One subject only
Crowd prompts turn into mush at small sizes. Keep one face in focus and let the town blur behind it.
Describe the setting, the gear and the light, and get a cowgirl portrait that still reads as a person when the frame is cropped tight.

Prompt Guide
Model
Reference Image

Three steps, one subject, no crowd
Say who stands in the frame: age, build, what they wear, what their hands are doing. One person beats five.
Pick the street, the canyon or the saloon, then add the light: low sun, lantern, neon, storm. Light decides whether the shot reads as western.
Check the grip on the revolver and the reins. Reroll or crop tighter if a thumb merges into the trigger guard.
Ten frames in four moods, from a night camp under stars to a neon saloon at closing time.
PortraitsA cowgirl riding a dark horse across open desert at sunset, long shadow ahead of her, dust cloud behind, low sun flare, wide cinematic framing, warm amber palette.
Try This Style
OutlawsThree cowboys standing shoulder to shoulder in a dusty field at sunset, backlit silhouettes, one with a sheriff star on his chest, hats low, wide shot, orange sky.
Try This Style
PortraitsA cowgirl holding two revolvers crossed at her chest, black studio backdrop, hard key light from the left, three-quarter framing, deep shadows, cool grey highlights.
Try This Style
PortraitsClose-up of a freckled cowgirl in a straw hat leaning on a weathered fence rail, late afternoon sun raking across her face, dust in the air, 85mm portrait, warm film tones, 3:4.
Try This Style
OutlawsFour riders silhouetted on a ridge above a valley town, reins loose in their hands, sunset behind them, dust haze, backlit wide composition, desaturated earth tones, 3:4.
Try This Style
OutlawsA dust-covered outlaw leaning on a post outside a marshal's office, hat pulled low, torn poncho, hard midday sun and long shadows on the plank boardwalk, full-body vertical frame, 3:4.
Try This Style
SaloonA cowgirl dealing cards at a saloon table, whiskey glasses and scattered chips, warm lantern light from above, smoke curling through the beam, three-quarter shot, amber palette, 3:4.
Try This Style
SaloonA barmaid pouring from a bottle behind a long wooden bar, mirrored shelves and bottle glow behind her, patrons blurred in the foreground, low amber light, medium vertical shot, 3:4.
Try This Style
RangesA cowgirl leading a pack horse across a shallow river at dusk, water breaking around the hooves, cottonwoods on the far bank, soft pink sky, natural light, vertical framing, 3:4.
Try This Style
RangesCattle moving along a dusty trail at first light, two riders holding the flank, long grass and a wire fence beyond, cool blue shadows with warm highlights, vertical composition, 3:4.
Try This StyleDecisions that keep the frame honest
Crowd prompts turn into mush at small sizes. Keep one face in focus and let the town blur behind it.
Bandana, holster, chaps and a hat with a worn brim do more work than a costume list. Name four pieces at most.
Lantern, low sun and storm light keep a scene in the 1880s. Studio flash reads as a modern costume shoot.
Revolver grips and bridles are where generations fail. Crop to the face or reroll before you download.
A little haze gives depth and hides a flat background. It also softens the seam where ground meets sky.
3:4 suits a single figure, 16:9 suits a canyon or a street. Pick before you write, not after.
I needed a portraits concept the same afternoon I found this page. I typed what I wanted for "Sunset Crossing", ran it twice, then changed only the light and the palette on the second pass. The version I kept went straight into a client deck and saved me a stock-photo search.
My first try for "Ridge Posse" came back too polished, more advertising than real photograph. I rewrote the sentence to describe the surface and the actual light in the room, and the third attempt looked like something I could have shot myself. That is the one I still use for ai cowgirl generator mockups.
I use it the way I use a sketchbook: one prompt for "Cards and Lantern", one for "Behind the Bar", then I pick the half that works and iterate from there. Two of those frames ended up in a package for a client who could not decide between the two directions, and seeing them side by side settled it in a minute.
Both are possible. With a reference image of yourself you get your features in the costume; with text only you get a generic cowgirl. If it has to look like you, upload and keep the face description short.
Text prompts cannot recover a face. The model invents features unless you give it a photo. Keep hair colour, build and one distinctive detail in words, and expect to re-roll two or three times.
Yes, as long as it does not impersonate a real person. Instagram and TikTok now ask creators to label AI-generated profile images, so disclose it and do not present it as a photo of yourself.
Describe clothing as clothing: denim jacket, felt hat, boots, long sleeves, and name the setting. Vague wording about a sexy western look is what pushes an image toward filters and failed generations.
The free allowance in the picker covers testing; paid models start around $5.80 a month. Nothing is watermarked.