AIAny
AI Image2026
Icon for item

Sun Direction LoRA (Flux2Klein 9B)

Applies or repositions directional sunlight in outdoor images by using a LoRA trained for Flux2Klein 9B to match a reference sun elevation and rotation. Workflow uses an overcast intermediate and a sphere (ball) reference; includes a ComfyUI node and Blender scene for rendering the reference.

Introduction

Most automatic relighting approaches struggle because existing shadows and highlights encode ambiguous or conflicting light directions. This LoRA encodes explicit sun elevation and rotation so an image-to-image Flux2Klein pipeline can place directional sunlight based on a small reference (a rendered sphere) while preserving scene content—especially useful for exterior scenes and timelapse sequences.

What Sets It Apart
  • Precise sun direction control: trained to match a reference sun elevation and rotation, so you can specify where the sun should be relative to the camera rather than relying on vague prompt hints.
  • Overcast-intermediate workflow: generates an overcast version of the target first to remove preexisting directional cues, so the relighting step produces more coherent shadows and shading.
  • Sphere reference + tooling: uses a rendered ball as the light-direction reference and provides a ComfyUI Sphere-Light node and a Blender scene, so users can generate consistent references without manual guesswork.
  • Lightweight integration: implemented as a LoRA for Flux2Klein 9B, enabling image-to-image relighting without fine-tuning a full backbone model; however, it was trained mainly on exteriors and may not generalize to indoor scenes.
Who it's for + tradeoffs

Great fit if you are an artist, environment/lighting artist, or photographer who needs to change sun direction in outdoor renders or photos and wants reproducible, seed-stable timelapse relighting. Look elsewhere if you need robust interior relighting, fine-grained control over light hardness/color/intensity (v1 focuses on direction), or a solution that avoids an overcast intermediate step. The package includes workflow nodes and reference assets but relies on Flux2Klein 9B and the image-to-image pipeline for inference.

Information

Categories

More Items

Hugging Face
AI Model2026

Combines text-to-image generation and reference-guided editing in one model, producing images around 16 megapixels and accepting up to 16 reference images.

Hugging Face
AI Image2026

Turns text prompts or up to eight reference images into high-resolution images, with support for editing. An eight-step sampler and built-in prompt upsampler streamline generation, though bf16 inference requires substantial GPU memory.

Hugging Face
AI Image2026

Generates high-resolution images and edits multiple references in ComfyUI in 8 steps. Based on MiniMax-H3, it offers int8 and w4a8 variants, with the smallest setup using about 21 GB of peak VRAM at 8 MP.