AnalysisOpinion / analysisImage, Video & 3D Generation · Large Language Models · Training & Scaling1 source · Oct 7, 2026

Show HN: TerrainSR – fast, realistic heightmap upscaling model

I wanted to have a 1:1 scale model of Europe, but my problem was that 100m data was too low-res while 10m LIDAR data was patchy, took hundreds of GBs to store and was full of manmade objects like mines, buildings and so on.

ProofWeights ↗1 community thread

Key points

  • This is a model that I made for a historical game.
  • I trained this model on undeveloped landscape so that it can quickly add plausible erosion features, rocks, etc to the low-resolution height data and sort of reconstruct what the terrain would look like before any human interference.

Sources (1)

  • [1]Show HN: TerrainSR – fast, realistic heightmap upscaling model
    Hacker News: LLM threads (40+ points) · Oct 7, 01:24 AM
    I wanted to have a 1:1 scale model of Europe, but my problem was that 100m data was too low-res while 10m LIDAR data was patchy, took hundreds of GBs to store and was full of manmade objects like mines, buildings and so on.
    This is a model that I made for a historical game.

Extractive summary: sentences quoted from the sources.

Before this

  1. Oct 6, 2026Introducing Mistral Large 4
  2. Oct 6, 2026[AINews] Reflection Beam - 501B-A23B American Open Model

Related