Shiva Viswanathan

01 404-GEN
02 Spectrogram
03 Dropmate
04 UTYPE
05 Keyblocks
06 Red Burns Fund


About
Resume
 



Shiva is a product and 3D designer bringing ideas to life and the intersection of digital design and physical products. 3D Designer at 404-GEN and currently based in Brooklyn NY. 


 

404-GEN

Role: 3D Designer, Marketing Designer, AI Technologist
Skills: Blender, Spline, Davinci Resolve, Touchdesigner  





404-GEN and Gaussian Splats

A Gaussian splat is a 3D visualization technique that represents objects using a collection of colored points (or "splats") rather than traditional polygons, allowing for detailed, lightweight models suited for real-time rendering and AI-generated content.

404-GEN is a technology studio focused on pioneering AI and web3 to create the largest open-source library of 3D models, with over 6 million assets. Their initiatives include integrating with Bittensor, a decentralized AI network, to enable scalable, collaborative model generation on web3, promoting accessibility and innovation in 3D content creation.

https://404.xyz/

https://doc.404.xyz/sn17-whitepaper





Animations created by me for Twitter/X made with 404-GEN


Tensorflow + Bittensor + XR

Gaussian splats tied to Bittensor are a cutting-edge, largely untested technology in XR applications, offering lightweight, AI-driven 3D models with potential for interactive experiences. My role involved exploring new use cases for these models, testing their feasibility in immersive environments, and developing marketing content to showcase their unique capabilities. This work aimed to position Gaussian splats as a pioneering tool in the XR space, highlighting their potential within decentralized, collaborative networks like Bittensor.





Scenes were created by me with models generated by 404-GEN



Creating Models + Expansion

In 404-GEN’s decentralized model creation pipeline, miners and validators work together to generate and assess 3D models. Miners create Gaussian splat models, while validators ensure quality by scoring each model based on accuracy and visual fidelity. My role involved testing validators and cross-referencing validation scores with the incentive curve, optimizing for higher-quality models by refining the feedback loop and maximizing rewards for precise model generation.

Happy Imp: Left validation score = .46 | Right validation score = .71 generated by 404 | 1.0 = perfect 



Public Releases 

To boost discovery and adoption of 404-GEN's Blender add-on, I led initiatives to increase its visibility and encourage user engagement. Since product adoption directly impacts the company’s profitability through web3 monetization, I contributed by designing and launching their public-facing Gumroad page and Unity Store page, making the add-on accessible to a wider audience. I also helped design their Discord bot, streamlining user interactions and fostering community engagement around the add-on.

https://404gen.gumroad.com/





Imagery from Gumroad and Blender Market storefront



Imagery from Unity Store page


Marketing for Public Releases

To support the release of our public storefronts on Gumroad and Unity Store, we launched a series of promotional videos showcasing the power and versatility of Gaussian Splat models. These videos highlighted real-world applications and ease of use for artists and developers, driving interest and engagement around the add-on.






Engagement and Education

I created a series of videos and tutorials to guide users through the installation and creative potential of Gaussian Splat models. These resources demonstrated unique applications in art and design, offering step-by-step instructions to help artists and developers integrate the models into their workflows. This content aimed to make the technology accessible and inspire creative exploration within the community.





1-4: From Twitter/X to announce discord and blender addon releases.
5: Video tutorial and User Guide for installing plugin and usage
6: Demo and showcase of a 3D scene built with 404-GEN models.


Potential + Experiments

Since these models are Gaussian splats generated with Blender's geometry nodes, they offer a unique foundation for playful, experimental animations. Built on a point cloud geometry structure, these models enable dynamic transformations, like morphing shapes or flowing movement, that go beyond traditional 3D models. This flexibility opens up new creative possibilities, allowing artists to explore animations that leverage the fluidity and adaptability of point-based geometry.


Left: Exploring methods of scaling and offsetting the points.
Right: Experiments with proximity to reveal the points and animation techniques with them.


Content Production

I created a range of videos to showcase the potential of 404-GEN's products, pushing emerging technologies to their limits while exploring practical use cases. Through these videos, I not only marketed the capabilities of our Gaussian Splat models but also rigorously tested and reported on their applications, sharing key insights with the public. This approach highlighted the technology’s creative versatility and opened the door for broader adoption.

Every single video below was made using a 404-GEN model or using one of the splats as a foundation for the mesh.






  

Takeaways and Next Steps

My work at 404-GEN combined technical exploration with strategic marketing to drive the discovery, usability, and adoption of cutting-edge Gaussian Splat models. By creating tutorials, testing product features, and launching storefronts, I contributed to expanding 404-GEN’s reach and showcasing the versatility of their tools to the creative community. This experience taught me the value of bridging emerging technology with user-centered design, pushing boundaries to make complex innovations accessible and inspiring for artists and developers alike.


Few of the key lessons learned:

  • Push limits and explore creativity.
  • New technologies require even more powerful tech to run it. My computer crashed multiple times while working with these projects.
  • Spread the word as fast as possible to increase adoption and let users innovate on the technology.




Thank you to the 404-GEN Team

Ben James
Monica Vasandani
Ruben Hohndorf
Maxim Bladyko
Alexander Tereshin
Rodion Stetsurin
“Sheesh”
Robin Simons