mo230761

mo230761 / UniGeo

Public

A framework for camera-controllable image editing using unified geometric guidance and video models.

46
6
100% credibility
Found Apr 28, 2026 at 46 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

UniGeo enables precise camera-controlled image editing by turning text descriptions of movements into new viewpoints using video generation models.

How It Works

1
📰 Discover UniGeo

You find UniGeo, a fun tool that lets you edit photos by adding smooth camera movements like panning or zooming using simple text descriptions.

2
📸 Gather your photos

Pick your favorite landscape or portrait images that are wider than tall, ready for cinematic edits.

3
✏️ Describe the camera action

Write easy phrases like 'Camera pans left by 16 degrees' or 'Camera tilts up by 7 degrees' for each photo in a simple list.

4
🔍 Preview the motion

See a quick 3D point cloud view of your camera path to tweak and perfect the movement before the magic happens.

5
🚀 Generate edited images

Hit go, and UniGeo blends your descriptions with powerful image tools to create new photos from the ending camera view.

😍 Enjoy your cinematic photos

You now have stunning images with realistic camera shifts, perfect for stories, videos, or sharing online.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 46 to 46 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is UniGeo?

UniGeo is a Python framework for camera-controllable image editing on GitHub, turning static landscape photos into views from new angles via text prompts like "Camera pans left 16 degrees" or "moves forward 2.5 meters." It uses unified geometric guidance from video diffusion models to generate consistent 3D edits—first previewing point clouds, then outputting the final image. Developers get precise control over pan, tilt, and zoom without 3D modeling tools.

Why is it gaining traction?

Unlike pixel-based editors, UniGeo leverages video models like Wan for geometry-aware results, taming inconsistencies in diffusion outputs. The two-step pipeline (prompt-to-point-cloud, then editing) lets users iterate previews fast, and recent updates support 23GB VRAM inference plus a Hugging Face Space demo. On GitHub framework ranking, it stands out for unigeo dataset handling and camera-controllable editing in desktop/laptop workflows.

Who should use this?

Vision ML engineers building AR previews from photos, game devs needing quick viewpoint shifts, or researchers extending video diffusion to images. Ideal for unigeo font tweaks in UI mockups or unigeo technology pvt ltd-style geometric estimation tasks.

Verdict

Promising for camera-controllable editing with a strong arXiv paper, but low maturity (46 stars, 1.0% credibility score) means test via HF Space first—docs are README-focused, no extensive tests yet. Worth forking if you're in unigeo taming video diffusion experiments.

(178 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.