Hi, this workflow uses Flux 8-16 Step lora by HyperSD team. This process fastens the generation speed. For me, in 8 steps with 3090 24GB, generation time is arraound 10-13 seconds.

Important Notes:

- This lora doesn't work with NF4 version of the model, so please use with default Flux Dev model.

- Your Lora strength should be 0.125 to 0.16, and the guidance has to be 3.5

- Whether you are using 8-step or 16-step, remember to update the sampling steps according to that.

In this workflow I used 8-step LoRA, but you can use the 16-step LoRA as well.

8-step Lora: https://huggingface.co/ByteDance/Hyper-SD/blob/main/Hyper-FLUX.1-dev-8steps-lora.safetensors

16-step Lora:  https://huggingface.co/ByteDance/Hyper-SD/blob/main/Hyper-FLUX.1-dev-16steps-lora.safetensors