Andrea Tagliasacchi 🇨🇦🏔️ @taiyasaki Twitter profile

Pinned Tweet

Andrea Tagliasacchi 🇨🇦🏔️

12 days

📢📢📢 3D Gaussian Splatting brought you real-time rendering, but at slightly lower PSNR compared to mipNeRF360... 𝐚𝐬 𝐨𝐟 𝐭𝐨𝐝𝐚𝐲, 𝐭𝐡𝐚𝐭 𝐢𝐬 𝐧𝐨 𝐥𝐨𝐧𝐠𝐞𝐫 𝐭𝐫𝐮𝐞. Introducing "3D Gaussian Splatting as Markov Chain Monte Carlo"

9

25

252

Last Seen Profiles

@SVHStigers

@cccu_ald

@zhenjun_zhao

@NateRosenzweig

@Nickycrecco

@davincii777

@REALRAVERANGERS

@far_dhela

@TheNewParkway

@setkab

@NCTrojanArmy

@amy13walsh

@tunduuk

@JasherJ2023

@sebastienroblin

@ICMRNICED

@BOKEPVIRAL122

@Filzime

@EHSCrimsonTide

@PiFitti

@miralasku

@chasercam

@HerbertLuna

@onyegbuleonyeka

@YT_HARUKA

@qwunser

@xsukyzie

@ODDLOREofficial

@dejokintv

@Ryan_Kreidler

@EvaDichand

@Flaminiesta

@ClassiqueGTA

@FaizullahSwati

@deitsu_official

@symphonnic

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

📢📢📢 Thrilled to introduce "𝐌𝐨𝐛𝐢𝐥𝐞𝐍𝐞𝐑𝐅: exploiting the polygon rasterization pipeline for efficient neural field rendering on mobile architectures" → with 𝐥𝐢𝐯𝐞 𝐝𝐞𝐦𝐨𝐬 (Internship project lead by @ZhiqinChen3 )

7

404

2K

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

📢📢📢 Happy to introduce that "Kubric: a scalable dataset generator" was accepted at #CVPR2022 . – Do you work on video and/or 3D vision from images? – Having a hard time applying the "scientific method" to your research? Look no further 🧵👇

14

185

1K

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

📢📢📢 𝗔𝘁𝘁𝗲𝗻𝘁𝗶𝗼𝗻 𝗕𝗲𝗮𝘁𝘀 𝗖𝗼𝗻𝗰𝗮𝘁𝗲𝗻𝗮𝘁𝗶𝗼𝗻 𝗳𝗼𝗿 𝗖𝗼𝗻𝗱𝗶𝘁𝗶𝗼𝗻𝗶𝗻𝗴 𝗡𝗲𝘂𝗿𝗮𝗹 𝗙𝗶𝗲𝗹𝗱𝘀 We ran A LOT of experiments to find the best way to make neural fields generalize... so you don’t have to!

1

126

722

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

3 years

📢📢📢 𝐒𝐮𝐧 𝐞𝐭 𝐚𝐥. "𝐂𝐚𝐧𝐨𝐧𝐢𝐜𝐚𝐥 𝐂𝐚𝐩𝐬𝐮𝐥𝐞𝐬: 𝐔𝐧𝐬𝐮𝐩𝐞𝐫𝐯𝐢𝐬𝐞𝐝 𝐂𝐚𝐩𝐬𝐮𝐥𝐞𝐬 𝐢𝐧 𝐂𝐚𝐧𝐨𝐧𝐢𝐜𝐚𝐥 𝐏𝐨𝐬𝐞" A network design that realizes the concept of "𝐦𝐞𝐧𝐭𝐚𝐥 𝐩𝐢𝐜𝐭𝐮𝐫𝐞" for unsupervised 3D deep learning.

12

140

664

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

📢📢📢 𝐍𝐞𝐮𝐫𝐚𝐥 𝐒𝐞𝐦𝐚𝐧𝐭𝐢𝐜 𝐅𝐢𝐞𝐥𝐝𝐬 (NeSF) accepted to #TMLR2022 – In deep learning, we have tools to understand sequences, images, point clouds... we introduce the first field-to-field neural network design for scene understanding.

1

104

528

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

3 years

📢📢📢 Introducing "Vector Neurons" Want a network (and latent space) that act by construction in an equivariant way w.r.t. SO(3) transformations? All you need is to do is to generalize the scalar non-linearity to a vector one (e.g. Vector ReLU)

4

99

481

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

📢📢📢 Introducing 𝐍𝐞𝐮𝐫𝐚𝐥 𝐃𝐞𝐬𝐜𝐫𝐢𝐩𝐭𝐨𝐫 𝐅𝐢𝐞𝐥𝐝𝐬 (𝐍𝐃𝐅) That's right, we 𝐭𝐞𝐚𝐜𝐡 𝐚 𝐫𝐨𝐛𝐨𝐭 to manipulate unseen objects, and unseen poses from 𝐣𝐮𝐬𝐭 𝟏𝟎 𝐞𝐱𝐚𝐦𝐩𝐥𝐞𝐬 🤯 Wanna know more? See this thread

Vincent Sitzmann

@vincesitzmann

2 years

Introducing “Neural Descriptor Fields: SE(3)-Equivariant Object Representations for Manipulation”! (w/ video!) NDFs are an object representation for robotic manipulation enabling imitation of pick-and-place tasks with pose generalization guarantees (1/n)

8

105

585

4

70

447

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

📢📢📢 thrilled to announce in Aug 2022 I will be 𝐣𝐨𝐢𝐧𝐢𝐧𝐠 @SFU 𝐚𝐬 𝐚𝐧 𝐀𝐬𝐬𝐨𝐜𝐢𝐚𝐭𝐞 𝐏𝐫𝐨𝐟𝐞𝐬𝐬𝐨𝐫 and visual computer research chair in @SFU_CompSci Looking for PhDs, applications due Jan 10th; More info at →

49

24

411

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

4 years

Our paper won the #CVPR2020 best student paper award!!! Of course, many thanks to my amazing collaborators Zhiqin Chen, and Prof. Hao (Richard) Zhang. p.s. this is the first paper I wrote with my PhD advisor after graduation :)

19

44

397

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

5 months

📢📢📢 thrilled to announce "𝟒𝐃-𝐟𝐲: 𝐓𝐞𝐱𝐭-𝐭𝐨-𝟒𝐃 𝐆𝐞𝐧𝐞𝐫𝐚𝐭𝐢𝐨𝐧 𝐔𝐬𝐢𝐧𝐠 𝐇𝐲𝐛𝐫𝐢𝐝 𝐒𝐜𝐨𝐫𝐞 𝐃𝐢𝐬𝐭𝐢𝐥𝐥𝐚𝐭𝐢𝐨𝐧 𝐒𝐚𝐦𝐩𝐥𝐢𝐧𝐠" Way to start @sherwinbahmani 🎉 PhD co-advised with @DaveLindell at #UofT .

11

80

392

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

I am sorry Point-NeRF authors... but I am going to "steal" this for my lectures 😍🥰 Source: "Point-NeRF: Point-based Neural Radiance Fields"

4

50

363

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

4 years

NASA: Neural Articulated Shape Approximation Representing characters and deformable shapes without polygonal meshes for 3D perception, games, ...

6

75

347

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

3 years

Our new paper "𝐔𝐧𝐬𝐮𝐩𝐞𝐫𝐯𝐢𝐬𝐞𝐝 𝐩𝐚𝐫𝐭 𝐫𝐞𝐩𝐫𝐞𝐬𝐞𝐧𝐭𝐚𝐭𝐢𝐨𝐧 𝐛𝐲 𝐅𝐥𝐨𝐰 𝐂𝐚𝐩𝐬𝐮𝐥𝐞𝐬" is out TL;DR: are newborns exposed to 14 million (i.e. ImageNet) labeled images? No! They learn by observing motion... in an unsupervised fashion

6

66

345

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

8 months

📢📢📢 Introducing "𝐁𝐚𝐲𝐞𝐬' 𝐑𝐚𝐲𝐬: Uncertainty Quantification for Neural Radiance Fields" TL;DR: 𝐜𝐚𝐧 𝐈 "𝐭𝐫𝐮𝐬𝐭" 𝐦𝐲 𝐍𝐞𝐑𝐅 𝐚𝐭 𝐩𝐨𝐬𝐢𝐭𝐢𝐨𝐧 𝐱? with @lily_goli , @sellan_s , @code_red7777 , @_AlecJacobson .

2

51

331

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

You can tell that @ak92501 is out of town... Follow-up to piGAN: "EG3D: Efficient Geometry-aware 3D Generative Adversarial Networks" ... 🤯🤯

4

51

315

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

Shen et al. "Deep Marching Tetrahedra: a Hybrid Representation for High-Resolution 3D Shape Synthesis" This is so cool, and eluded my twitter radar :)

0

43

311

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

📢📢📢 𝐕𝐨𝐥𝐮𝐦𝐞 𝐑𝐞𝐧𝐝𝐞𝐫𝐢𝐧𝐠 𝐃𝐢𝐠𝐞𝐬𝐭 (𝐟𝐨𝐫 𝐍𝐞𝐑𝐅) Tech-report: Hope it's helpful, it's a self-contained step-by-step derivation of the volume rendering math (I needed to go carefully through it for another paper). w/ @BenMildenhall

1

43

296

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

4 years

Do you like implicit functions to model 3D geometry? We show that "Neural Articulated Shape Approximation" improves the approximation power by over 50% ABSOLUTE!! Fast-forward of our #ECCV paper, full talk, additional material, and code comings soon.

3

66

265

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

3 years

I just published the cleaned-up official #CVPR2022 latex template on @overleaf – might it save you some precious time :) Best of luck to all your submissions! 🍀

CVPR 2022 Author Kit

An online LaTeX editor that’s easy to use. No installation, real-time collaboration, version control, hundreds of LaTeX templates, and more.

www.overleaf.com

3

34

255

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

1 year

📢📢📢 Excited to announce “𝐧𝐞𝐫𝐟𝟐𝐧𝐞𝐫𝐟: 𝐏𝐚𝐢𝐫𝐰𝐢𝐬𝐞 𝐑𝐞𝐠𝐢𝐬𝐭𝐫𝐚𝐭𝐢𝐨𝐧 𝐨𝐟 𝐍𝐞𝐮𝐫𝐚𝐥 𝐑𝐚𝐝𝐢𝐚𝐧𝐜𝐞 𝐅𝐢𝐞𝐥𝐝𝐬” → PDF: → Homepage: → Code: → Video:

2

53

254

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

Enjoy physics? Fascinated by neural nets? Have a fetish for @neural_fields ? This is a must read by @lipmanya

9

50

250

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

📢📢📢 "Novel View Synthesis with Diffusion Models" Single shot novel view synthesis results of quality never seen before. And, surprisingly, it does _not_ use NeRF to obtain a 3D consistent model, but rather 3D consistency is learnt during training.

Daniel Watson

@watson_nn

2 years

Excited to announce our work on novel view synthesis with diffusion models! Our model can lift a single 2d image into 3d. Joint work w/ @wchan212 @rmbrualla @hojonathanho @taiyasaki @mo_norouzi

67

945

5K

2

53

250

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

6 months

As you work on your #CVPR2024 , don't forget you are human. Sleep, exercise, eat well, find some time to spend with friends. ... Best of luck ❤️‍🔥

2

26

241

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

📢📢📢 Have you ever wondered what "capsules" are in deep learning? For primary capsules, it is actually very very simple, and explained excellently by Weiwei in his upcoming talk at #NeurIPS21 , linked below: Thread 🧵⬇️

Canonical Capsules @ NeurIPS2021

A self-supervised capsule architecture that canonicalizes data while simultaneously decomposing point clouds into parts to perform unsupervised representation learning.

canonical-capsules.github.io

17

48

239

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

Reverse-Engineering Deep ReLU Networks i.e. begging computational geometry people to pay attention to developments in deep learning :)

5

26

221

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

A @CVPR reviewer is asking a comparison against a black-box commercial package ($4k+) and I have nothing in the #CVPR2022 policies to point out such ask is unacceptable!!! 😮😮 1) a dangerous path towards "pay-to-publish"? 2) for all I know, the reviewer could own the company?

22

13

219

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

Huge fan of this piece of research (i.e. wish it was mine). Time to bring geometry processing to the new era! And _plenty_ of ways to extend this paper... "Geometry Processing with Neural Fields"

4

34

216

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

6 months

This is how you get a 2-column (SIGGRAPH style) teaser image in #CVPR2024 (code and outcome below)

3

32

212

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

1 year

#CVPR2023 is too much about bold numbers. And it is WAY too easy to cheat on those (few keyboard strokes, right?) We found two CVPR (orals) that mis-represented their results –– figuring out what to do in regards... You WILL be found out. Don't – just don't

10

18

211

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

1 month

Cool results with surfels rather than splats. ~2024 feels like ~2004 all over again (the era of point based rendering)

2

24

209

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

If you have a friend that works at Google... stop and give him/her a hug 🫂 It is "perf season", and generally a pretty painful moment of the year... especially for researchers.

11

4

204

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

5 years

Do you need differentiable convex hulls? Take a look at our CvxNet work

5

51

185

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

3 years

@mmitchell_ai The fact that you will lose your healthcare BECAUSE you lost your job is the terrifying bit in your post.

5

2

186

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

NeRF training (apparently) in 5s? :O (up up you go to the top of the reading stack)

AK

@_akhaliq

2 years

Instant Neural Graphics Primitives with a Multiresolution Hash Encoding paper: project page: github:

21

360

2K

2

24

183

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

Neural rendering goes brrrrrr...

Wenzel Jakob

@wenzeljakob

2 years

We can also jointly optimize geometry and appearance. In the example below, we determine the albedo and roughness of a Disney BSDF (shown under new viewing/illumination conditions). (6/8)

4

50

249

1

20

179

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

📢📢📢 𝐂𝐔𝐅 – 𝐂𝐨𝐧𝐭𝐢𝐧𝐮𝐨𝐮𝐬 𝐔𝐩𝐬𝐚𝐦𝐩𝐥𝐢𝐧𝐠 𝐅𝐢𝐥𝐭𝐞𝐫𝐬 Neural-fields beat classical CNNs in (regressive) super-res: AFAIK, a first for @neural_fields in 2D deep learning? Mostly their wins are in sparse, higher-dim signals ~NeRF

2

44

176

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

6 months

Want to generate tables programmatically in overleaf? And just copy-paste from your google spreadsheet Check this out:

1

25

172

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

1 year

Considering a PhD? Interested in 3D deep learning, neural scene representations, NeRFs, and applications to robotics and graphics? Come to Vancouver, 𝐒𝐢𝐦𝐨𝐧 𝐅𝐫𝐚𝐬𝐞𝐫 𝐔𝐧𝐢𝐯𝐞𝐫𝐬𝐢𝐭𝐲 (SFU) Deadline is 𝐉𝐚𝐧𝐮𝐚𝐫𝐲 𝟏𝟖 𝟐𝟎𝟐𝟑 Apply here:

1

42

166

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

4 months

📢📢📢 introducing 𝐩𝐢𝐱𝐞𝐥𝐒𝐩𝐥𝐚𝐭: feed-forward Gaussian splats from image pairs! pixelNeRF + 3D Gaussian Splatting == Led by @DavidCharatan and @sizhe_lester_li , collaboration with @vincesitzmann

2

23

154

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

1 month

📢📢📢 TC4D: Trajectory-Conditioned Text-to-4D Generation Homepage: Lead by @sherwinbahmani , with collaborators @AlvinLiu27 @toomanyyifans @isskoro @victor__rong @liuziwei7 @XihuiLiu @jjpark3D @SergeyTulyakov @GordonWetzstein @taiyasaki @DaveLindell

2

35

156

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

4 years

If you see me a bit hazed these days... it is because our 2nd daughter "Mila" was born mid-September!!! and now... off to the CVPR deadline 🤣😂

16

2

152

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

Finally somebody run a nice ablation on all the (Ne)RF-class training acceleration papers! (InstantNGP, DVFo, Plenoxels) This is a *must* read

Dmytro Mishkin 🇺🇦

@ducha_aiki

2 years

ReLU Fields: The Little Non-linearity That Could @AnimeshKarnewar , Tobias Ritschel, Oliver Wang, Niloy J. Mitra tl;dr: grid-based 3D representation + ReLU are almost NERFs, but much faster.

10

58

347

3

18

145

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

1 year

📢📢📢 Thrilled to announce that nerf2nerf was accepted to ICRA 2023 – Huge congrats to @lily_goli on publishing her first PhD project (in her first year!)

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

1 year

📢📢📢 Excited to announce “𝐧𝐞𝐫𝐟𝟐𝐧𝐞𝐫𝐟: 𝐏𝐚𝐢𝐫𝐰𝐢𝐬𝐞 𝐑𝐞𝐠𝐢𝐬𝐭𝐫𝐚𝐭𝐢𝐨𝐧 𝐨𝐟 𝐍𝐞𝐮𝐫𝐚𝐥 𝐑𝐚𝐝𝐢𝐚𝐧𝐜𝐞 𝐅𝐢𝐞𝐥𝐝𝐬” → PDF: → Homepage: → Code: → Video:

2

53

254

3

19

144

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

1 year

TL;DR: COLMAP completely fails in few shot settings, while neural bundle adjustment (SparsePose) regresses quite accurate camera poses. Congratulations @_sam_sinha_ on a flawless execution! See you at #CVPR2023

Samarth Sinha

@_sam_sinha_

1 year

Super excited to share that my first project as a PhD student was accepted at CVPR 2023! #CVPR2023 😄 We show how to recover accurate camera poses from as few as 3 images! More details to come! w/ @jasonyzhang2 @taiyasaki @igilitschenski @DaveLindell

27

121

763

2

25

140

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

5 months

📢📢📢 𝐀𝐜𝐜𝐞𝐥𝐞𝐫𝐚𝐭𝐢𝐧𝐠 𝐍𝐞𝐮𝐫𝐚𝐥 𝐅𝐢𝐞𝐥𝐝 𝐓𝐫𝐚𝐢𝐧𝐢𝐧𝐠 𝐯𝐢𝐚 𝐒𝐨𝐟𝐭 𝐌𝐢𝐧𝐢𝐧𝐠 by Shakiba Kheradmand et al. TL;DR: importance sampling for accelerating your novel view synthesis workloads (...yes, it should also work for 3DGS)

2

19

139

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

10 months

We have had an _amazing_ week in Banff at the BIRS “3D Generative Models” workshop Thank to @BIRS_Math for hosting us!

4

19

135

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

4 years

Source code and full talk for 𝗡𝗔𝗦𝗔: 𝗡𝗲𝘂𝗿𝗮𝗹 𝗔𝗿𝘁𝗶𝗰𝘂𝗹𝗮𝘁𝗲𝗱 𝗦𝗵𝗮𝗽𝗲 𝗔𝗽𝗽𝗿𝗼𝘅𝗶𝗺𝗮𝘁𝗶𝗼𝗻 is now available – links 👇👇 Talk @ #ECCV2020 → Project page → Source code →

NASA: Neural Articulated Shape Approximation (ECCV 2020)

Project Page: https://nasa-eccv20.github.io/.Code: https://github.com/tensorflow/graphics/tree/master/tensorflow_graphics/projects/nasa.Paper: https://arxiv....

www.youtube.com

2

46

132

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

10 months

📢📢📢 CC3D: Layout-Conditioned Generation of Compositional 3D Scenes was accepted by #ICCV2023 Congratulation @sherwinbahmani on your first AAA first-author paper, and very much looking forward to help you grow at @UofT with @DaveLindell

1

16

133

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

10 months

Congrats for the #SIGGRAPH2023 best paper award @GKopanas 🎉🎉

George Kopanas

@GKopanas

1 year

We will be in @siggraph 2023 with "3D Gaussian Splatting for Real-Time Radiance Field Rendering", have you ever seen radiance fields with 100+ FPS and MipNeRF360 quality? Check out our website here:

13

147

823

1

16

130

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

1 year

Alright, the word it out 😊 is one of the 12 finalists for best paper award at #CVPR2023 🎉🎉🎉

#CVPR2024

@CVPR

1 year

The twelve #CVPR2023 award candidate papers are listed at . Congratulations to all of the authors on this achievement!

1

27

98

7

10

128

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

Implicit functions meets kernel regression. Excellent work by @frncswllms and @NVIDIAAI (had reviewed this work as part of Francis' PhD thesis at NYU): Is the "era" of Poisson surface reconstruction officially over? MeshLab plugin @ALoopingIcon ? :)

0

19

127

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

5 months

📢📢📢 𝐔𝐧𝐬𝐮𝐩𝐞𝐫𝐯𝐢𝐬𝐞𝐝 𝐊𝐞𝐲𝐩𝐨𝐢𝐧𝐭𝐬 𝐟𝐫𝐨𝐦 𝐏𝐫𝐞𝐭𝐫𝐚𝐢𝐧𝐞𝐝 𝐃𝐢𝐟𝐟𝐮𝐬𝐢𝐨𝐧 𝐌𝐨𝐝𝐞𝐥𝐬 TL;DR: semantic keypoints within a dataset emerge naturally from latent-diffusion model's text embeddings.

1

23

121

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

Wang et a. "SunStage: Portrait Reconstruction and Relighting using the Sun as a Light Stage" Very cool work, i.e. using the sun as a light-probe (i.e. point light source in a light-stage) ... and yet another to scratch off the list of ideas :)

1

25

116

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

3 years

Rebain et al. "𝐃𝐞𝐜𝐨𝐦𝐩𝐨𝐬𝐞𝐝 𝐑𝐚𝐝𝐢𝐚𝐧𝐜𝐞 𝐅𝐢𝐞𝐥𝐝𝐬 (𝐃𝐞𝐑𝐅)" That is, the render time of NeRFs can be improved (by +𝟑𝟎𝟎%!) by decomposing the scene representation into many lower-capacity networks. Collaboration {UBC, Google, SFU}

1

20

112

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

Very neat paper to deal with equivariance in point clouds ( #CVPR2022 ). Recommended read if you work on point clouds! Core idea? Simple! Learn a local coordinate frame by message passing.

2

13

110

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

InstantNGP won one of the #SIGGRAPH2022 best paper awards. Congrats @mmalex and colleagues! Transformative work.

Aaron Hertzmann

@AaronHertzmann

2 years

@siggraph @budmonde @qisun0 "Instant Neural Graphics Primitives with a Multiresolution Hash Encoding", which can train neural graphics primitives in seconds and render them in milliseconds. 4/6

1

8

58

2

8

111

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

📢📢📢 Do you have verifiable research record in #NeuralFields / #NeRF and would like to come to @GoogleAI ? I collaborate with the SynthX team, that now have several Student Researcher positions available (eg. past project ) Apply → synthx-jobs @google .com

1

22

109

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

Why is peer-reviews becoming more and more toxic? (i.e. trying to find negativity in an idea at all costs, and discount all but perfection). The more toxic reviews you write, the more toxic reviews you are to expect for your own papers (i.e. what goes around, comes around).

2

9

107

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

4 years

Deng et al. CVPR 2020 "CvxNet: Learnable Convex Decomposition" Oral Presentation:

4

25

104

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

1 year

If you haven't had a change to read NeRFPlayer... you should. Been pondering whether this makes HyperNeRF _completely_ obsolete? Is there anything that HyperNeRF does better? I don't think so, but feel free to argue otherwise! Well done @AutoVisionGroup

NeRFPlayer: A Streamable Dynamic Scene Representation with Decomposed...

NeRFPlayer: A Streamable Dynamic Scene Representation with Decomposed Neural Radiance FieldsLiangchen Song, Anpei Chen, Zhong Li, Zhang Chen, Lele Chen, Juns...

www.youtube.com

2

21

102

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

Shouldn't this be fixed ASAP? This is 🤮 cringe-worthy 🤮for anybody with a signal processing background. Also, this matters: "Impact of Aliasing on Generalization in Deep Convolutional Networks"

Impact of Aliasing on Generalization in Deep Convolutional Networks

We investigate the impact of aliasing on generalization in Deep Convolutional Networks and show that data augmentation schemes alone are unable to prevent it due to structural limitations in...

arxiv.org

Alfredo Canziani

@alfcnz

2 years

I love figure 3! 😍😍😍 A few keywords for future searches: PyTorch, TensorFlow, OpenCV, PIL, image rescaling, downsample, decimation, aliasing.

3

52

261

7

8

100

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

1 year

The social media ban is completely absurd... "Nobody" will know this is from NVIDIA, right?

AK

@_akhaliq

1 year

Magic3D: High-Resolution Text-to-3D Content Creation abs: project page:

24

299

1K

6

4

98

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

4 years

Our CVPR 2020 oral paper "Binary Space Partitioning Networks (BSP-Net)" has now source code available! TL;DR: a neural network that generates low-polycount meshes code → preprint →

1

23

96

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

Alright, name sadly scooped, although alas we moved away from fusion and towards semantics (i.e. NeSF). ... but, how well does it work off trajectory? Anybody seen ScanNet results? NeRFusion: Fusing Radiance Fields for Large-Scale Scene Reconstruction

1

21

96

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

Next year #CVPR2023 I play at home ;) Looking forward to welcome you to Vancouver!

1

3

92

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

The more I AC/PC, the more I realize we need to show PhD students what happens during AC review consolidation. This will contribute to significantly higher quality reviews, which will also make judging outcomes faster/easier.

3

2

90

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

1 year

📢📢📢 Excited to co-chair 3DV 2024 with Siyu Tang @SiyuTang3 and Federico Tombari @fedassa Smaller conferences are where you create the networking that will last a lifetime.

International Conference on 3D Vision

@3DVconf

1 year

📣 Exciting news! 📣 The call for papers is now open for 3DV2024, taking place in the stunning Davos, Switzerland! 📝 Submission: by July 31. 🔗 Details: 🌟 Don't miss the chance to share your research at #3DV2024 🎉 #CallForPapers #Davos #Switzerland

1

37

123

2

8

87

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

10 months

See you at #CVPR2023 – come say hi 😊

1

4

80

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

1 year

Working on #CVPR2023 ? A simple trick to organize your floats. Store your figure data in the fig/ folder, and automate the rest of your life. Your filenames then become your \ref{...} keys See snapshots👇👇

0

16

85

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

3 years

WARNING: controversial opinion incoming it's BULLSHIT to put a blanket ban on social media ads, yet allow seniors present unpublished (to be submitted) work at major venues. This only gives an unfair advantage to the big labs, w/ active senior members on the topic. #ICCV #CVPR

4

2

82

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

3 years

📢📢📢 Happy to announce "Canonical Capsules" (or CaCa 💩) will be presented at #NeurIPS 2021! Congrats especially to Weiwei!! @wsunid

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

3 years

📢📢📢 𝐒𝐮𝐧 𝐞𝐭 𝐚𝐥. "𝐂𝐚𝐧𝐨𝐧𝐢𝐜𝐚𝐥 𝐂𝐚𝐩𝐬𝐮𝐥𝐞𝐬: 𝐔𝐧𝐬𝐮𝐩𝐞𝐫𝐯𝐢𝐬𝐞𝐝 𝐂𝐚𝐩𝐬𝐮𝐥𝐞𝐬 𝐢𝐧 𝐂𝐚𝐧𝐨𝐧𝐢𝐜𝐚𝐥 𝐏𝐨𝐬𝐞" A network design that realizes the concept of "𝐦𝐞𝐧𝐭𝐚𝐥 𝐩𝐢𝐜𝐭𝐮𝐫𝐞" for unsupervised 3D deep learning.

12

140

664

5

7

81

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

5 months

The gaussian SLAM war has begun.

Hide

@HideMatsu82

5 months

Happy to share Gaussian Splatting SLAM We show the first 3DGS-based Monocular RGB SLAM, the hardest SLAM setting. Using 3D Gaussians as a unified representation, the method only requires RGB images - No need for SfM, depth sensor, or learned prior.

18

138

650

0

8

77

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

1 year

How did I miss this? 😀 Quite a change. The CVPR’23 program will not have orals/spotlights. Only Best Paper Award nominees (~20-30 in total) will be given a plenary presentation at the conference.

8

1

79

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

1 year

Did I again miss something major?!?! @CVPR ? Going backward and paywall'ing research?!!? 🤯 Was there even a CVF vote for this? What's wrong with

4

11

78

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

1 year

"Attention Beats Concatenation" has been accepted (as is) at #TMLR – congratulations to Daniel (PhD student) for yet another successful project! Now quite sure how to post it at @TmlrPub ? ;) )

Attention Beats Concatenation for Conditioning Neural Fields

Neural fields model signals by mapping coordinate inputs to sampled values. They are becoming an increasingly important backbone architecture across many fields from vision and graphics to biology...

openreview.net

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

📢📢📢 𝗔𝘁𝘁𝗲𝗻𝘁𝗶𝗼𝗻 𝗕𝗲𝗮𝘁𝘀 𝗖𝗼𝗻𝗰𝗮𝘁𝗲𝗻𝗮𝘁𝗶𝗼𝗻 𝗳𝗼𝗿 𝗖𝗼𝗻𝗱𝗶𝘁𝗶𝗼𝗻𝗶𝗻𝗴 𝗡𝗲𝘂𝗿𝗮𝗹 𝗙𝗶𝗲𝗹𝗱𝘀 We ran A LOT of experiments to find the best way to make neural fields generalize... so you don’t have to!

1

126

722

3

10

77

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

"Self-Calibrating Neural Radiance Fields" Noteworthy :)

1

8

76

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

4 years

I am happy to announce that all our submissions to #NeurIPS2020 were accepted! Special congratulations to the first (student!) authors Emre Aksan (ETHZ), Max Jiang (UCBerkeley), and Xiaogang Wang (SFU)

PIE-NET: Parametric Inference of Point Cloud Edges

We introduce an end-to-end learnable technique to robustly identify feature edges in 3D point cloud data. We represent these edges as a collection of parametric curves (i.e.,lines, circles, and...

arxiv.org

3

9

76

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

“Taking a NeRF” 😂 Selfies are so 2010s Hail to 3D perception

Franc Lucent

@franclucent

2 years

We went scootering and decided to take a NeRF. I think it came out pretty cool! What do you think? #NVIDIA #AI #photogrammetry #NeRFs #3D @NVIDIAAIDev @NVIDIAAI

5

55

192

1

6

76

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

3 months

How to evaluate novelty fairly in a paper? 🧐 @CVPR this might not be a bad article to broadcast to reviewers about to read rebuttals for #CVPR2024 Thanks again to @Michael_J_Black for writing it.

Novelty in Science

A guide to reviewers

medium.com

2

13

75

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

1 month

The @3DVconf PC. Couldn’t have asked for a better team, really 🥰

1

0

75

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

📢📢📢 Urban Radiance Fields @ #CVPR2022 → proof on how to integrate depth/lidar in NeRF → elegant modelling of exposure/balance → ... on Google StreetView data!

Vittorio Ferrari

@VittoFerrariCV

2 years

Five papers accepted to #CVPR22 ! 1/5: Urban Radiance Fields Novel view synthesis and 3D reconstruction in urban outdoor environments (StreetView). With @krematas , @ndrewLiu , @_pratul_ , @jon_barron , @taiyasaki , Funkhouser

1

16

155

2

8

73

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

How can you get a NeRF model that runs at interactive frame rate on most devices? (...CUDA is not an option) The idea is to exploit the rasterization pipeline as much as possible, and employ (𝐧𝐞𝐮𝐫𝐚𝐥) 𝐝𝐞𝐟𝐞𝐫𝐫𝐞𝐝 𝐬𝐡𝐚𝐝𝐢𝐧𝐠.

2

5

73

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

7 months

📢📢📢 Thrilled to announce that both our papers were accepted by #NeurIPS2023 - Neural Fields with Hard Constraints of Arbitrary Differential Order () - Unsupervised Semantic Correspondence Using Stable Diffusion ()

Unsupervised Semantic Correspondence Using Stable Diffusion

Text-to-image diffusion models are now capable of generating images that are often indistinguishable from real images. To generate such images, these models must understand the semantics of the...

arxiv.org

0

6

73

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

Fantastic work by @akanazawa 's team i.e. 32dB PSNR in 9minutes (vs. NeRF 1.6 days) Angjoo... acronym way too long? It is not as friendly as NeRF {NeRFW, SNErG, DeRF, DNERF, ...}.

Plenoxels: Radiance Fields without Neural Networks

Plenoxels: Radiance Fields without Neural NetworksAlex Yu*, Sara Fridovich-Keil*, Matthew Tancik, Qinhong Chen, Benjamin Recht, and Angjoo KanazawaProject pa...

www.youtube.com

0

9

74

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

. @CVPR many students from China won't be able to travel given the 30 days quarantine imposed upon re-entrance from the US. Might be worth considering allowing for local events so that PhD students can (finally) have an in person experience?

0

3

72

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

This is a big deal (for PyTorch devs)

NVIDIA AI Developer

@NVIDIAAIDev

2 years

Are you interested in #NeRFs ? NVIDIA Kaolin Wisp provides a framework and the building blocks for neural field research.

0

46

247

0

7

72

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

And this is the core technical idea – super simple!!! (...and in no way limited to point clouds)

0

4

71

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

3 years

Simply amazed that @ICCV_2021 does not have 3D deep learning, geometric deep learning, 3D representations in their subject areas? The closest are: – 3D from a single image and shape-from-x – Stereo, 3D from multi-view and other sensors And they are barely appropriate...

9

5

72

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

Sometime in the future, we’ll have more 3D data than 2D. Until then, distilling 2D self-supervised understanding to 3D seems quite effective. Great work @vincesitzmann + team: Also, parallel work (~identical) by Vedaldi + team:

2

6

72

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

3 months

The best #CVPR2024 reviewers clearly and concisely tell the authors what they expect to see in the rebuttal, and whether it will change their review score. This is the best content for the "justification for recommendation" section.

1

4

71

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

5 months

And if you have not noticed... 𝐬𝐨𝐮𝐫𝐜𝐞 𝐜𝐨𝐝𝐞 𝐢𝐬 𝐩𝐮𝐛𝐥𝐢𝐬𝐡𝐞𝐝. AFAIK the first open source codebase on the topic:

GitHub - sherwinbahmani/4dfy: 4D-fy: Text-to-4D Generation Using Hybrid Score Distillation Sampling

4D-fy: Text-to-4D Generation Using Hybrid Score Distillation Sampling - sherwinbahmani/4dfy

github.com

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

5 months

📢📢📢 thrilled to announce "𝟒𝐃-𝐟𝐲: 𝐓𝐞𝐱𝐭-𝐭𝐨-𝟒𝐃 𝐆𝐞𝐧𝐞𝐫𝐚𝐭𝐢𝐨𝐧 𝐔𝐬𝐢𝐧𝐠 𝐇𝐲𝐛𝐫𝐢𝐝 𝐒𝐜𝐨𝐫𝐞 𝐃𝐢𝐬𝐭𝐢𝐥𝐥𝐚𝐭𝐢𝐨𝐧 𝐒𝐚𝐦𝐩𝐥𝐢𝐧𝐠" Way to start @sherwinbahmani 🎉 PhD co-advised with @DaveLindell at #UofT .

11

80

392

1

15

70

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

Awesome paper :)

David Lindell

@DaveLindell

2 years

Pleased to announce our #CVPR2022 paper on BACON: band-limited coordinate networks! This is an interpretable network architecture with a controllable Fourier spectrum for multiscale scene representation. video: paper: 1/n

5

38

231

1

5

68

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

The core of the paper is how to "bake" a NeRF into a mesh and a (feature+opacity) texture map. The idea is to replace the NeRF quadrature points with the intersection points of a ray with the mesh (of fixed topology, but optimizable vertex positions).

1

2

66

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

Confirm, this is an excellently written paper, with strong technical ideas. TL;DR = NSVF + VoxelHashing + CUDA = 💕💕 Amazing engineering work by Thomas Muller at @NVIDIAAI

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

NeRF training (apparently) in 5s? :O (up up you go to the top of the reading stack)

2

24

183

2

5

66

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

This is quite cool, diffusion on occupancy functions to predict (a distribution of) 3D cell shapes. "A Diffusion Model Predicts 3D Shapes from 2D Microscopy Images"

1

8

66

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

3 months

As an advisor, I find it odd that I am required to pose questions to my own students during a defense. We have worked together for years... I have already asked them all my questions. What is the point? Is it from an era when students were "on their own"?

8

1

65

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

4 years

So... >10% of ECCV is working on 3D! Papers with "3D" in the title: 132 out of 1,361 total entries Can I just say that this website is AMAZING? Snappy to find what will appear (if only mouse overlays would give authors/affiliation/abstract...)

4

8

64

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

3 years

I'm not sure how I feel about the fact that I can see other reviewer's identities at #NeurIPS ... how is that not biasing the review system? (i.e. junior folks altering their scores after seeing a senior providing a polarizing opinion)

2

1

62

Andrea Tagliasacchi 🇨🇦🏔️

@taiyasaki

2 years

📢📢📢 If you are working on 3D deep learning, and you are annoyed by the #CVPR2022 ban, I would STRONGLY suggest you to give a try to the SIGGRAPH conference track. Thanks to the @siggraph team for making this happen, this is fantastic 🎉🥳🎉

Aaron Hertzmann

@AaronHertzmann

2 years

#SIGGRAPH2022 this year has a new Conference Papers track! Submissions may be riskier/less-polished than in the existing journal track. Papers limited to 7-pages plus references. Deadline Jan 27. @siggraph

3

70

232

1

9

62