# Tsinghua University releases PPMem, a plane-based memory for AI video models

Digest AI · Generative AI & Models · published 2026-10-20T15:43:40Z

Canonical: https://digestai.news/story/tsinghua-university-releases-ppmem-a-plane-based-memory-for-ai-video-m

## Summary

Tsinghua University announced Persistent Planar Memory for Video World Models (PPMem), a new approach that stores 3D memory as compact plane primitives instead of dense point clouds. The system builds on the Wan 2.2‑5B video diffusion model and was trained on more than 80,000 clips from DL3DV and RealEstate10K, using depth data from MoGe‑2 and VGGT reconstructions. In tests, PPMem outperformed Gen3C, Voyager, VMEM and SPMem on DL3DV and RealEstate10K, achieving higher PSNR, LPIPS, CLIP and DINOv2 similarity scores and preserving scene consistency over 300–400 frames with a 0.27‑point margin at frame 300. The authors note that after about 400 frames an entire scene can be represented by roughly 100,000 plane primitives, occupying only 5 MB of memory—about a thousand times more compact than a dense point cloud.

The paper argues that plane primitives provide a lightweight, structured memory that is orders of magnitude smaller than per‑pixel representations, enabling longer, more consistent AI‑generated video while reducing storage and inference costs. The authors suggest that this method could help future AI models retrieve and reference fixed assets more efficiently, though they acknowledge that the approach still relies on a “breadcrumb” strategy rather than a full RAG system.

## Key points

- PPMem uses plane primitives to store 3D memory, reducing 100k planes to 5MB, outperforming Gen3C, Voyager, VMEM, and SPMem on DL3DV and RealEstate10K.
- The system builds on Wan 2.2‑5B, training on 80,000 clips from DL3DV and RealEstate10K, with depth from MoGe‑2 and VGGT reconstructions.
- PPMem preserves scene consistency over 300–400 frames, achieving higher PSNR, LPIPS, CLIP and DINOv2 similarity than competitors, with a 0.27 margin at frame 300.

## Why it matters

By compressing 3D memory into plane primitives, PPMem enables longer, more consistent AI‑generated video with far less storage, potentially easing deployment on edge devices and improving realism for virtual environments.

## Sources

1. [Giving AI Video Some Remembrance of Things Past](https://unite.ai/giving-ai-video-some-remembrance-of-things-past) (Unite.AI, 2026-10-20)

## Cite

Digest AI, "Tsinghua University releases PPMem, a plane-based memory for AI video models", 20 October 2026, https://digestai.news/story/tsinghua-university-releases-ppmem-a-plane-based-memory-for-ai-video-m

---

Written by Digest AI's editorial model from the linked sources; the sources are the record. Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse
JSON: https://digestai.news/story/tsinghua-university-releases-ppmem-a-plane-based-memory-for-ai-video-m.json