>>3079413
>>3079414

You don't give enough information to give you good advice.

What graphics card do you have?

Is everything installed on a Solid State Drive?

What starting resolution?

What is your upscale % and Denoising strength?

How many Sampling steps, for both first gen and Hires?

What Sampler method and Scheduling type?

What user interface are you working with?

What type of checkpoint (SDXL ; SD 1.5 ; Flux ; etc.) are you are using?

What ADetailer models are you activating?

How many Loras are you trying to add?

All of those things change the amount of time you will need. So you have a baseline, my setup is:

4060 TI with 16 gigs Vram ; Everything on SSD ; Reforge (which is no longer updated, but is about 10% faster then reg forge) ; SDXL Checkpoint (illustrious) that needs 6.5 gigs. One lora that needs 220 MiB

Starting resolution of 1024*1024 with 50 steps using "DDPM" Sampling method and "SGM Uniform" Schedule type.

Hires.fix at 35 steps with 0.4 Denoising strength, upscaled by 1.5* = 1536*1536 pixels

2 Adetailer models: "Anzhc Face segmant v2" and "CockandBallYolo8x".

First generation @ 50 steps takes 24 sec. HiresFix 40% strength @ 35 steps takes another 42 sec. Each ADetailer detection adds about another 10 secs.

It usually takes about 100 to 120 secs per picture generated.