README.md

---
title: FE2E Depth + Normal (CPU)
emoji: 🗺️
colorFrom: blue
colorTo: green
sdk: gradio
sdk_version: 6.14.0
app_file: app.py
python_version: "3.11"
pinned: false
license: mit
short_description: Depth + Normal estimation from a single image on CPU
tags:
  - depth-estimation
  - normal-estimation
  - cpu
  - mcp-server
---

# FE2E: Depth + Normal Estimation (CPU)

Estimate depth and surface normals from a single image using [FE2E](https://github.com/AMAP-ML/FE2E) (CVPR 2026).

- Model: Step1X-Edit DiT + LDRN LoRA, pre-quantized INT8 (12 GB)
- Single denoise step (not iterative diffusion)
- Outputs both depth AND normal maps simultaneously
- Takes ~29 min for 768x1024 on free CPU Basic