Загрузка видео...

Не удалось загрузить видео

На главную

EDGS: Eliminating Densification for Efficient Convergence of 3DGS Contributions: • We show that initial triangulation based on 2D correspondences can replace the incremental refinement process, fundamentally changing how 3DGS models allocate resources. • Our method reduces the path each Gaussian must travel in parameter space. Careful initialization not only...

janusch_patas's profile picture

MrNeRF

16,811 subscribers

124,076 просмотров • 1 год назад •via X (Twitter)

Здоровье и велнесНаука и технологииОбразование
LIVE

Anya Rossi• Live Now

Private livecam show

Комментарии: 0

Нет доступных комментариев

Здесь появятся комментарии из оригинального поста

Похожие видео

Human Hair Reconstruction with Strand-Aligned 3D Gaussians Contributions (cited): – We propose a new 3D line lifting scheme that uses a modified 3DGS reconstruction technique to lift 2D orientation maps into a 3D field while also providing refinement of the camera parameters; – We introduce a dual representation of hair strand polylines and 3D Gaussians to achieve differentiable rasterization of hair strands and leverage photometric constraints for strand-based hair reconstruction; – Based on these components, we propose a coarse-to-fine optimization method for prior-guided hair reconstruction that leverages both latent and explicit representations of the hairstyle.
0:13

Human Hair Reconstruction with Strand-Aligned 3D Gaussians Contributions (cited): – We propose a new 3D line lifting scheme that uses a modified 3DGS reconstruction technique to lift 2D orientation maps into a 3D field while also providing refinement of the camera parameters; – We introduce a dual representation of hair strand polylines and 3D Gaussians to achieve differentiable rasterization of hair strands and leverage photometric constraints for strand-based hair reconstruction; – Based on these components, we propose a coarse-to-fine optimization method for prior-guided hair reconstruction that leverages both latent and explicit representations of the hairstyle.

MrNeRF

106,525 просмотров • 1 год назад

Many 3D generators output Gaussian Splats (3DGS) for fast rendering, flexible deployment, and high visual fidelity. Static 3DGS aren't world models (no dynamics/semantics) but a true world model must allow distilling 3D-consistent representations for any given time step (3DGS/meshes). This post-distillation serves a dual purpose: 1) validates physical consistency of the model. 2) extracting explicit representations avoids continuously running a heavy generator, thus saves compute and facilitates real-time interaction.
0:19

Many 3D generators output Gaussian Splats (3DGS) for fast rendering, flexible deployment, and high visual fidelity. Static 3DGS aren't world models (no dynamics/semantics) but a true world model must allow distilling 3D-consistent representations for any given time step (3DGS/meshes). This post-distillation serves a dual purpose: 1) validates physical consistency of the model. 2) extracting explicit representations avoids continuously running a heavy generator, thus saves compute and facilitates real-time interaction.

Matthias Niessner

26,226 просмотров • 4 месяцев назад

Self-Calibrating Gaussian Splatting for Large Field of View Reconstruction Note: Check below for full video. Abstract (cited): "In this paper, we present a self-calibrating framework that jointly optimizes camera parameters, lens distortion, and 3D Gaussian representations, enabling accurate and efficient scene reconstruction. Our technique is particularly effective for high-quality scene reconstruction from large field-of-view (FOV) imagery taken with wide-angle lenses, allowing the scene to be modeled from a smaller number of images. We introduce a novel method for modeling complex lens distortions using a hybrid network that combines invertible residual networks with explicit grids. This design effectively regularizes the optimization process, achieving greater accuracy than conventional camera models. Additionally, we propose a cubemap-based resampling strategy to support large FOV images without sacrificing resolution or introducing distortion artifacts. Our method is compatible with the fast rasterization of Gaussian Splatting, adaptable to a wide variety of camera lens distortions, and demonstrates state-of-the-art performance on both synthetic and real-world datasets."
0:16

Self-Calibrating Gaussian Splatting for Large Field of View Reconstruction Note: Check below for full video. Abstract (cited): "In this paper, we present a self-calibrating framework that jointly optimizes camera parameters, lens distortion, and 3D Gaussian representations, enabling accurate and efficient scene reconstruction. Our technique is particularly effective for high-quality scene reconstruction from large field-of-view (FOV) imagery taken with wide-angle lenses, allowing the scene to be modeled from a smaller number of images. We introduce a novel method for modeling complex lens distortions using a hybrid network that combines invertible residual networks with explicit grids. This design effectively regularizes the optimization process, achieving greater accuracy than conventional camera models. Additionally, we propose a cubemap-based resampling strategy to support large FOV images without sacrificing resolution or introducing distortion artifacts. Our method is compatible with the fast rasterization of Gaussian Splatting, adaptable to a wide variety of camera lens distortions, and demonstrates state-of-the-art performance on both synthetic and real-world datasets."

MrNeRF

17,206 просмотров • 1 год назад

Official Launch of the MrNeRF 3DGS Bounty 2: We're offering 🏆 $1600 + $500 bonus for improving initialization & training without densification for 3D Gaussian Splatting! RT & tag friends who might crush this. Details in thread 👇
0:12

Official Launch of the MrNeRF 3DGS Bounty 2: We're offering 🏆 $1600 + $500 bonus for improving initialization & training without densification for 3D Gaussian Splatting! RT & tag friends who might crush this. Details in thread 👇

MrNeRF

35,133 просмотров • 11 месяцев назад

Procedural Gaussian Splats in PlayCanvas 🔥 Everything becomes splats: Images. Text. Lines. One unified render, seamlessly blending with 3DGS models.
0:17

Procedural Gaussian Splats in PlayCanvas 🔥 Everything becomes splats: Images. Text. Lines. One unified render, seamlessly blending with 3DGS models.

Martin Valigursky

19,026 просмотров • 5 месяцев назад

DroneSplat: 3D Gaussian Splatting for Robust 3D Reconstruction from In-the-Wild Drone Imagery Abstract: Drones have become essential tools for reconstructing wild scenes due to their outstanding maneuverability. Recent advances in radiance field methods have achieved remarkable rendering quality, providing a new avenue for 3D reconstruction from drone imagery. However, dynamic distractors in wild environments challenge the static scene assumption in radiance fields, while limited view constraints hinder the accurate capture of underlying scene geometry. To address these challenges, we introduce DroneSplat, a novel framework designed for robust 3D reconstruction from in-the-wild drone imagery. Our method adaptively adjusts masking thresholds by integrating local-global segmentation heuristics with statistical approaches, enabling precise identification and elimination of dynamic distractors in static scenes. We enhance 3D Gaussian Splatting with multi-view stereo predictions and a voxel-guided optimization strategy, supporting high-quality rendering under limited view constraints. For comprehensive evaluation, we provide a drone-captured 3D reconstruction dataset encompassing both dynamic and static scenes. Extensive experiments demonstrate that DroneSplat outperforms both 3DGS and NeRF baselines in handling in-the-wild drone imagery.
0:19

DroneSplat: 3D Gaussian Splatting for Robust 3D Reconstruction from In-the-Wild Drone Imagery Abstract: Drones have become essential tools for reconstructing wild scenes due to their outstanding maneuverability. Recent advances in radiance field methods have achieved remarkable rendering quality, providing a new avenue for 3D reconstruction from drone imagery. However, dynamic distractors in wild environments challenge the static scene assumption in radiance fields, while limited view constraints hinder the accurate capture of underlying scene geometry. To address these challenges, we introduce DroneSplat, a novel framework designed for robust 3D reconstruction from in-the-wild drone imagery. Our method adaptively adjusts masking thresholds by integrating local-global segmentation heuristics with statistical approaches, enabling precise identification and elimination of dynamic distractors in static scenes. We enhance 3D Gaussian Splatting with multi-view stereo predictions and a voxel-guided optimization strategy, supporting high-quality rendering under limited view constraints. For comprehensive evaluation, we provide a drone-captured 3D reconstruction dataset encompassing both dynamic and static scenes. Extensive experiments demonstrate that DroneSplat outperforms both 3DGS and NeRF baselines in handling in-the-wild drone imagery.

MrNeRF

21,346 просмотров • 1 год назад

Gaussian Head Avatar: Ultra High-fidelity Head Avatar via Dynamic Gaussians paper page: Creating high-fidelity 3D head avatars has always been a research hotspot, but there remains a great challenge under lightweight sparse view setups. In this paper, we propose Gaussian Head Avatar represented by controllable 3D Gaussians for high-fidelity head avatar modeling. We optimize the neutral 3D Gaussians and a fully learned MLP-based deformation field to capture complex expressions. The two parts benefit each other, thereby our method can model fine-grained dynamic details while ensuring expression accuracy. Furthermore, we devise a well-designed geometry-guided initialization strategy based on implicit SDF and Deep Marching Tetrahedra for the stability and convergence of the training procedure. Experiments show our approach outperforms other state-of-the-art sparse-view methods, achieving ultra high-fidelity rendering quality at 2K resolution even under exaggerated expressions.
0:12

Gaussian Head Avatar: Ultra High-fidelity Head Avatar via Dynamic Gaussians paper page: Creating high-fidelity 3D head avatars has always been a research hotspot, but there remains a great challenge under lightweight sparse view setups. In this paper, we propose Gaussian Head Avatar represented by controllable 3D Gaussians for high-fidelity head avatar modeling. We optimize the neutral 3D Gaussians and a fully learned MLP-based deformation field to capture complex expressions. The two parts benefit each other, thereby our method can model fine-grained dynamic details while ensuring expression accuracy. Furthermore, we devise a well-designed geometry-guided initialization strategy based on implicit SDF and Deep Marching Tetrahedra for the stability and convergence of the training procedure. Experiments show our approach outperforms other state-of-the-art sparse-view methods, achieving ultra high-fidelity rendering quality at 2K resolution even under exaggerated expressions.

AK

65,834 просмотров • 2 лет назад

Wonderland: Navigating 3D Scenes from a Single Image Contributions: • First, we introduce a representation for controllable 3D generation by leveraging the generative priors from camera-guided video diffusion models. Unlike image models, video diffusion models are trained on extensive video datasets. This enables them to capture comprehensive spatial relationships within scenes across multiple views and embed a form of "3D awareness" in their latent space, which allows us to maintain 3D consistency in novel view synthesis. • Second, to achieve controllable novel view generation, we empower video models with precise control over specified camera motions. We introduce a novel dual-branch conditioning mechanism that effectively incorporates desired diverse camera trajectories into the video diffusion model. This enables expansion of a single image into a multi-view consistent capture of a 3D scene with precise pose control. • Third, to achieve efficient 3D reconstruction, we directly transform video latents into 3DGS. We propose a novel latent-based large reconstruction model (LaLRM) that lifts video latents to 3D in a feed-forward manner. With this design, during inference, our model directly predicts 3DGS from a single input image, effectively aligning the generation and reconstruction tasks—and bridging image space and 3D space—through the video latent space. Compared with reconstructing scenes from images, the video latent space offers a 256× spatial-temporal reduction while retaining essential and consistent 3D structural details. Such a high degree of compression is crucial, as it allows the LaLRM to handle a wider range of 3D scenes within the reconstruction framework, with the same memory constraints.
0:12

Wonderland: Navigating 3D Scenes from a Single Image Contributions: • First, we introduce a representation for controllable 3D generation by leveraging the generative priors from camera-guided video diffusion models. Unlike image models, video diffusion models are trained on extensive video datasets. This enables them to capture comprehensive spatial relationships within scenes across multiple views and embed a form of "3D awareness" in their latent space, which allows us to maintain 3D consistency in novel view synthesis. • Second, to achieve controllable novel view generation, we empower video models with precise control over specified camera motions. We introduce a novel dual-branch conditioning mechanism that effectively incorporates desired diverse camera trajectories into the video diffusion model. This enables expansion of a single image into a multi-view consistent capture of a 3D scene with precise pose control. • Third, to achieve efficient 3D reconstruction, we directly transform video latents into 3DGS. We propose a novel latent-based large reconstruction model (LaLRM) that lifts video latents to 3D in a feed-forward manner. With this design, during inference, our model directly predicts 3DGS from a single input image, effectively aligning the generation and reconstruction tasks—and bridging image space and 3D space—through the video latent space. Compared with reconstructing scenes from images, the video latent space offers a 256× spatial-temporal reduction while retaining essential and consistent 3D structural details. Such a high degree of compression is crucial, as it allows the LaLRM to handle a wider range of 3D scenes within the reconstruction framework, with the same memory constraints.

MrNeRF

52,801 просмотров • 1 год назад

The (Processing) Level Required to Be Super We’ve had clients ask us whether our 3D/4DGS capture and processing pipeline holds up in high-end VFX cinematic shots. Last year we were fortunate to work on the Superman production with Framestore (capturing Superman’s parents for a hologram), with our work shown on IMAX screens ~30 m high—so we’re confident the quality delivers. For this R&D preview, we ran test captures using an affordable off-the-shelf temporal plugin to reduce flicker and improve stability over time. We recorded @ 48fps from our 196x machine vision cameras and trained the 4DGS using Postshot at Ultra quality ~20 million splat resolution. What you’re seeing are stop-motion 3DGS frames. Each frame is an individual Gaussian Splat—processed to ~20 million splats for the full scene, then cropped and cleaned to ~6 million splats per frame. Our system—and the 3DGS process—picks up fine details like stray fine hairs, skin pores and woven cloth threads. Thanks for watching & stay super! Learn more @ #3DGS #4DGS #VFX #GaussianSplatting #IMAX #VolumetricCapture #Superman 4K Version here -
0:19

The (Processing) Level Required to Be Super We’ve had clients ask us whether our 3D/4DGS capture and processing pipeline holds up in high-end VFX cinematic shots. Last year we were fortunate to work on the Superman production with Framestore (capturing Superman’s parents for a hologram), with our work shown on IMAX screens ~30 m high—so we’re confident the quality delivers. For this R&D preview, we ran test captures using an affordable off-the-shelf temporal plugin to reduce flicker and improve stability over time. We recorded @ 48fps from our 196x machine vision cameras and trained the 4DGS using Postshot at Ultra quality ~20 million splat resolution. What you’re seeing are stop-motion 3DGS frames. Each frame is an individual Gaussian Splat—processed to ~20 million splats for the full scene, then cropped and cleaned to ~6 million splats per frame. Our system—and the 3DGS process—picks up fine details like stray fine hairs, skin pores and woven cloth threads. Thanks for watching & stay super! Learn more @ #3DGS #4DGS #VFX #GaussianSplatting #IMAX #VolumetricCapture #Superman 4K Version here -

Infinite-Realities

44,109 просмотров • 9 месяцев назад

3DGS render blender add-on v3.0 ( was just open-sourced! With KIRI tools, you can use "3DGS to mesh" function to get a mesh from Gaussian splatting, and now you can convert the mesh back to Gaussian splatting. It's kind of ironic, but we just try to bring handy tools to everyone. #GaussianSplatting #Blender
0:13

3DGS render blender add-on v3.0 ( was just open-sourced! With KIRI tools, you can use "3DGS to mesh" function to get a mesh from Gaussian splatting, and now you can convert the mesh back to Gaussian splatting. It's kind of ironic, but we just try to bring handy tools to everyone. #GaussianSplatting #Blender

Chris make some 3D scans

39,221 просмотров • 1 год назад

🚀 Just added SPZ export to my 360° → 3DGS pipeline! 📊 8M splats → only 33MB (vs ~466MB PLY) ✨ New features: SPZ v3 format for 10x smaller files Adaptive sampling - more Gaussians where detail matters Quality presets: Low→Ultra (250K to 8M splats) Transform any equirectangular panorama into a Gaussian splat VR scene Painting: Joakim Skovgaard. Christ in the Kingdom of the Dead 🌐 #GaussianSplatting #unity3d #GameDev #VR #3DGS #Voxels #Physics #madewithunity 1892
0:13

🚀 Just added SPZ export to my 360° → 3DGS pipeline! 📊 8M splats → only 33MB (vs ~466MB PLY) ✨ New features: SPZ v3 format for 10x smaller files Adaptive sampling - more Gaussians where detail matters Quality presets: Low→Ultra (250K to 8M splats) Transform any equirectangular panorama into a Gaussian splat VR scene Painting: Joakim Skovgaard. Christ in the Kingdom of the Dead 🌐 #GaussianSplatting #unity3d #GameDev #VR #3DGS #Voxels #Physics #madewithunity 1892

Daniel Skaale

29,855 просмотров • 6 месяцев назад

3DGS needs to fix this if they want to be used for virtual tours: Gaussian splats struggle with ceilings. NeRFs ALWAYS beat out quality in this respect. I captured this over the weekend to compare. I have ideas on how to fix it. Anyone else have success with ceilings?
0:19

3DGS needs to fix this if they want to be used for virtual tours: Gaussian splats struggle with ceilings. NeRFs ALWAYS beat out quality in this respect. I captured this over the weekend to compare. I have ideas on how to fix it. Anyone else have success with ceilings?

Jonathan Stephens

12,340 просмотров • 2 лет назад

Create a 3D model from a single image, set of images or a text prompt in < 1 minute 😮‍💨 This new AI paper called CAT3D shows us that it’ll keep getting easier to produce 3D models from 2D images — whether it’s a sparser real world 3D scan (a few photos instead of hundreds) or your favorite 2D image generator like Midjourney (just an image). How does this magic work? “This architecture is similar to video diffusion models, but with camera pose embeddings for each image instead of time embeddings. The generated views are passed into a robust 3D reconstruction pipeline to create the 3D representation (Zip-NeRF or 3DGS)”
0:18

Create a 3D model from a single image, set of images or a text prompt in < 1 minute 😮‍💨 This new AI paper called CAT3D shows us that it’ll keep getting easier to produce 3D models from 2D images — whether it’s a sparser real world 3D scan (a few photos instead of hundreds) or your favorite 2D image generator like Midjourney (just an image). How does this magic work? “This architecture is similar to video diffusion models, but with camera pose embeddings for each image instead of time embeddings. The generated views are passed into a robust 3D reconstruction pipeline to create the 3D representation (Zip-NeRF or 3DGS)”

Bilawal Sidhu

92,792 просмотров • 2 лет назад

Arnold can now import, relight and render 3D Gaussian Splats Check out the other new features in the renderer, including updates to toon shading and experimental free cloud rendering service Flow Render #ArnoldRenderer #3DGS #VFX Arnold
0:16

Arnold can now import, relight and render 3D Gaussian Splats Check out the other new features in the renderer, including updates to toon shading and experimental free cloud rendering service Flow Render #ArnoldRenderer #3DGS #VFX Arnold

CG Channel

18,843 просмотров • 20 дней назад

A blueberry. 90 stacks, 68 photos each. 1.48M splats. #3dgs
0:12

A blueberry. 90 stacks, 68 photos each. 1.48M splats. #3dgs

Dany Bittel

721,625 просмотров • 1 месяц назад

Times are changing. Service providers are soon going to have to adapt. Or at least have the ability to generate your own diffusion models. 360 spin (Pixverse) 3DGS (Inria Sibr viewer) Synthetic Human 3DGS SD+LORA -> Pixverse 360 -> Metashape -> Postshot -> Inria Sibr (.PLY)
0:28

Times are changing. Service providers are soon going to have to adapt. Or at least have the ability to generate your own diffusion models. 360 spin (Pixverse) 3DGS (Inria Sibr viewer) Synthetic Human 3DGS SD+LORA -> Pixverse 360 -> Metashape -> Postshot -> Inria Sibr (.PLY)

Infinite-Realities

62,774 просмотров • 1 год назад

Triangle Splatting for Real-Time Radiance Field Rendering Contributions: (i) We propose Triangle Splatting, a novel approach that directly optimizes unstructured triangles, bridging traditional computer graphics and radiance fields. (ii) We introduce a differentiable window function for soft triangle boundaries, enabling effective gradient flow. (iii) We demonstrate qualitatively and quantitatively that Triangle Splatting outperforms concurrent methods in terms of visual quality and rendering speed, and achieves superior perceptual quality compared to the state-of-the-art Zip-NeRF on indoor scenes. (iv) The optimized triangles are directly compatible with standard mesh-based renderers, enabling seamless integration into traditional graphics pipelines.
0:16

Triangle Splatting for Real-Time Radiance Field Rendering Contributions: (i) We propose Triangle Splatting, a novel approach that directly optimizes unstructured triangles, bridging traditional computer graphics and radiance fields. (ii) We introduce a differentiable window function for soft triangle boundaries, enabling effective gradient flow. (iii) We demonstrate qualitatively and quantitatively that Triangle Splatting outperforms concurrent methods in terms of visual quality and rendering speed, and achieves superior perceptual quality compared to the state-of-the-art Zip-NeRF on indoor scenes. (iv) The optimized triangles are directly compatible with standard mesh-based renderers, enabling seamless integration into traditional graphics pipelines.

MrNeRF

51,407 просмотров • 1 год назад

🔥 #3DV2026 Oral: We present a matrix-free Levenberg-Marquardt optimizer that makes second-order optimization practical for 3DGS! 🚀
0:13

🔥 #3DV2026 Oral: We present a matrix-free Levenberg-Marquardt optimizer that makes second-order optimization practical for 3DGS! 🚀

Marc Habermann

25,676 просмотров • 3 месяцев назад

📢Pix2NPHM: Learning to Regress NPHM Reconstructions From a Single Image📢 We directly regress neural parametric head models (NPHMs) from a single image — fast, stable, and significantly more expressive than classical 3DMMs such as FLAME. Face tracking & 3D reconstruction are often limited by the representational capacity of PCA-based face models. By lifting NPHMs to a first-class reconstruction primitive, we enable more accurate geometry, richer expressions, and finer animation control. Pix2NPHM obtains fast and reliable NPHM reconstructions on real-world data. Inference-time optimization against surface normals and canonical point maps can further increase fidelity. Key to successful and generalized training of our ViT-based network are: (1) large-scale registration of existing 3D head datasets, and (2) self-supervised training on vast in-the-wild 2D video datasets using pseudo ground-truth surface normals. Finally, we show that geometry-aware pretraining on pixel-aligned reconstruction tasks significantly outperforms generic visual pretraining (e.g., DINO-style features) in terms of generalization. 🌍 🎥 Great work by Simon Giebenhain, Tobias Kirschstein, Liam Schoneveld, Davide Davoli, Zhe Chen
0:17

📢Pix2NPHM: Learning to Regress NPHM Reconstructions From a Single Image📢 We directly regress neural parametric head models (NPHMs) from a single image — fast, stable, and significantly more expressive than classical 3DMMs such as FLAME. Face tracking & 3D reconstruction are often limited by the representational capacity of PCA-based face models. By lifting NPHMs to a first-class reconstruction primitive, we enable more accurate geometry, richer expressions, and finer animation control. Pix2NPHM obtains fast and reliable NPHM reconstructions on real-world data. Inference-time optimization against surface normals and canonical point maps can further increase fidelity. Key to successful and generalized training of our ViT-based network are: (1) large-scale registration of existing 3D head datasets, and (2) self-supervised training on vast in-the-wild 2D video datasets using pseudo ground-truth surface normals. Finally, we show that geometry-aware pretraining on pixel-aligned reconstruction tasks significantly outperforms generic visual pretraining (e.g., DINO-style features) in terms of generalization. 🌍 🎥 Great work by Simon Giebenhain, Tobias Kirschstein, Liam Schoneveld, Davide Davoli, Zhe Chen

Matthias Niessner

37,850 просмотров • 6 месяцев назад

3D Gaussian Splatting for Real-Time Radiance Field Rendering paper page: Radiance Field methods have recently revolutionized novel-view synthesis of scenes captured with multiple photos or videos. However, achieving high visual quality still requires neural networks that are costly to train and render, while recent faster methods inevitably trade off speed for quality. For unbounded and complete scenes (rather than isolated objects) and 1080p resolution rendering, no current method can achieve real-time display rates. We introduce three key elements that allow us to achieve state-of-the-art visual quality while maintaining competitive training times and importantly allow high-quality real-time (>= 30 fps) novel-view synthesis at 1080p resolution. First, starting from sparse points produced during camera calibration, we represent the scene with 3D Gaussians that preserve desirable properties of continuous volumetric radiance fields for scene optimization while avoiding unnecessary computation in empty space; Second, we perform interleaved optimization/density control of the 3D Gaussians, notably optimizing anisotropic covariance to achieve an accurate representation of the scene; Third, we develop a fast visibility-aware rendering algorithm that supports anisotropic splatting and both accelerates training and allows realtime rendering. We demonstrate state-of-the-art visual quality and real-time rendering on several established datasets.
0:18

3D Gaussian Splatting for Real-Time Radiance Field Rendering paper page: Radiance Field methods have recently revolutionized novel-view synthesis of scenes captured with multiple photos or videos. However, achieving high visual quality still requires neural networks that are costly to train and render, while recent faster methods inevitably trade off speed for quality. For unbounded and complete scenes (rather than isolated objects) and 1080p resolution rendering, no current method can achieve real-time display rates. We introduce three key elements that allow us to achieve state-of-the-art visual quality while maintaining competitive training times and importantly allow high-quality real-time (>= 30 fps) novel-view synthesis at 1080p resolution. First, starting from sparse points produced during camera calibration, we represent the scene with 3D Gaussians that preserve desirable properties of continuous volumetric radiance fields for scene optimization while avoiding unnecessary computation in empty space; Second, we perform interleaved optimization/density control of the 3D Gaussians, notably optimizing anisotropic covariance to achieve an accurate representation of the scene; Third, we develop a fast visibility-aware rendering algorithm that supports anisotropic splatting and both accelerates training and allows realtime rendering. We demonstrate state-of-the-art visual quality and real-time rendering on several established datasets.

AK

633,532 просмотров • 2 лет назад