l2r (Qwen-Image 2.1)
The node pack is required to use this properly.
Generates the right eye for a VR180 left eye. Give it a 2048×2048 half-equirectangular left eye, and it redraws the same scene from the right eye's position. Intended for use with the Flat to 180 VR LoRA to generate a full VR180 image from a flat input.
Limits
It learned an average eye spacing, so depth is plausible but not exact, but from my experience it's still far more convincing than an approach like using DepthAnything.
Some seeds drift, but the node pack's Right Eye (best of N) node retries seeds against stereo checks, and Join Eyes fixes brightness and divergence.

