Understanding the Radiance Fields landscape with Michael Rubloff

Tue, May 21, 2024

Imagine a technology that blurs the line between the digital and physical worlds, transforming flat images into vivid three-dimensional experiences. This is the realm of Neural Radiance Fields (NeRF), a revolutionary visual technology that is redefining realism across various industries.

At Tryolabs, we are dedicated to not only staying at the forefront of knowledge and pushing boundaries but also to actively sharing our expertise. From our weekly internal knowledge sharing sessions to our "This Week In" AI initiative, rich blog content, and presentations at conferences - our team of experts is committed to sharing insights with the broader community.

In this blog post, we dive into an enlightening interview with Michael Rubloff, founder of radiancefields.com. Michael has collaborated with big companies such as NVIDIA and Shutterstock, to showcase the impact of Radiance Fields. He has also spoken at events like GTC 2024 (on “The Future of Extended Reality and Generative AI) and SIGGRAPH 2023. Together, we explore the genesis, applications, and future potential of NeRF and how it promises to revolutionize our interaction with digital content.

Michael Rubloff

From hobby to hyper-realism

Michael’s journey into the realm of Radiance Fields began somewhat serendipitously. He was engaged with Lidar scans (3D point clouds representing the surfaces of objects) during the pandemic, trying to capture real life into 3D. But discovering NeRF on social media marked a pivotal shift.


“I saw NeRF and realized pretty much instantly that that solved every problem I was trying to address with Lidar.” he recalls.

This revelation propelled him down what he describes as a "rabbit hole" of self-education and experimentation, leading to the acquisition of a powerful NVIDIA GPU and countless hours of trial and error.

“I actually ended up convincing my friend to lend me his computer for a little bit, which ended up turning into a ten month loan.” Michael said.

His initial fascination was rooted in a simple yet ambitious goal: to bring photography into three dimensions, transcending the flat plane of traditional images. This endeavor evolved from a personal challenge into a robust exploration of NeRF's potential, which Michael soon recognized as vast and largely untapped.

The birth of radiancefields.com

Frustrated by the lack of coverage of such a transformative technology, Michael decided to take matters into his own hands, launching what would eventually become radiancefields.com.

“I was thinking, this is an amazing technology, and here are these concrete examples of use cases for it, and nobody is writing about it.” he expressed. His platform not only showcases the extensive capabilities of NeRF (and newer Radiance Fields models) but also demystifies the technology, making it accessible to enthusiasts and professionals without the need for a deep technical background.

Understanding NeRF

The first NeRF model was proposed by Mildenhall et al. in their paper NeRF: Representing Scenes as Neural Radiance Fields for View Synthesis, which triggered mass research in this field. As explained in the original work:

We present a method that achieves state-of-the-art results for synthesizing novel views of complex scenes by optimizing an underlying continuous volumetric scene function using a sparse set of input views.

Our algorithm represents a scene using a fully-connected (non-convolutional) deep network, whose input is a single continuous 5D coordinate (spatial location (x, y, z) and viewing direction (θ, φ)) and whose output is the volume density and view-dependent emitted radiance at that spatial location.

We synthesize views by querying 5D coordinates along camera rays and use classic volume rendering techniques to project the output colors and densities into an image. Because volume rendering is naturally differentiable, the only input required to optimize our representation is a set of images with known camera poses. We describe how to effectively optimize neural radiance fields to render photorealistic novel views of scenes with complicated geometry and appearance, and demonstrate results that outperform prior work on neural rendering and view synthesis.

At its core, NeRF utilizes sparse standard 2D images to produce hyper-realistic 3D models. The process involves capturing multiple images around an object and then using a technique called 'structure from motion' to establish camera poses. These images are fed into a neural network, which meticulously renders a three-dimensional output, complete with view-dependent effects that mimic real-world lighting and reflections.

“NeRF understands how light should behave as you move around a scene,” Michael explained, highlighting the model’s ability to simulate the nuances of light and reflection with astonishing accuracy. This capability enhances visual realism and opens up new possibilities for applications in various industries.


Commercial and creative potential

The commercial implications of NeRF are profound. Michael cites the example of Google’s Immersive view, which uses NeRF in Google Maps for detailed visual representations of urban scenes. Another potential use appears in e-commerce, where consumers could view products in three-dimensional detail before purchasing. This level of detail could dramatically reduce return rates and increase consumer satisfaction by providing a more accurate representation of products.

Moreover, the artistic and creative prospects are equally compelling. Michael envisions radiancefields.com as a platform that explores these potentials and inspires others to consider how three-dimensional rendering can be integrated into various aspects of digital and physical experiences.

The future of NeRF and its integration with GenAI

Despite its numerous applications, NeRF is just one of many tools in the rapidly advancing field of digital rendering. Technologies like 3D Gaussian splatting and Instant NGP (Neural Graphics Primitives) offer alternative methods that might cater to different needs, such as higher frame rates or different rendering styles.

However, Michael believes that these technologies are complementary rather than competitive. He suggests that as these technologies evolve, they will likely converge, integrating their strengths to create even more powerful tools. This view offers a gentler perspective to the common tech narrative that one technology must inevitably replace another.

Looking ahead, Michael sees significant advancements in NeRF technology through its integration with Generative AI (GenAI). This synergy could drastically reduce the time and effort required for 3D modeling, allowing creatives to focus on more complex challenges and produce richer, more detailed virtual environments.

The combination of NeRF and GenAI might soon allow us to not just recreate small objects or scenes but potentially entire city blocks from simple text descriptions,” he speculated. These capabilities have the potential to disrupt industries ranging from virtual production to advertising, offering unparalleled flexibility and creative freedom.

Tools and accessibility

For those interested in exploring Radiance Fields, Michael recommends several tools catering to varying technical expertise levels.

For beginners, cloud-based platforms like Luma AI and Polycam offer a user-friendly introduction without requiring advanced hardware. For those with access to a GPU, Nerfstudio provides more hands-on control, allowing users to witness their virtual worlds come to life in real-time. In either case, those curious about this technology should seriously consider checking the Platforms section on radiancefields.com to understand which tools suit them best.

Examples generated with Polycam for Android using 111 input images.


In conclusion, NeRFs are a technological innovation and a paradigm shift in how we perceive and interact with digital content. As we stand on the brink of this new visual frontier, initiatives like radiancefields.com play a crucial role in advancing the technology and ensuring it is comprehensible and accessible to a broader audience. The journey of NeRF, as narrated by Michael Rubloff, is not just about technological achievement but also about the democratization of cutting-edge tools that redefine our visual and virtual landscapes.

His call to action is simple yet profound—experiment with capturing your surroundings using Radiance Fields technology. With the barriers to entry lowering each day, now is an exciting time to explore the potential of 3D modeling, whether for professional purposes or personal curiosity.

If this technology has caught your attention and you're interested in adapting it to fit your organization or business, we're ready to assist. At Tryolabs, we can help you ideate and execute innovative projects tailored to your specific needs.

Let’s work together to bring AI to your organization!

This website uses cookies to improve user experience. Read more