Sumitc13

These are LoRAs trained to control the exposure of images generated with FLUX.1 Kontext (dev), a state-of-the-art image editing model from Black Forest Labs. These LoRAs allow you to increase or decrease the exposure value (EV) of the output image. - To decrease exposure, use the LoRAs with negative EV values and their corresponding trigger words (e.g., `ev-2` for `ev-2.safetensors`, `ev-4` for `ev-4.safetensors`). - To increase exposure, use the LoRAs with positive EV values and their corresponding trigger words (e.g., `ev+2` for `ev+2.safetensors`, `ev+4` for `ev+4.safetensors`). - Adjust the LoRA weight to control the intensity of the effect. A weight of 1 is a good starting point. - Trigger words are the same as the LoRA name without the extension (e.g., `ev+2`, `ev-2`). These LoRAs work perfectly with ComfyUI to generate multiple exposure images that can be combined into HDR (High Dynamic Range) images. HDR Processing with ComfyUI For creating professional HDR images using these exposure LoRAs, use the Luminance Stack Processor custom nodes: This custom ComfyUI node allows you to: - Merge different exposure level images (generated with these LoRAs) - Create high dynamic range images using advanced algorithms - Export professional 32-bit EXR files with preserved HDR values - Use the Radiance Fusion Algorithm for enhanced HDR processing Workflow Process: 1. Generate images with different LoRAs (`ev-4`, `ev-2`, `ev+2`, `ev+4`) 2. Use the Luminance Stack Processor nodes to combine exposures 3. Export as HDR EXR files with full dynamic range preservation Complete ComfyUI workflow examples are available in the repository's `/workflow` directory. Complete workflow setup showing how to use the exposure LoRAs in ComfyUI Comparison showing the same bulb scene with different EV settings: Original, EV+4, EV+2, EV-2, EV-4 Comparison showing the same landscape with different EV settings: Original, EV+4, EV+2, EV-2, EV-4 Example showing reduced highlight clamping for better dynamic range preservation > Note: These comparison strips demonstrate how each LoRA affects the same scene with different exposure values, from underexposed (EV-4) to overexposed (EV+4).