FLUX LORA ENHANCER/FACE DETAILER

This ComfyUI workflow takes a Flux text prompt and LoRA settings to create portraits or character-focused images, then uses Face Detailer to reinforce character traits and enhance facial details. It produces an initial image and a refined portrait.

It generates an initial image from the text prompt and LoRA settings, then applies Face Detailer using those settings.

Without a character LoRA, it can work as a standalone Face Detailer to sharpen and refine a portrait.

Inputs include a Flux text prompt, LoRA settings, and optionally a character LoRA.

Outputs include an initial image and a portrait refined by Face Detailer.

Use it in ComfyUI with Flux.1 D as the base model.

The listing provides links to required files and optional enhancements.

The listed model files are ae.safetensors, bbox/face_yolov8m.pt, clip_l.safetensors, flux1-dev.safetensors, sam_vit_b_01ec64.pth, and t5xxl_fp8_e4m3fn.safetensors.

Listed node types include AV_FaceDetailer, BasicGuider, BasicScheduler, CLIPTextEncode, ConditioningZeroOut, DualCLIPLoader, EmptyLatentImage, FluxGuidance, Image Comparer (rgthree), and KSamplerSelect.

They also include Power Lora Loader (rgthree), RandomNoise, SAMLoader, SamplerCustomAdvanced, SaveImage, UltralyticsDetectorProvider, UNETLoader, VAEDecode, and VAELoader.

The workflow includes guidance on general and Face Detailer settings, including how each component affects the final result.

It includes Flux prompting tips and names the LoRA model as “Vanessa” from random flux girlfriends by TechnicianSpecial666.

Before using it, verify that the listed model files and node types are available in your ComfyUI setup.

Suggestion · non vérifié

The listing does not specify a license.