NVIDIA has published a paper greatly helping detail recovery thanks to Generative AI.

HTGS gets rid of popping artifacts in 3DGS and more!

Another novel Radiance Field representation is here with interesting capabilities.

A new Radiance Field representation has been published by the Technical University of Munich.

A GitHub repo from Pablo Vela has integrated GLOMAP with HLOC.

3D Convex Splatting is another new way to render a Radiance Field. It is not Gaussian Splatting.

This GitHub repository is making it easy to convert 3DGS to dense point clouds.

The original implementation is receiving major training speed boosts.

Rendering speeds are getting a big boost, with some pesky features addressed.

Another Ray Tracing Radiance Field emerges, this time from Google.

Gaussian Splatting research might be speeding up, thanks to a new contribution from the first author of 3DGS.

The free Gaussian Frosting for Blender is making it easy to edit & animate.

CMU and Google unveil a new approach to Radiance Caching that reduces bias and improves rendering quality.

This method from the 3DGS MCMC team pushes NeRF rendering rates well into real time and up to 500fps.

Have you ever spent time at a museum, carefully examining how light filters through a marble statue or observing the soft glow that radiates from a wax candle?

Gaussian Splatting is so Julia.

A second paper examining Ray Tracing Gaussians, instead of Splatting has emerged.

We speak to kidachi, whose GaussMR is currently on display at SIGGRAPH

A criticism of NeRFs is their rendering rates. Quietly a couple of papers have been published over the last two months which push NeRFs into real time rates.

It significantly improves the fidelity of reflections and other view-dependent effects, making scenes look more realistic.

Today, things are taking an exciting step forward with the introduction of 3D Gaussian Ray Tracing (3DGRT).

Research topics on relighting continue to increase sharply.

3DGS-DR steps up reflection fidelity, while remaining viable on consumer GPUs

There are a tremendous number of applications where relighting will greatly aid the use of NeRFs in practical applications.

Have you ever captured a dataset only to realize that your camera was set to manual mode, causing exposure to fluctuate?

Over the last month, we have repeatedly seen methods that have been exploring Gaussian Splatting to increase fidelity and surface reconstruction. These have ranged from 2DGS, to Surfels, and more. Today, we’re looking at a slightly different take on the problem with AtomGS. 

If you've interacted with Gaussian Splatting, you might have noticed an issue described as "popping," where details appear to flicker as you move through a scene.

Late last year we looked at Uni-SDF which introduced dual radiance fields to better represent reflections in a scene. However, I just happened to see NeRF-Casting on Github a little while ago

We very recently were looking at RealmDreamer, which generates scenes from prompts. Just over a month later, CAT3D, short for "Create Anything in 3D," has emerged and takes things up a notch or two.

Tri-MipRF was one of the more underrated NeRF papers to be released last year. Now we're seeing a progression of the work Tri-Mip created with Rip-NeRF.

SMERF, or Streamable Memory Efficient Radiance Fields was originally announced late last year from Google Research, to great excitement.

As SIGGRAPH 2024 approaches, some of the early papers have started to surface.

If you think you've seen a large-scale NeRF before, have you seen a 25km capture, spread across 260,000 images?

NVIDIA's recently announced RefFusion, however, takes a different approach by employing Gaussian Splatting.

Gaussian Splatting has quickly become one of the most exciting research topics in Radiance Fields, thanks to its fast training, real time rendering rates, and easy to create pipeline. The one critique that emerged was the resulting file size from captures, often venturing into the high hundreds of megabytes and up.

Since the unveiling of the Sora's large-scale generative Radiance Fields, the tech world has been buzzing with anticipation about the future of 3D scene generation. There hasn't been much public work since then showcasing what could be coming, but today we're looking at RealmDreamer, which creates scene level generations based on original text prompts.

Playing as yourself in a video game has always seemed like a fun idea. Now, we're one step closer to making that a reality with PhysAvatar.

Gaussian Splatting methods have continued to pour in over the first three months of the year. With the rate of adoption, being able to merge and compare these methods, shortly after their release would be amazing.

There have been two recurring questions that researchers have been working towards: stronger outputs with less data and speed. COLMAP is usually unreliable and fragile under sparse-view settings with insufficient matched features.

There has almost always been a demand to pull meshes out of radiance fields. When Antoine Guédon and Vincent Lepetit first released SuGaR,

Excitingly, we're seeing the arrival of the first result of another widely hyped event. The meeting of NeRFs and Gaussian Splatting.

Traditionally, when we discuss radiance fields, we focus on radiance itself. However, today there's a subtle yet significant shift towards radiation—yes, that kind of radiation.

Researchers from Aalto University and Naver Labs Europe have introduced DUSt3R, a groundbreaking method for Dense Unconstrained Stereo 3D Reconstruction, propelling forward the field of geometric 3D vision.

The newly announced VastGaussian project introduces a new approach to high-quality reconstruction and real-time rendering of large scenes, showcasing some of the most extensive radiance fields to date.

The first six weeks of 2024 have seen remarkable advancements in the realm of converting prompts to detailed 3D models.

The pursuit of making Gaussian Splatting editable in real-time, without compromising its renowned high visual fidelity, represents an exciting topic in computer graphics research.

There has been a surge in interest in VR based radiance field methods since the release of the Apple Vision Pro. I admittedly spent significantly more time in VR than I had alotted myself, contributing to this article being published a few days after its release.

Diving into a world where reality can take on the aesthetics of imagination, Geometry Transfer for stylizing radiance fields (Geo-SRF) showcases a glimpse, offering artists and technologists alike the whimsical power to reshape the very fabric of 3D spaces.

The recent release of a groundbreaking technology has sparked considerable excitement across the tech community. I am of course referring to the code release of Trilinear Point Splatting for Real Time Radiance Field Rendering, otherwise known as TRIPS.

When physics based radiance fields landed last year, they certainly made a splash with PhysGaussian and PieNeRF. Now, a follow up paper is taking it up a notch with Gaussian Splashing.

Last year, we saw the introduction of the first semantic based NeRF papers such as LERF, which expanded into LERF TOGO. Now we are getting another paper from the same group of authors, with GARField. GARField stands for Group Anything with Radiance Fields and it lives up to the name.

Quietly over the weekend, underwater NeRF method, SeaThru NeRF has been added as an officially supported method to nerfstudio. We already know that radiance fields can model exactly what they see, but we typically imagine that the locations are your everyday interiors or exteriors. But what happens when that place is underwater?

Several of the most compelling radiance field papers and methods have been released by Google over the last year. With each paper release, there's often a clamoring for their release.

Gaussian Splatting has been the hottest radiance field method for the past four months, since SIGGRAPH. Well, it seems like we might have a new radiance field method called TRIPS, Trilinear Point Splatting for Real-Time Radiance Field Rendering that might have something to say about that.

Back in middle school, I really thought that more of my life would involve decoding hidden messages in images for some reason. With Noise-NeRF, we're seeing that it's possible and it actually works pretty well.

With the rise of editable NeRF methods last year, such as Instruct NeRF2NeRF and InpaintNeRF 360, the industry has been eagerly anticipating advancements in 3D scene editing.

Shortly after SIGGRAPH in late August, I began having conversations about the place of NeRFs in Ecommerce. The almost universal feedback I received was that the file sizes were still too large and to come back when they approach roughly a megabyte.

UniSDF comes on the heels on a few advancements to create better geometry out of radiance field methods, specifically NeRFs.

Human Gaussian Splats (HUGS) from the Apple Machine Learning team is designed to create lifelike, animatable human avatars within a dynamic scene, is groundbreaking in its efficiency and detail. It's over 100 times faster AND it can also be incorporated directly into NeRF based scenes as well.

For the long time readers of this site, earlier this year, we looked into Google Research's Memory Efficient Radiance Fields (MERF). Now, they're back with another groundbreaking method: Streamable Memory Efficient Radiance Fields, or SMERF.

Google Research has been on fire lately. They just recently came out with Reconfusion and now joining a very busy year for the Google team is Nuvo.

In the evolving landscape of robotics and vision systems, the quest for a method that not only navigates but also reconstructs unknown environments with precision and efficiency has been relentless.

In the evolving landscape of Neural Radiance Fields (NeRFs), each new development brings us closer to overcoming the challenges of rendering lifelike, scalable digital worlds.

If you've been trying out Gaussian Splatting, you may have noticed that the scene tends to degrade quickly, especially when you pull outside of the capture path or change the focal length, compared to NeRF.

GaussianEditor emerges as an exciting new Gaussian Splatting editor, allowing for more precise editing and inpainting. At its core, the tool integrates two groundbreaking techniques: Gaussian Semantic Tracing and Hierarchical Gaussian Splatting (HGS).

Just when we thought the week couldn't get more exciting in the world of physics-based radiance fields, we're introduced to another groundbreaking development: PIE-NeRF (Physics-based Interactive Elastodynamics with Neural Radiance Fields).

I knew the second I saw jam spreading itself across a piece of toast, that this was going to be a good paper. Gaussian Splatting has had a tremendous amount of functionalities added with great speed and today we're seeing an exploration into a new area surrounding real world physics.

There are a lot of eyes on NVIDIA's Neuralangelo that just this past week was named one of Time Magazine's Top Inventions of 2023. Instant-Angelo comes from CUHK-Shenzhen MPhil Student, Chongjie Ye.

In early September, we took at a look at Improving NeRF Quality by Progressive Camera Placement by Georgios Kopanas and George Drettakis from Inria.

Jonathon Luiten, author of Dynamic 3D Gaussians released the code this morning in a post on Twitter.

With ICCV behind us and the ton of NeRF papers that were shown, one seriously caught my eye. It was one of the finalists for best paper (along with Zip NeRF) and for a good reason. The paper is called Tri-MipRF; now let me tell you why.

NeRFs have revolutionized the realm of photorealism, but they've historically had a glaring limitation: the inability to modify lighting.

Catching up with my sister has been an exercise in bridging distances. She recently moved to Copenhagen, trading the familiar landscapes of our shared childhood for the charming streets of the Danish capital.

While there was the exciting news on the platform side, with Polycam implementing Gaussian Splatting, we have an equally exciting research paper, NeuRBF, that greatly increases the quality of NeRFs.

Continuing the recent trend of papers looking to optimize neural radiance fields (NeRFs) is a team from the Technical University of Munich and Meta Reality Labs Zurich.

It's no secret there are a large amount of use cases for NeRFs. Recently, I took a look at Infinite Realities and examined some of human based ones.

Two authors of the wildly popular 3D Gaussian Splatting paper are back and have released a new paper, but about NeRFs?

Summer is coming to a close and it seems as though, the team at Google went away to CamP. No, this is not a summer camp designed around NeRF, as fun as that would be, but it is the next best thing! CamP stands for Camera Preconditioning for Neural Radiance Fields.

Recently, we've seen Google publish blogs on implementing NeRFs into Google Maps and Immersive View. At the end of their post, they gave a sneak peak of something—an outdoor scene.

I often liken NeRFs to magic, in that they can understand view dependent effects. That becomes a bit more literal with today's featured paper: Magic NeRF Lens.

Accurate 3D scene and object reconstruction is crucial for various applications such as robotics, photogrammetry, and AR/VR. NeRFs has been successful in synthesizing novel views but falls short in accurately representing the underlying geometry.

By traditional means, the storage sizes of the NeRF do not appear to be huge, like some new datasets, but they do have wider implications.

We've continued to get higher fidelity NeRFs in the last few months, but the trade off has always been sacrificing frame rate or being more computationally intensive to run.

In a nutshell, InpaintNeRF-360 allows for the removal or editing of objects within a unbounded 360 degree NeRF.

3D reconstructions of humans have been notoriously tough because of the large amount of minute details. Additionally, we have a very low tolerance when something looks out of place.

One of the big differentiators of NeRF compared to existing methods is the ability to process reflections.

Even though the name suggests it, no this doesn't have anything to do with NeRF streaming.

I have received several questions about whether or not NeRFs are able to generate a point cloud and how the two concepts correlate with one another.

A groundbreaking new method called Neural Fields for LiDAR (NFL) has been developed to optimize a neural field scene representation from LiDAR measurements.

When NVIDIA first announced Maxine and the ability to have eye contact be redirected on a video call, my own eyes nearly popped out of my skull.

Over the last six months, we have seen more and more papers come in that will help power real time rendering of NeRFs and Radiance Fields.

For the last year+, Instant NGP has been seen as the fastest local training NeRF method that would run on a consumer grade GPU.

Researchers from the National University of Singapore and Tencent BCG Business School have developed a new method called HOSNeRF (Human-Object-Scene Neural Radiance Fields) that can create 360° free-viewpoint renderings of dynamic scenes with human-environment interactions from just a single video.

Over the past few weeks, we have seen a few papers pushing image segmentation forward, but today we get one specific to NeRFs.

NeRFs off the bat bake in the lighting and while creators such as Jonathan Stephens have explored relighting a scene through Instant-NGP, today there is another paper that not only allows for Novel View Relighting, but also extends to Virtual Object Insertion.

The new NeRF papers continue to pour in. This time with the introduction of F2-NeRF (Fast Free NeRF).

Stop me if this has happened to you before. You've gone out, shot a NeRF, processed the transform and launched the GUI only to realize, it's full of artifacts and floaters.

From Google to Preimage, there has been a lot of recent hype surrounding large scale NeRFs. Each have their own method and today another one has been introduced.

With GTC and GDC in the immediate future, we have seen an increase of consequential NeRF papers get published.

Rounding out the busy month of February for NeRFs, another paper has been announced by Google Research.

Almost coincidentally, just hours after Luma AI announced in web browser realtime rendering, a new method of radiance fields has been published, Memory-Efficient Radiance Fields or MERF.

The team from Meta AI have released their paper MAV3D, a revolutionary new technology that can create stunning three-dimensional videos from just a simple text description.

Cornell Tech and Google Research have released a new paper that addresses the challenges of moving subjects due to motion, camera paths, and video length.