGPT and DALL-E 2 are catching the attention of investors, startup entrepreneurs are looking to monetize new business models built around them. One of the more interesting ventures to emerge from space recently is Poly, which allows designers to create video games and other virtual assets, including textures for 3D models, using only text prompts.
Poly is essentially a stock asset library along the lines of Adobe Stock and Shutterstock, but populated solely by generations of AI. While platforms like Getty Images have banned AI-generated content for fear of potential legal backlash, Poly is moving full steam ahead.
“Almost everyone knows the all-too-common pain of searching online for that perfect icon, illustration, font or sound effect, only to give up and settle for something imperfect. Poly is trying to dramatically improve on this with a suite of generative tools aimed at creators,” CEO Abhay Agarwal told businessupdates.org in an email interview.
Before co-founding Poly with Sam Young, Agarwal was a research fellow at Microsoft, where he published articles on AI for social impact. Agarwal then started Polytopal, a “human-centered AI” consultancy that partnered with brands such as Spotify, Meta and Nestlé to develop various intelligent systems. Among other projects, Polytopal co-created a dance choreography algorithm for the BeatSaber game and launched a virtual baking assistant for Toll House that helps design a cookie recipe that fits users’ nutritional needs.
“Young and I started Poly in early 2022 out of a shared passion to ‘amplify the world’s creative capacity,’ and joined Y Combinator’s S22 batch,” said Agarwal.
Poly’s first tool in its planned web-based suite generates 3D textures with physically based rendering maps. In modeling, “physical-based rendering” refers to a technique that aims to render images in a way that mimics the flow of light in the real world.
Poly allows designers to describe a texture (e.g. “Tree bark with moss”) and optionally provide a reference image to get textures generated for 3D modeling. The models come in customizable resolutions and with normal and inverted maps – maps often used in game development to add volume, depth and detail to the surfaces of 3D objects.
“Poly trains its generative AI models using a variety of proprietary methods, such as extracting texture information from normal images to enhance the model’s learning capabilities,” said Agarwal.
When asked how Poly handles more sensitive content that developers may request, such as violent and overtly sexually generated imagery, Agarwal gave few details, but said Poly monitors its products “carefully and responsibly.” “No cases of damage have been reported to us yet,” he added.
Poly sees itself competing against both traditional asset markets and developers’ manual design processes. In addition to portals like GameDev Market and OpenGameArt, major game engine vendors like Unity host and sell assets through their own platforms.
Poly is also not the first to apply AI to generate game assets. Direct competitors are Hotpot and Pixela.ai, which use similar algorithms to create custom backgrounds, sprites, and other art content.
Agarwal claims that Poly’s generative AI is superior to most in terms of the quality of assets it produces. The jury is out on that. But Poly wants to differentiate itself further by expanding its generative AI service to asset types such as artwork, sprites, sound effects, and more. It plans to monetize through partnerships with enterprises, premium integrations for design tools, and by charging subscription fees for royalty-free access to assets, including commercial and resale rights.
Agarwal claims that “thousands” of developers are currently using Poly’s free service, which generates unlimited assets for non-commercial use, while “hundreds” pay for Poly’s pro plan. To date, the platform has generated over two million textures.
That momentum attracted investors including Felicis, Bloomberg Beta, NextView Ventures, Y Combinator, Figma Ventures and the AI Grant, who collectively contributed $3.9 million in venture capital to Poly on Y Combinator demo day in September.
“Poly’s clients range from professionals at Fortune 500 companies to individual freelancers in game design, AR/VR, interior design, architecture and 3D rendering for e-commerce and marketing,” said Agarwal. “Poly has a multi-year runway and can focus on building the best possible technology as a higher quality product is required to stand out and win in this emerging and highly active space.”
Assuming Poly catches on widely, it and its generative AI rivals risk upsetting the artist community – not only because they could threaten livelihoods, but also because generative AI systems have been shown to regurgitate the data they were trained on (e.g. ). On art community portal ArtStation, which allowed AI-generated art on its platform for the first time earlier this year, members started protesting widely by posting “No AI Art” images in their portfolios.
The aforementioned legal questions surrounding the technology also remain unresolved. One class action lawsuit claims that GitHub’s code-generating system, Copilot, regurgitates portions of licensed code without providing credit, which could have implications for art-generating AI systems and those who use art created by them. In an unrelated case, the US Copyright Office recently ended copyright protection for a comic book created with generative AI after it was initially granted, saying only works created by humans are entitled to protection.
However, Agarwal isn’t worried – or if he is, he won’t show it.
“Generative AI is heavily criticized by makers and seen as ‘anti-creator’ as many companies in this industry want to replace makers with automated systems. However, Poly’s focus has always been to give makers easier access to design resources,” said Agarwal. “Building on the current momentum, Poly plans to continue its relentless focus on its own generative AI innovation, model training and product development to support more types of design assets and be embedded in designers’ day-to-day workflows.”
Poly currently has three employees and plans to double the team in the next six to 12 months.