As a next question you may be thinking what’s the relationship between a Style Reference and Stylize Parameter? How do different values of the Stylize parameter affect images that have a style reference?

Let's test this with both photorealistic and illustrative images.

--sref Image URL

VS

--stylize value

EXAMPLE 1: Photorealism


REFERENCE IMAGE: photo, an astronaut in a futuristic space suit running away from volcanic lava at night, blue, red, orange, black, foggy, transfixing marine scenes, neon sparks --ar 3:2 --v 5.1

REFERENCE IMAGE: photo, an astronaut in a futuristic space suit running away from volcanic lava at night, blue, red, orange, black, foggy, transfixing marine scenes, neon sparks --ar 3:2 --v 5.1

NEW PROMPT:

cinematic, photo, A horrifying archaic creature from the underworld is walking away from volcanic lava at night, motion, blue, red, orange, black, high-speed camera, transfixing marine scenes, neon sparks --sref https://s.mj.run/lsflnUniu-8

--stylize {0,100,250,500,1000}

--ar 16:9 --v 6.0

Untitled

Despite minor differences, I haven't noticed a considerable effect of the stylize parameter on the final images.

To ensure that there is no significant impact, I decided to test the same parameter with an illustrative type of prompt that includes an illustrative type of image reference.

EXAMPLE 2: Illustrative


REFERENCE IMAGE: Painting, robots are working in the field, Beautiful Thailand scenery, sci-fi, Jakub Różalski style --s 750 --ar 16:9 --v 6.0

REFERENCE IMAGE: Painting, robots are working in the field, Beautiful Thailand scenery, sci-fi, Jakub Różalski style --s 750 --ar 16:9 --v 6.0

portrait, a Chinese young woman dressed in traditional clothing for the new year, she wears a red qipao dress and hairpins, she holds a decorative fan, posed indoors standing against a blank background, lighting illuminates her festive outfit, heritage and cultural preservation --sref [<https://s.mj.run/yam9uFmYteQ>](<https://s.mj.run/yam9uFmYteQ>)

--stylize {0,100,250,500,1000}

--ar 16:9 --v 6.0

Untitled