Introduction by Nanji Bear: For a long time, 3D printing enthusiasts and creators have faced the pain points of limited model sources and high production thresholds, making it difficult to quickly transform inspiration into printable products. To address this issue, Ren Lifeng, a member of the founding team of Douyin, established “Shumei Wanshu”, which has currently completed a total financing of 500 million RMB. The series of products launched by the company covers the entire process from creativity to production, aiming to empower designers and creators with AI and enhance the overall efficiency from design to production.
On September 15, 2025, industry friends informed Nanji Bear that Tuozhu has integrated the Hitem3D API model from Shumei Wanshu into their community Makerworld. With Hitem3D, users only need to upload an image or provide three views to generate a high-definition 3D model that is ready for printing, with almost no additional model repair required. 3D enthusiasts who have learned about or used Hitem3D can quickly realize that a new turning point is approaching.
△ The tool page makerlab in the Makerworld community has integrated Hitem3D models
From Creator Tools to Industry Interfaces
Compared to previous AI tools that could only generate models with a resolution of 1024^3 and blurry details, Hitem3D directly increases the resolution to 1536^3, combined with geometric restoration and texture mapping capabilities, for the first time allowing AI-generated models to truly meet production and manufacturing standards.
The team at Shumei Wanshu recognized early on that while 2D raw images could meet users’ basic creative needs, one of the key steps in transforming a creator’s inspiration from an image into a tangible product, even into a commodity, is the capability of producing commercial-grade 3D models. Therefore, from day one of their startup, they chose the path of self-developing 3D generation models.
This choice has proven to be correct. Through independently designed sparse convolution networks and 3D VAE, Hitem3D achieves nearly lossless three-dimensional data compression and reconstruction, reducing the Chamfer Distance metric by 40% compared to traditional models. This technological breakthrough means that the generated models can almost be directly put into the production process.
Hitem3D v1.5: Faster, Higher Definition, and More Realistic 3D Generation Tool
In the field of 3D printing, in addition to high-resolution models, sometimes the generation speed is also one of the factors affecting user experience. In the paper on the new algorithm ultra3d model published by Shumei Wanshu, we also see progress in this key factor. Traditional AI modeling often takes several minutes or even longer, while Ultra3D can shorten the generation time to just a few minutes.
Moreover, the release of Hitem3D v1.5 has added a 512 mode compared to the 1.0 model, further reducing the model generation time to less than one minute. Even with the 512 model, the details of the generated model are still very sufficient, making it very user-friendly for printing enthusiasts.
For creators, this means that inspiration can be materialized in real-time; for corporate clients, it means a significant improvement in sampling efficiency. A studio manager engaged in figure design shared that in the past, it took them 4-5 days to create and refine a mecha prototype, but with Hitem3D, if using the normal 512 mode, they can complete the model in under a minute, suitable for ordinary sampling references. If using the highest resolution 1536pro, although it takes 12-15 minutes, the time saved in post-model repair can be 1-2 days, and only artistic modifications are needed before production. “We shift our focus from ‘basic modeling and spending a lot of time on fine-tuning’ to ‘creative expression’, which is the tool that designers should have.”
The manager particularly appreciates the geometric capabilities of Hitem3D v1.5: “The latest v1.5 1536pro model has excellent geometric performance; even without three views, the back of the object is restored much better than competitors. I can only say that Hitem3D absolutely kills the competition.”
The algorithm team at Shumei Wanshu explained to us, “The 1536^3pro resolution has achieved significant improvements in geometric accuracy and texture capabilities, not only generating more detailed structures but also getting closer to real material in texture performance, providing higher quality raw materials for industrial design and creative production.”
Other Model – 1
Other Model – 2
Hitem3D v1.5 512^3
Hitem3D v1.5 1536^3
Version 1.5 has also been specifically optimized for input scenarios such as half-body and close-up photos, allowing for more accurate restoration of facial features and details. Our team ran several tests, and the results were indeed stunning, with very accurate restoration of facial proportions and positions.
△ Input Image (Left) Hitem3D Portrait Model Comparison Effect (Right)
Validation of Popularity in Overseas Communities
The potential of this technology is not just theoretical. After Shumei Wanshu uploaded the algorithm model Sparc3D in the development stage to the Hugging Face community in mid-June, it quickly topped the space trend list within a week and remained in the top three for three consecutive weeks. On YouTube and X platforms, many 3D industry users have published reviews, with some users calling it “the best 3D model generator of 2025.”
After the release of version 1.5, many overseas users have shared the effects of the models generated using Hitem3D. An independent game developer from the United States showcased a character model generated with Hitem3D, stating, “The level of detail is close enough to be directly imported into the engine, saving a lot of initial modeling time”; another figure blogger generated a model from an image of their favorite character, exclaiming, “Even the texture of the clothing has been restored.” This cross-regional word-of-mouth effect opens up the imagination for Hitem3D in the global market.
Ecological Expansion Beyond Technology
Although geometric restoration capability is Hitem3D’s strong suit, the “multi-view generation” feature undoubtedly brings Hitem3D closer to industrial application scenarios. Designers only need to provide front, side, and back views, and the system can automatically stitch together to generate a complete model. Combined with the existing high-definition geometric restoration capability, this shows great potential in higher precision design needs such as automotive parts and industrial prototypes.
At the same time, the collaboration with Tuozhu is just the beginning. In the Makerlab module of Tuozhu Makerworld, more manufacturers and individual developers can experience the capabilities of Hitem3D, thereby building a richer upstream and downstream ecosystem. This is not just a “creation tool” but more like an emerging AI 3D Creative Operating System.
The “Express Lane” from Creativity to Production
The logic behind this is clear: the market for AI creation is transitioning from experimental to productive, and Hitem3D is opening up the express lane from creativity to production lines. Whether for personal hobbyists’ figure printing, small and medium enterprises’ product sampling, or asset production in the gaming and film industries, Hitem3D is gradually becoming an efficient foundational tool. The collaboration with Tuozhu brings this capability to the most vibrant 3D printing community.
“We hope to use AI to enable more people to possess the creativity to go from 0 to 1,” summarized Zhang Bo, product manager of Shumei Wanshu. “When creativity becomes three-dimensional, the efficiency of the entire industry chain will be redefined.”
Industry insiders point out that the upgrade of Hitem3D v1.5 will greatly expand the application boundaries of Hitem3D: from figures, prototypes, and components to virtual humans and digital images, all will benefit from this. Shumei Wanshu has stated that in the future, Hitem3D will maintain a rapid product iteration pace, continuously adding specialized features for different scenarios while maintaining high definition and efficiency, helping more creators and industry users achieve a seamless connection from “creativity” to “reality.”