I recommend using the "AziibPixelMix" model.
LoRAs are your preference, but I recommend Pixel art (there's a girl with short hair on the cover).
And in the advanced parameters I recommend:
Size: 512x
VAE Model: *
Steps: 34
Sampling Method: DDIM
CFG Scale: 5~10
Why use "DDIM"? Because it costs less credits and the images maintain a good quality. Using DDIM you can generate 4x for a low price using 34 Steps
