360 Diffusion LoRA for Flux

Overview

This LoRA was trained for 24,000 iterations with rank 64 on a diverse dataset of over 2000 images, for the purpose of creating 360 degree panorama images. Generally the seam is almost nonexistent and polar distortion artifacts are rare, but some careful editing or in-painting may be required to make the image appear seamless.

Usage

There is no trigger phrase, but as with all 360° Flux LoRAs, you should include some variation of '360 degree equirectangular view' in your prompt to improve output quality.

When rendering images, it's recommended that you choose a 2:1 aspect ratio, like 1024x512, 1536x768, or 2048x1024. Afterwards, you can use an upscaler of your choosing to make the resolution high enough for skyboxes and VR.

Additional Tools

HTML 360 Viewer

To make the viewing and sharing of 360 images & video easier, I built a web browser based HTML 360 viewer that runs locally on your device.

• You can try it out here on Github Pages: https://progamergov.github.io/html-360-viewer/

  • Github source code: https://github.com/ProGamerGov/html-360-viewer

• You can append ?url= followed by a link to your image in order to automatically load it into the 360 viewer, making sharing your 360 creations extremely easy.

• Example: https://progamergov.github.io/html-360-viewer/?url=https://upload.wikimedia.org/wikipedia/commons/7/76/Dauderi.jpg

Recommended ComfyUI Nodes

• ComfyUI_preview360panorama

  • For viewing 360 images & video inside of ComfyUI (may be slower than my web browser viewer).

  • Link: https://github.com/ProGamerGov/ComfyUI_preview360panorama

• ComfyUI_pytorch360convert

  • For editing 360s and for applying circular padding to models in order to improve output quality.

  • Link: https://github.com/ProGamerGov/ComfyUI_pytorch360convert

Limitations

The proportion of training images containing humans was limited to make training faster for version 1.0.