3DGS AI slop is over.
Training 3DGS is now faster than loading the whatever slop 3D model into to GPU, 1M Gaussians in 20s, low memory footprint, insane PSNR.
We are looking for one or two interns working on 4DGS and/or feedforward 4DGS models.
We have elegant rig hardware, looooots of real multi-view cap data, mature data infrastructure, the very fastest sfm/splatting pipeline, collaborators, and okayish local computing resources.
Relocation to Cleveland is required. Recent graduate with a strong interest in 4DGS is preferred. DM me here or on LinkedIn.
Sharing large volumetric microscopy datasets for interactive visualization has always been painful. The files are massive, difficult to move, and hard to render smoothly in real time.
With 3DGS, this can change.
Volumetric microscopy data can now be rendered directly in the browser, with streaming level-of-detail support using standard 3DGS infrastructure (as well as more optimized rasterizer). This opens the door to faster visualization, new analysis workflows, hybrid rendering modalities, and more efficient deep learning pipelines for large-scale microscopy data.
At OpsiClear, we are building both optimization-based and feedforward models that turn terabytes of volumetric microscopy data into manageable point clouds with rich, analyzable features.
The goal is simple: make large-scale microscopy data easier to visualize, share, analyze, and learn from. Instead of voxels, researchers and models can understand the data from a much more compact point cloud).
Reach out if you are interested in collaborating or learning more.
Teaser:
https://t.co/FSAzLPk4bG
Want to catch bad images in your SFM reconstruction?
Introducing ColmapView v0.7 with Gaussian Splatting QA (and more).
- Detects all PLY in your dataset and URL.
- Overlay images exactly at the 3DGS reconstruction.
- Compute PSNR and SSIM in your browser.
- Sort your images based on reconstrction quality.
Plug and play at:
https://t.co/j9Ebve1jSA
Happy to share our work: Real-World Point Tracking with Verifier-Guided Pseudo-Labeling. #CVPR2026
We improve the pseudo-label training pipeline for real-world videos using a verifier that selects the most reliable predictions across multiple trackers.
🔗https://t.co/wIKqTAeFzO
Introducing https://t.co/HfzR0VgI8m v0.5
Beyond mobile support, now you can remove unwanted camera from the COLMAP data and export a copy, edit the camera model for compatibility need, save config presets, export demo videos, and more.
This is video… but interactive 🎥
@ijustine checks out OpsiClear’s 4D capture system, turning video into explorable models for medical education and beyond.
Working with QUEL imaging, we produced a 3DGS digital twin of their 3D printed Digital Mouse Atlas. Unlike traditional 3D scanning, 3DGS natively captures the translucent optical properties of materials.
https://t.co/zrnbKLFHgP
No doubt that 3DGS will be becoming a foundation tools for biphotonic and medical research.
4 simple steps to add 3DGS digital twins to Squarespace, Google Sites & other no-code platforms. Minimal code needed. #3DGS#DigitalTwin#NoCode#Web3D