AIToday
Image GenerationTop Companies' AI MovesTop Companies AIPublished: Nov 22, 2026, 06:30 JST

Apple's Depth Pro 2 handles photos and video in one model

Send on LINE
Apple's Depth Pro 2 handles photos and video in one model

3 Key Points

  1. What happened

    Apple researchers Peiyun Hu, Marko Mihajlovic, Lyndon Duong, Lahav Lipson, Wei Dong, Amaël Delaunoy, Stephan R. Richter and Vladlen Koltun present Depth Pro 2, a single network that predicts depth in meters for a photo, video, or live feed at the input's full resolution.

  2. Why it matters

    Prior multi-frame methods favored temporal consistency while single-frame methods favored boundary sharpness; Depth Pro 2 is presented as achieving both at once, which could simplify setups that need stable depth across frames without losing edge detail.

WHO IT HITSDevelopers and research teams building depth sensing into camera apps, video pipelines, or AR prototypes may find a single model that covers stills and video, reducing the need to pair separate single-frame and multi-frame systems.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, every morning.

Context & Analysis

Depth Pro 2 arrives as a follow-up to Apple's earlier Depth Pro, which synthesized high-resolution depth maps with absolute scale without relying on camera metadata such as intrinsics, producing a 2.25-megapixel depth map in 0.3 seconds on a standard GPU. The new work positions itself against a split in monocular depth estimation: multi-frame methods favor temporal consistency, while single-frame methods favor boundary sharpness. By training one scalable network on both images and video and predicting at native resolution, the team says it avoids that trade-off. Apple also points to related work, including SHARP, which regresses a 3D Gaussian representation from a single image in under a second on a standard GPU and can then be rendered in real time. Depth Pro 2 is released with code and weights, suggesting the team intends outside developers to build on it rather than keep it internal.

FAQ
What inputs can Depth Pro 2 handle?
One network handles a single photo, a whole video, or a live camera feed, always at the input's full resolution.
Is Depth Pro 2 available to use?
Yes. Apple has released the code and weights for a family of models.
How is Depth Pro 2 different from prior monocular depth methods?
Prior methods commonly trade off temporal consistency and boundary sharpness: multi-frame methods favor consistency, single-frame methods favor sharpness. Depth Pro 2 is presented as achieving both.
Top Companies AIRead Original Article
Send on LINE

AI news that matters for your work, delivered every morning.

Pick your industry and the AI tools you use, and get news related to your work every day.

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.

Questions and answers are published on this page.

Related Articles

Next articleBlackRock's Mitchnick Saw Bitcoin Inversely Tracking AI Stocks — Then Both Fell