Christian Diller

AI Research Scientist @ Beyond Presence | Previously @ Meta Reality Labs, PhD @ 3D AI Lab (TU Munich) | 3D CV & GenAI

[Angela Dai @angelaqdai]

Check out our #CVPR'24 papers on 3D human interactions, generative 3D modeling, and uncertainty-aware and unsupervised 3D semantic scene understanding!

Congrats to @craigleili @david_roz_ @chrdiller @yawarnihal @shivangi2201 @jiapeng_tang @AnhQuanCAO for their amazing work!

[Matthias Niessner @MattNiessner]

(1/2)
Check out DPHMs: Diffusion Parametric Head Models for Depth-based Tracking!

3D head reconstruction from noisy & sparse depth? Our diffusion prior constrains identity/expression latents, and maps them to high-quality samples.

https://t.co/GHrC7NTUvb
https://t.co/1zK8ntsNHq

[Matthias Niessner @MattNiessner]

(1/2)
Check out 𝐆𝐚𝐮𝐬𝐬𝐢𝐚𝐧𝐀𝐯𝐚𝐭𝐚𝐫𝐬: Photorealistic Head Avatars with Rigged 3D Gaussians!

We create photorealistic head avatars by animating 3D Gaussians on a parametric face model - edited and rendered in *real-time*!

https://t.co/R90VnWJB9Y
https://t.co/Gv5gED01SG

[Angela Dai @angelaqdai]

Check out @DaoyiGao's DiffCAD - introducing probabilistic CAD retrieval and alignment to an RGB image.

We captures ambiguities in depth/scale, inexact CAD matches, and don't require any training on real data!

https://t.co/eFZgP9JP4j
https://t.co/giMJl6R9Kl

[Matthias Niessner @MattNiessner]

(1/2) 📢📢𝐃𝐢𝐟𝐟𝐮𝐬𝐢𝐨𝐧𝐀𝐯𝐚𝐭𝐚𝐫𝐬 📢📢

High-fidelity 3D head avatars with precise control over viewpoint, expression, and pose.

-> Our parametric 3D model enables control & consistency + 2D diffusion makes it photoreal.

https://t.co/XCmGtehL07
https://t.co/nTBuBFaDlY

[Angela Dai @angelaqdai]

Can we synthesize 3D human-scene interactions without learning from any 3D data?

Yes! Check out @craigleili's GenZI, a novel zero-shot approach to generating 3D interactions by distilling priors from large vision-language models.

https://t.co/2zAgsqgXvD
https://t.co/aMWmc2tjVs

[#CVPR2026 @CVPR]

Starting today, #CVPR2024 Area Chairs will begin the process of recommending reviewers for each of the 11,523 valid submissions. We expect to distribute paper assignments to reviewers on December 9th!

[Matthias Niessner @MattNiessner]

(1/2)
📢📢𝗦𝗰𝗲𝗻𝗲𝗧𝗲𝘅 📢📢

Given scene geometry and text prompt -> SceneTex generates high-quality textures.

Main idea: directly optimize scene texture with gradients from a score-distillation objective with view sampling.

https://t.co/Rf1YLit6Qo
https://t.co/AKNpQDKOUX

[Christian Diller @chrdiller]

We generate 3D full-body human-object interactions just from text and object geometry.
Explicitly modeling contact is key for realistic interactions :)

[Angela Dai @angelaqdai]

Check out @chrdiller's CG-HOI :)

We generate realistic 3D human-object interactions, from object geometry and text description.

A key ingredient is explicit modeling of contact, during training and as guidance during inference.

https://t.co/Cl5Jw9oFBO
https://t.co/FVIFqEpjHi

[Angela Dai @angelaqdai]

Excited to have a distinguished guest lecture today by Turing Award laureate, three time Academy Award recipient Pat Hanrahan, speaking on programmable graphics today at @TU_Muenchen!

For TUM students, check it out online here: https://t.co/S4IqztI8aB

[Matthias Niessner @MattNiessner]

(1/2)
Check out 𝐌𝐞𝐬𝐡𝐆𝐏𝐓!

MeshGPT generates triangle meshes by autoregressively sampling from a transformer model that produces tokens from a learned geometric vocabulary.

As a result, we obtain clean and compact meshes :)

https://t.co/ynrf0qjYVF
https://t.co/rQe7ipP15t

[Matthias Niessner @MattNiessner]

Diffusion models are awesome!

Check out our survey on 𝐃𝐢𝐟𝐟𝐮𝐬𝐢𝐨𝐧 𝐌𝐨𝐝𝐞𝐥𝐬 𝐟𝐨𝐫 𝐕𝐢𝐬𝐮𝐚𝐥 𝐂𝐨𝐦𝐩𝐮𝐭𝐢𝐧𝐠! We give an introduction to diffusion models and highlight how they are used by state-of-the-art methods in graphics and vision.

https://t.co/FqaqF7tMPM

[Angela Dai @angelaqdai]

Check out @ABokhovkin's Mesh2Tex!

From a real-world image and a shape mesh to texture, we generate high-res and realistic texturing without requiring any matching geometry or pose alignment to the image!
w/ @shubtuls
#ICCV2023
https://t.co/mdaErke58q
https://t.co/U8UV5YKTKJ

[Angela Dai @angelaqdai]

Looking for a challenging dataset for novel view synthesis and 3D semantics?

Check out ScanNet++ at #ICCV23 (Oral)!
460+ scenes w/ 1mm laser scans, semantics, DSLR images, iPhone RGBD video
@chandan__yes @liuyuehcheng @MattNiessner
https://t.co/3oe5yXUEkM
https://t.co/8WCLF7C9n4

[Matthias Niessner @MattNiessner]

Camera frame registration with little or no overlap?

Check out @cangumeli's ObjectMatch: We leverage object semantics to find indirect correspondences between frames. Great for any SLAM and SfM pose optimization!

Vid: https://t.co/NowioCRTqq
#CVPR2023 @angelaqdai

[Matthias Niessner @MattNiessner]

(1/2)
Excited to share "Learning Neural Parametric Head Models" #CVPR2023!

We capture over 5200 high-quality 3D human head scans from which we build a neural parametric head model that disentangles & expressions and deformations.

https://t.co/ynlcqd8ynr
https://t.co/npsZulVUHO

[Matthias Niessner @MattNiessner]

📢Panoptic Lifting for 3D Scene Understanding with Neural Fields #CVPR23 highlight!

Given only posed RGB images of a scene, we optimize a panoptic radiance field representing color, depth, semantics, and instances at any point in space.

Vid: https://t.co/vFpb02zsZQ
@yawarnihal

[Idil @idilsulo]

On weird LaTeX packages today: I found this cute little snowman package to make your day ☃️

https://t.co/87ivaWeYnk

[Dave Zhenyu Chen @davech2y]

❓ Can a single neural model solve BOTH visual grounding and dense captioning in RGB-D scans?
✅ We present a self-critical speaker-listener architecture to unify both tasks.
📢 Come and chat today at 15:50 in Hall B 32

#ECCV2022

[David Rozenberszki @david_roz_]

Gonna present a new benchmark ScanNet200 and our language grounded pretraining strategy at today's @ECCV2022 poster session starting at 11 am! A huge thanks for my amazing supervisors @angelaqdai and @orlitany! Come and chat if you are interested