v2.1 Update: HighSodium Optimization & Custom Depth Maps

Immersive 3D Generation
Inside ComfyUI

Transform any 2D image or video into stunning Side-by-Side (SBS) stereoscopic 3D. Powered by Depth-Anything (V2 & V3), built for VR headsets, 3D displays, and cross-eyed viewing.

Unmatched 2D to 3D Capabilities

Everything you need to create production-ready 3D content directly within your existing AI workflows.

AI-Powered Depth

Utilizes state-of-the-art Depth-Anything (V2 & V3) models for automatic, pixel-perfect depth estimation without external maps.

HighSodium Optimization

Toggle the performance mode to utilize a vectorized algorithm, resulting in up to 4x faster rendering speeds for bulk tasks.

Full Video Pipeline

Includes dedicated nodes for video uploading, frame-by-frame processing, and recombination. Turn flat MP4s into spatial video.

Granular Control

Resolution-relative depth scaling prevents artifacting. Adjust depth scale, blur radius, and easily switch between Parallel or Cross-eyed modes.

External Depth Maps

Already have a depth map from another node? The V2.1 engine allows you to feed custom depth data directly into the fast generator.

How the SBS Workflow Operates

Creating VR-ready content has never been easier. ComfyUI_SSStereoscope integrates directly into your node graph. By feeding an image into the main SBS V2 node, the system utilizes advanced machine learning to estimate distance and depth.

  • 1. Input Pass your AI-generated art, photograph, or video frame into the node.
  • 2. Processing The node calculates the offset for the left and right eyes based on your specified depth_scale.
  • 3. Output A pristine Side-by-Side (SBS) image is generated, ready to be viewed on the Meta Quest, Apple Vision Pro, Xreal glasses, or directly on your screen using the cross-eye technique.

Quick Start Installation

Install via ComfyUI Manager or manually via the command line.

cd ComfyUI/custom_nodes
git clone https://github.com/SamSeenX/ComfyUI_SSStereoscope.git
cd ComfyUI_SSStereoscope
pip install -r requirements.txt

Note: The AI depth model will automatically download on its first run. Requires Python >= 3.8.