Genur.art

Using Flux Kontext with quantized models, no lora: Flux_kontext_Q5_1 (gguf quantized) clip_l (default) t5xxl_fp8_e4m3fn (default quantized) vae (default) No Lora Parameters: Using two chained ReferenceLatent nodes in the conditionning pipeline. cfg: 1.0 Flux_guidance: 2.5 Workflow intermediate step1: dress the model in a bikini (1 ReferenceLatent) Workflow step2: transfer clothing (2 ReferenceLatent) and: Using ClipSeg to detect the person and highlight the subject's contour in red Optional LLM prompt refining (see prompt below)