Stable Diffusion Prompt Tricks

The content is taken from this Reddit post. The post is about how to enhance your prompt image generation. We added some examples made with Stable Diffusion.

Canon50:

Prompt: Avocado, Canon50

Makes the picture into a camera photograph

Forces picture to be realistic.

Fantasy characters become pictures of kid toys.

Rendered by octane:

makes it movie-like. Cinematic effects. Can be more Disney like.

8k:

Avocado, 8k

It does increase the definition slightly, though can lead to it looking more fake.

It makes the image more camera-like and realistic-

Close-up:

Ninja, Close-up

Zooms into the face and upper body.

Seductive:

Makes females more adult, and adds quality breasts to most photos.

Symmetric face:

halloween witch woman, Symmetric face

Certainly focuses the model on the face and does clearly enhance its symmetry.

Often enhances the symmetry of the body as well.

Somehow leads to more issues with the face, sometimes?

By art germ:

digital illustration of neotokyo cyberpunk cityscape, by art germ

stunning modifier. Makes the colors power out. Auras everywhere. The lighting becomes spectacular. The reliable dramatic increase in photo quality.

(Removes the need for symmetric face)

Dramatic cinematic lighting:

digital illustration of neotokyo cyberpunk cityscape, dramatic cinematic lighting

Creates strong effects on the skin and background lighting. Overall increase the potential of the image.

Removing ‘dramatic’ does reduce the quality of the image.

Removing ‘lighting’ actually seems to increase the quality of the image.

Dramatic

Increases the emotional expressivity of the face. Overall substantial increase in photo potential / variability. +1 for variability, important for getting the max hit.

Cinematic

“Cinematic” makes the image more movie-like.

Golden Corset

Sometimes generates dresses

Often generates golden jewelry (but perhaps this is because it is a princess I’m generating)

Portrait

Portrait of Arnold Schwarzenegger

Focuses image on the face / headshot.

Increases photo quality.

Improves lighting on face / skin (this is what I was seeing and missing from the great one)

Makes 1/4th of images usable

Making portrait primary focuses the shot on the upper body and face.

wlop and ross tran:

Makes images intense in a surprising way.

Less kiddie than Artgerm, more adult.

Adds scenery in many cases – water, housing.

Totally unclear that this is an improvement in this case, unlike art germ.

wlop

Adds some rich scenery.

Adds lipstick consistently.

Adds a lot of makeup.

Ross tran:

More movie-like

facially intense

shimmers and glimmers in scenes.

Does seem like an improvement.

Hyper realistic

Definitely makes the faces more likely to be real, less fantasy world.

Make the closing look like clothing, rather than art. Clothing is less perfectly fit.

Young with long hair

Arnold Schwarzenegger, Young with long hair

This definitely makes the images more realistic.

Overall, this seems to decrease the average quality of the images.

Maybe 1/9 photos hair visibly improves.

In a cathedral

Makes the woman look more realistic.

Adds cathedral surroundings in most portraits.

Putting the woman ‘in a cathedral’ takes the photos to the next level, reliably.

Vivacious

Images become more finely detailed, and the woman’s body language improves dramatically

Moving vivacious to the front increases the sensuality of the poses but damages face quality

Very sexy pose

Huge improvement in head tilting

Proportions start to get out of whack

eyes angles with tilted head over breasts start to show up, same head angel as in saved portrait

Stable Diffusion Prompt Tricks
Scroll to top