Launching today

VISTAlabs

Launching today

Exploring physical spaces as an authentication signal

0 followers

Exploring physical spaces as an authentication signal

0 followers

Visit website

Explore room images as depth, 3D points and spatial reconstructions. VISTAlabs is an Astra assisted research preview investigating physical geometry for same place verification with interactive demos, research notes and early access.

Free

Launch tags:Artificial Intelligence•Tech•OpenAI Day

Launch Team / Built With

OpenAI

GPT-6 Astra Challenge

What became possible in your product with Astra that was not practical before?

Maker

Astra made it practical to connect our existing spatial-reconstruction research to a complete, inspectable product demonstration. We already had pieces of a Blender/Unreal data generation and automation system. With Astra, we could work through a sustained loop: inspect reconstruction failures, form hypotheses, target data or geometry changes, run experiments, evaluate outputs, and repeat. It also helped build and verify the launch pipeline: aligned RGB/depth/point-cloud/Gaussian visualizations, cinematic camera paths, reusable video assets, an interactive research website, and a validated early-access form. The result is VISTAlabs: a research preview exploring physical geometry as a signal for same-place verification. Astra helped us turn fragmented experiments into a coherent demonstration with explicit evidence boundaries; independent metric accuracy, authentication performance, and spoof resistance remain future evaluation work.

Report

3h ago

Maker

📌

Hey Product Hunt! 👋 We’re building VISTAlabs around a question: can the same physical place be reconstructed consistently enough to become an authentication signal? The hardest part was the data. Studying rooms across viewpoints, lighting and object changes requires controlled geometric supervision. Before Astra, we had already been building pieces of a Blender / Unreal data generation and automation system. Astra helped us work through the research loop: identify a failure, form a hypothesis, target a data or geometry change, run an experiment, evaluate, and repeat. Today we’re sharing an interactive research preview. You can explore RGB, depth, derived normals, 3D points, camera geometry and downstream Gaussian visualizations across three indoor demonstrations, plus a separate single-image reconstruction path. The visual demonstrations use retained pipeline outputs and saved camera views. The website lets you inspect those results; it isn’t running a live reconstruction or authentication service in your browser. Our longer term direction is repeatable spatial templates and smaller models suitable for consumer devices. Authentication performance, independent metric accuracy and spoof resistance still need dedicated evaluation. We’d love your feedback: what evidence would you need to trust physical place verification, and where could it be useful? Thanks for exploring VISTAlabs!

Report

12h ago

OpenAI

GPT-6 Astra Challenge

What became possible in your product with Astra that was not practical before?

Maker

Report

3h ago

Maker

📌

Report

12h ago