So over on Reddit, PromptGeek shared a video detailing his findings for what works with photorealistic images. He was kind enough to create a PDF with the details, and I took that and put the findings in a dynamic prompt wildcard file, and created a prompt based on it.

Have fun!

An example prompt

${subject=he-man} 
${features=with immaculate shoulder-length blonde hair}
${details=with a fervent look in his eye}
${pose=brandishing a sword above his head}
${setting=a rocky crevasse}
${background=greyskull castle looms in the background}

__promptgeek/prompt__

The PromptGeek template

[STYLE OF PHOTO] photo of a [SUBJECT], [IMPORTANT FEATURE], [MORE DETAILS], [POSE OR ACTION], [FRAMING], [SETTING/BACKGROUND], [LIGHTING], [CAMERA ANGLE], [CAMERA PROPERTIES], in style of [PHOTOGRAPHER]

The underbelly

__photorealistic/style__ photo of ${subject:_},
${features:_},
${details:_},
${pose:_},
${action:_},
__photorealistic/framing__,
${setting:_},
${background:_},
__photorealistic/lighting__,
__photorealistic/camera/angle__,
__photorealistic/camera/properties/shot_on**__,
{1-$$__photorealistic/camera/properties/film_type__
|__photorealistic/camera/properties/lens__
|__photorealistic/camera/properties/filters_and_effects__},
{3::_|1::in the style of __photorealistic/photographer__}