Hyojun Go

ELLIS PhD Student @ ETH / Google

[Hyojun Go @gohyojun3]

Our recent finding on Diffusion Alignment: a reward model in pixel space can be easily transferred to score noisy diffusion latents directly — at small finetuning cost, via stitching.

This makes Faster & Better for both Training & Inference Alignment.

Meet StitchVM👇

1/

[Grigory Bartosh @GrigoryBartosh]

🚀 Excited to share my @GoogleDeepMind student researcher project: Dual-Rate Diffusion✨

⚡ A simple construction that speeds up both regular diffusion and distilled models by interleaving a heavy context encoder with a light conditional denoiser.

🧵👇

[Hyungjin Chung @hyungjin_chung]

For alignment you need V, but is hard to compute. Most methods try to approximate with

1) Tweedie, which is biased
2) MC roll-outs, which is slow with high var.

Training V was often neglected since it's hard.

We beg to differ. StitchVM enables this! Led by @gohyojun3

👇

[Hyojun Go @gohyojun3]

For more, check out our work 👇

@hyungjin_chung , @prunetruong , Goutam Bhat, @ZhaochongAn , @zixiangzhao_ , @DNarnhofer , @fedassa , @SergeBelongie , Konrad Schindler

paper: https://t.co/qeVubJM2Om
Page: https://t.co/ZCzaYLQRIl

[Hyojun Go @gohyojun3]

Our recent finding on Diffusion Alignment: a reward model in pixel space can be easily transferred to score noisy diffusion latents directly — at small finetuning cost, via stitching.

This makes Faster & Better for both Training & Inference Alignment.

Meet StitchVM👇

1/

[Hyojun Go @gohyojun3]

Result 4️⃣ — Training-time alignment with DRaFT & DiffusionNFT

No need for full rollouts. Just stop denoising at an intermediate step and use StitchVM's inference as the reward signal.

Now we have much faster convergence

7/

[Yuanwen Yue @YueYuanwen]

Want a lighter yet stronger Point Transformer? Meet LitePT ✨

LitePT is a lightweight, high-performance 3D point cloud architecture for a wide range of point cloud processing tasks. Our smallest variant LitePT-S, features 3.6× fewer parameters, 2× faster runtime and 2× lower memory footprint than PTv3, while already matching or outperforming it across a range of benchmarks.

💻Code: https://t.co/WtMSKJHfRB
🌐Project page: https://t.co/FEQbeOtHUB
📰Paper: https://t.co/8cQcS4Nvtt

with Damien Robert, @jianyuan_wang , Sunghwan Hong, Jan Dirk Wegner, Christian Rupprecht, and Konrad Schindler

[Michael Niemeyer @Mi_Niemeyer]

Combining video diffusion and 3D feedforward models by simply stiching them together in latent space - very cool idea! Make sure to check out this novel work from my collagues at Google and ETH!

[Dominik Narnhofer @DNarnhofer]

Want to leverage the power of SOTA 3D models like VGGT & Video LDMs for 3D generation? Now you can! 🚀
Introducing VIST3A — we stitch pretrained video generators to 3D foundation models and align them via reward finetuning.

📄 https://t.co/MctMyuDev4
🌐 https://t.co/XQMW4mfjWI

[Prune Truong @prunetruong]

🎺Meet VIST3A — Text-to-3D by Stitching a Multi-view Reconstruction Network to a Video Generator.
➡️ Paper: https://t.co/sFqbbUiGOO
➡️ Website: https://t.co/QWMLwXyVcB

Collaboration between ETH & Google with Hyojun Go, @DNarnhofer, Goutam Bhat, @fedassa, and Konrad Schindler.

[Hyungjin Chung @hyungjin_chung]

Even the SOTA VideoLLMs see videos in 1 fps, and you CANNOT perceive fine-grained motion 💃 with this frequency 🥲

📣 Presenting Video Parallel Scaling (VPS), an inference-time strategy that lets VideoLLMs see more frames by scaling compute in the parallel-axis 🤩

[Hyungjin Chung @hyungjin_chung]

Excited to share that 3 papers are accepted to #ICCV2025 at EverEx 🎉

📌 SteerX: https://t.co/t9BQouecfI
📌 VideoRFSplat: https://t.co/rSrlGiZvCv
📌 CapeLLM: https://t.co/Sh75vXsQID

See you in Hawaii 🌴

👇 link to some threads

[Hyungjin Chung @hyungjin_chung]

🚨Introducing VideoRFSplat📽️, a feed-forward text-to-3DGS generative model with high-quality scene-level results without post-optimization (e.g. SDS)

Led by collaborators at EverEx AI - @gohyojun3, @bypark___, @namhyelin99, Byung-Hoon

https://t.co/jfzlfejO8A

A 🧵 👇

1/n

[Zhenjun Zhao @zhenjun_zhao]

SteerX: Creating Any Camera-Free 3D and 4D Scenes with Geometric Steering

@bypark___, @gohyojun3, @namhyelin99, Byung-Hoon Kim, @hyungjin_chung, Changick Kim

tl;dr: MV-DUSt3R+ and MonST3R->geometric reward functions->geometric consistency

https://t.co/vJSExQFGBV

[Hyungjin Chung @hyungjin_chung]

3D consistent videos are hard to generate 🙁

What if we could steer them to be consistent during generation?

Introducing SteerX🛞, a plug-and-play sampling method that works with *any* video diffusion to make videos physically plausible🤩

w/ @bypark___ @gohyojun3 @namhyelin99

[MrNeRF @janusch_patas]

SplatFlow: Multi-View Rectified Flow Model for 3D Gaussian Splatting Synthesis

TL;DR: SplatFlow is a unified framework that combines a latent-space multi-view generator and a Gaussian Splatting Decoder to enable efficient 3D generation, editing, and inpainting directly from text prompts.

Abstract (excerpt):
SplatFlow comprises two main components: a multi-view rectified flow (RF) model and a Gaussian Splatting Decoder (GSDecoder).

The multi-view RF model operates in latent space, generating multi-view images, depths, and camera poses simultaneously, conditioned on text prompts, thus addressing challenges like diverse scene scales and complex camera trajectories in real-world settings.

Then, the GSDecoder efficiently translates these latent outputs into 3DGS representations through a feed-forward 3DGS method. Leveraging training-free inversion and inpainting techniques, SplatFlow enables seamless 3DGS editing and supports a broad range of 3D tasks-including object editing, novel view synthesis, and camera pose estimation-within a unified framework without requiring additional complex pipelines.

We validate SplatFlow's capabilities on the MVImgNet and DL3DV-7K datasets, demonstrating its versatility and effectiveness in various 3D generation, editing, and inpainting-based tasks.