Big 3D model company VAST received tens of millions of dollars in financing Tripo Studio, the world's first AI 3D workbench: from “leading algorithms” to “closed loop workflow”

The Zhitong Finance App learned that the big 3D model company VAST has once again completed a pre-A+ round of financing of tens of millions of dollars. This round of investment was led by the Beijing Artificial Intelligence Industry Investment Fund, followed by Jingya Capital. VAST's previous investment shareholders include Oasis Capital, Dachen Financial Intelligence, Chunhua Venture Capital, Inno Angel Fund, and Mizuki Tsinghua Alumni Seed Fund.

Meanwhile, VAST officially released Tripo Studio, the world's first AI-driven one-stop 3D workbench, and will soon launch the new algorithm Tripo 3.0. The financing will focus on the development of the Tripo series of large-scale models and the construction of Tripo Studio products and ecological platforms, accelerate the construction of the “AI+3D” entire industry chain, and build an end-to-end product system of “basic model + ecological plug-in + native workbench”, thus building a complete gradient of creator portraits covering professional (PGC producers), expert (PUGC creators) to popular (UGC users), further consolidating the team's leading position in the world of 3D generation.

Established in March 2023, VAST is an AI company focusing on the development of large general-purpose 3D models. It is committed to establishing a 3D UGC content platform by creating popular 3D content creation tools, making 3D-based space a core element of user experience upgrade, content expression innovation, and new quality productivity improvement.

Since the beginning of 2024, VAST has continued to iterate on large Tripo models, successively launched large 3D model series with multi-billion parameters such as Tripo1.0 to Tripo2.5, and also released 3D basic models widely recognized by the global open source community, such as TripOSR, TripOSG, and TripOSF, and developed a series of DCC software ecosystem plug-ins.

Currently, VAST has provided Tripo3D large model services to 2 million+ 3D creators, 20,000+ small to medium developers, and 700+ large enterprises around the world, generating nearly 30 million models.

Recently officially released, Tripo Studio is the world's first AI-driven one-stop 3D workbench. It received high praise from users around the world in the first week of closed beta: the platform's payment rate increased 2.5 times, and ARR R exceeded 3 million US dollars.

Through Tripo Studio and the upcoming Tripo3.0, as the world's top 3D model team, VAST will continue to contribute to technological breakthroughs in the field of 3D generation.

From “selling tools” to “delivering results”

In over 1,000 hours of in-depth interviews with global users, the VAST team met many creators with diverse backgrounds: a 76-year-old retired American art teacher, a Japanese WebAR charity creator, a Russian miniature model craftsman, an Israeli game full stack developer, a senior Disney concept artist, a former Blizzard art director...

Through the scenery outside the window from all over the world during video calls, and through the creative pipeline that pulls through screen sharing, these practices spanning the fields of gaming, film and television, XR, industrial design, public welfare art, etc., have continuously given the VAST team “use-case epiphany” moments.

VAST was happy for 3D newbies to complete their first model with Tripo, but it also clearly understood the technical boundaries from professional user feedback: if the initial model generated by Tripo is connected to an industrial pipeline, it still requires extensive manual topology adjustments, map calibration, and structural optimization.

This kind of user pain point of “coming with an idea but leaving with a semi-finished product” is the original purpose of Tripo Studio — what VAST wanted to solve was to enable every idea to move from a vague idea to a complete result, so that 3D creation no longer stops at model generation, but rather obtains complete, usable, and valuable interactive 3D content.

The VAST team designed new algorithms around the user's core experience and process, from generation to disassembly and editing, from stickers to bone binding, and launched Tripo Studio, the world's first AI-driven one-stop 3D workbench, which redefined the AI-driven 3D content creation process.

Intelligent component segmentation

Since the development of video and image generation models, orbital editing or layer operations are still not possible, but Tripo Studio is the first in the world to achieve accurate and controlled asset editing in 3D mode.

This breakthrough not only significantly improved creative efficiency, but also opened up huge commercial application space.

For 3D printing, Tripo's intelligent segmentation allows the model to be automatically disassembled according to components, and colors can be directly set for different regions in the slicing software. There is no need to manually cut the surface or reconstruct the topology, greatly improving the efficiency and accuracy of multi-color printing. For game developers, split model parts can be retained or replaced as needed, and they can quickly own a component-level asset library. This algorithm comes from HoloPart developed by VAST.

magic sticker brush

In traditional 3D creation processes, UV expansion and texture processing have always been key bottlenecks limiting efficiency. Especially in the field of game character modeling, dealing with sticker defects often takes a lot of time and money.

With the magic brush, users only need to circle the target area, and the system can quickly complete defect repair and maintain a high degree of consistency in texture style. If you use the “Smart Part Segmentation” function, you can modify certain parts individually — for example, change the character's face, directly select the “head” section, and gently paint the sticker to immediately “change the head”.

Intelligent low mode generation

In the digital world of real-time rendering, there is always a conflict between high-fidelity models and performance bottlenecks. Traditional optimization requires manual topology reduction, which not only takes time and effort, but also easily causes problems such as texture stretching and structural distortion.

“Intelligent low mode generation” pioneered the introduction of autoregressive models into the field of 3D generation. Through intelligent recognition of key model features, the number of surfaces can be plummeted from hundreds of thousands to thousands or even hundreds while retaining model details.

Everything is automatically bone-bound

When animating 3D characters, binding bones is always the most time-consuming part. This process has also faced a “digital divide” for a long time — basic skeletal binding alone takes animators 8-10 hours in traditional DCC tools such as Blender.

Behind Tripo Studio's bone binding function is the algorithm UniRig [link open source project] developed by VAST, which can automatically analyze the shape structure of the model to complete skeletal binding. For example, when entering a fish model, the system can accurately identify key movement dimensions such as fin swing axes and spine bending paths, quickly complete bone position estimation and skin allocation, and support export in mainstream formats such as FBX.

Most of Tripo Studio's core features, smart component segmentation, magic sticker brushes, intelligent low-mode generation, and automatic bone binding for everything originated from VAST Open Source Month in April.

Among them, the basic model TripOSF redefined the “upper limit of 3D generated model quality” and achieved breakthroughs in detail, complex structure, and computational scalability. The full experience of TripOSF was quickly presented in the studio in the form of Tripo3.0, which has 10 billion parameters.

Trust first, then see

Traditional 3D modeling requires multi-tasking collaboration, such as art, topology, binding, and stickers. Today, non-professional users can independently complete the entire process through Tripo Studio. From simply providing algorithms to using a complete workbench to undertake creative intentions, VAST expects Studio not to simply improve efficiency, but to break professional barriers through “process re-engineering” so that technology can truly serve the full implementation of creative intentions.

Smartphones use cameras to objectify the world, driving the logic of “images establish the world” — exporting images from reality. And VAST is engaged in another media revolution: deriving reality itself from images. Enabling more people to create and share 3D content has been VAST's original intention for a long time.

Over the past two years, Tripo has provided services to many partners, such as Tencent Games, NetEase Games, ByteDance, Alibaba, and Premium Choice. These companies have completed the full link from concept generation to production and application within the pipeline.

In terms of industrial design, VAST collaborates with many of the world's top household 3D printer teams, including Pioneering Bamboo, Vertical and Dimensional, and 3D, to provide creative space for mass makers around the world. In order to better restore the details of the generated model, VAST collaborates with the full-color 3D printing team to provide users with customized printing services at home.

In the field of architecture and space design, VAST collaborates with platforms such as Cool Kara and Homestyler to integrate AI generation solutions into the design pipeline to provide personalized home experiences for global households.

At the same time, VAST also sees more possibilities for 3D generation to become the gameplay itself: the highlight gameplay “Yanyun's Sixteen Sounds”, which was launched at the beginning of the NetEase game year, allows users to follow the virtual world and bring players a new interactive experience; the home system allows users to graffitize 3D furniture models and preview placement effects — 3D generation has gone from being an efficient tool in professional pipelines to become an “interactive engine” for the public to participate in digital creation.

In the future, VAST expects 3D content to move further towards “creation is play”, from behind the scenes of industrial production to the front desk of user interaction. It is no longer the designer's professional output behind closed doors, but rather a creative expression that naturally arises during user interaction.

Song Yachen, founder and CEO of VAST, said:

“For the past two years, we've been focusing on one thing: making technology a global SOTA; now, we're focusing on commercialization, engineering, and commercialization. Because we believe that 3D creators need not only a “model is a product” semi-finished product, but also a new AI-centered workflow, which allows 3D creators to maintain their control and creativity while improving efficiency. The team gathered a group of content consumers and creators, and we are also creating this product for ourselves. In the foreseeable future, we believe 3D's mass creator tools and content platforms will be created, and production relationships for 3D creation will change.”

The Beijing Artificial Intelligence Industry Investment Fund stated:

“We are optimistic about VAST's potential in the field of 3D model generation. As an organization that has long focused and invested in the field of AI generative models, we are pleased to see VAST continue to iterate and innovate in models and products over the past year. More than just a technology innovation company, VAST is a key force driving the development of the global 3D content creation field. With their deep understanding of AI technology and fearless innovative spirit, the young founding team successfully developed Tripo Studio, a revolutionary product. As a “3D Cursor,” the “3D Cursor” had the opportunity to drastically lower the technical threshold of 3D creation and continue to stimulate the potential of creators. Looking forward to the future, as VAST continues to deepen its layout in the “AI+3D” industry chain, we believe that in the future, 3D creation will be as popular and creative as today's photography, truly transforming from professional pipelines to public participation, and leading the industry to a wider world.”

Jingya Capital said:

“Jingya Capital is very happy to invest in VAST, and we are strongly optimistic about the market potential of 3D generation driven by GenAI technology. In terms of market space, real application scenarios such as games, film and television, industrial design, metaverse, and 3D printing are driving surging demand; at the time point, the 3D generation circuit is on the eve of an explosion, and breakthroughs in technology, data, and computing power will accelerate commercialization. We believe the next 5 years may reshape trillion-level markets such as gaming, industry, and metaverse.

In communicating with the VAST team led by Song Yachen, we deeply felt the team's accurate insight into user needs and were able to develop products based on customer needs and actual work processes. At the same time, the team has the world's top technical capabilities to continuously iterate and upgrade products. Its products have quickly completed commercial verification, won high recognition from many Prosumers and B-side companies, and has accumulated key data assets. We are confident of the VAST team's growth and leadership in 3D generation and look forward to their continued leadership in the industry.”