Camera prompts control framing and mood with two ingredients: a shot type (close-up, medium, full body, wide) and an angle (from above, from below, eye level, dutch). Add a focal length like 85mm and depth cues like bokeh to fine tune. The same subject reads completely differently depending on these choices. Keep baseline safety negatives on every render.
Two identical prompts with different camera tags produce two different photographs. Shot type and angle are not decoration, they are composition, and composition is what separates an amateur render from one that looks like it came from a real shoot. Most people leave the camera slot empty and let the model default to a flat, eye level, medium shot every time. That is why so much AI output looks the same.
This is a practical library of camera, shot, and lens phrasings, what each one does to the image, and copy and paste examples for every model family. Every subject is an adult (18+), fictional, AI-generated character, never a real identifiable person, never a minor or minor-appearing subject. Baseline safety negatives stay on every single render.
Test any framing live in our generator.
The two layers of camera language
Think of camera prompting as two layers that stack.
Shot type is how much of the subject is in frame: close-up, portrait, medium shot, cowboy shot, full body, wide shot. This controls distance and how much setting you see.
Angle is where the camera sits relative to the subject: eye level, from above, from below, dutch angle, over-the-shoulder, point of view. This controls the psychological feel and the perspective distortion.
Layer a lens and depth cue on top (85mm, shallow depth of field, bokeh) and you have full control. Get these right and the prompt formula slot six does heavy lifting for you.
| Layer | Example tokens | Controls |
|---|---|---|
| Shot type | close-up, medium shot, full body, wide shot | How much is in frame |
| Angle | from above, from below, eye level, dutch angle | Perspective and feel |
| Lens | 35mm, 50mm, 85mm, 135mm | Compression and intimacy |
| Depth | shallow depth of field, bokeh, deep focus | Background separation |
Shot types, from tight to wide
Close-up and portrait
Close-ups emphasize the face and expression and minimize background. They are the most flattering and the most forgiving on anatomy because less body is in frame. Great for mood.
1woman, adult, close-up, looking at viewer, soft smile, natural makeup,
blurred background, soft key light, rim light, 85mm, shallow depth of field,
bokeh, photorealistic, best quality, highly detailed
Negative: child, minor, underage, loli, shota, bad anatomy, deformed face,
watermark, text, low qualityMedium shot
The workhorse. Frames roughly head to waist, balancing subject and a hint of setting. The safest general purpose choice.
score_9, score_8_up, score_7_up, 1girl, adult, mature female, medium shot,
standing, looking at viewer, off-shoulder top, indoor, window light,
depth of field, 50mm
Negative: score_1, child, minor, underage, loli, shota, bad hands,
extra digits, deformed, watermarkCowboy shot
A booru and film term for framing from mid thigh up. Shows more of the outfit and pose than a medium shot while keeping the face readable. Very popular for full outfit display.
masterpiece, best quality, 1girl, adult, mature female, cowboy shot,
standing, contrapposto, looking at viewer, fitted dress, studio backdrop,
soft lighting, depth of field, anime style
Negative: child, minor, underage, loli, shota, worst quality, bad anatomy,
bad hands, watermarkFull body and wide shot
Full body shows the entire figure, which is essential for poses and outfits but raises anatomy risk because the model must get the whole body right at once. Wide shots add environment and make the setting a character.
1woman, adult, full body shot, standing, dynamic pose, athletic wear,
outdoor setting, bright daylight, deep focus, 35mm, photorealistic,
best quality, highly detailed
Negative: child, minor, underage, loli, shota, bad anatomy, bad hands,
extra limbs, deformed, watermark, textCamera angles and what they do
Eye level
Neutral and natural. The default. Use it when you want the subject presented plainly without psychological slant.
From above (high angle)
The camera looks down at the subject. Tends to feel intimate or vulnerable and elongates the upper body. A common flattering choice for lying poses shot from overhead.
1woman, adult, lying on bed, on back, looking at viewer, from above,
silk sheets, dim bedroom, low key lighting, 35mm, shallow depth of field,
photorealistic, film grain, best quality
Negative: child, minor, underage, loli, shota, bad anatomy, bad hands,
extra limbs, deformed, watermark, textFrom below (low angle)
The camera looks up. Feels powerful, dominant, and dramatic, and emphasizes height. Watch for perspective distortion of the lower body.
score_9, score_8_up, 1girl, adult, mature female, standing, from below,
looking down at viewer, confident expression, fitted outfit, studio,
dramatic lighting, low angle, depth of field
Negative: score_1, child, minor, underage, loli, shota, bad anatomy,
deformed, watermarkDutch angle
The camera tilts so the horizon is diagonal. Adds energy, tension, and a stylized editorial feel. Use sparingly because it draws attention to itself.
Over-the-shoulder
Frames the subject from behind a foreground shoulder. Adds depth and a candid, observed quality. Strong for storytelling compositions.
Point of view (POV)
The camera is positioned as if it were a participant’s eyes. Creates immersion. Phrase it as pov, looking at viewer and expect to iterate, since POV framing is one of the harder things to get clean from text.
1woman, adult, pov, looking at viewer, lying on bed, reaching toward viewer,
bedroom, warm lighting, 28mm, shallow depth of field, photorealistic,
best quality, highly detailed
Negative: child, minor, underage, loli, shota, bad anatomy, bad hands,
extra fingers, deformed, watermark, textWant to feel how each angle shifts the mood? Render the same subject three ways here.
Lens and focal length: the secret realism lever
Focal length is the most underused camera token. It changes facial proportions and background compression, and naming a real lens nudges the model toward photographic realism.
| Focal length | Effect | Best for |
|---|---|---|
| 24mm to 28mm | Wide, slight distortion, lots of scene | Environmental, POV |
| 35mm | Natural, documentary feel | Full body, candid |
| 50mm | Neutral, true to eye | General, medium shots |
| 85mm | Flattering compression, soft background | Portraits, close-ups |
| 135mm | Strong compression, heavy background blur | Tight beauty shots |
Pair the lens with a depth cue. 85mm, shallow depth of field, bokeh is the classic portrait recipe that throws the background into creamy blur and isolates the subject. 35mm, deep focus keeps the whole scene sharp for environmental shots. This is where realism specific work pays off, covered in how to make realistic AI images.
1woman, adult, portrait, close-up, looking at viewer, soft expression,
blurred bedroom background, warm window light, rim light, 135mm,
very shallow depth of field, bokeh, photorealistic, film grain, best quality
Negative: child, minor, underage, loli, shota, bad anatomy, deformed face,
watermark, text, low qualityDepth of field and background separation
Depth of field decides how much of the scene is in focus. Shallow depth of field blurs the background and pulls the eye to the subject, which is why it reads as professional and intimate. Deep focus keeps everything sharp, which suits storytelling and environmental images but can make backgrounds compete with the subject. Combine your depth choice with clean background phrasing from the settings guide so the blur has something tidy to blur.
A practical rule: when in doubt, go shallow. Most NSFW work wants the subject dominant and the background suggestive rather than detailed. Shallow depth of field plus a simple background is the most reliable path to a clean, focused image, and it also hides messy background artifacts the model might otherwise generate.
Per-model camera differences
The camera vocabulary is largely shared, with one caveat. Booru trained models (Pony, Illustrious) know danbooru framing tags precisely: cowboy shot, from above, from below, dutch angle, pov are all real dataset tags and fire reliably. The full reference is in danbooru tags for NSFW AI. SDXL realistic checkpoints understand both the tags and plain descriptions. Flux prefers a descriptive sentence: “shot from a low angle on an 85mm lens with a shallow depth of field.” See the Flux guide and pick your base from best NSFW checkpoints.
The takeaway is the same as everywhere in this cluster: the intent is universal, the dialect is per model. Tags for booru models, plain sentences for Flux, either for SDXL.
Common camera mistakes
| Mistake | Symptom | Fix |
|---|---|---|
| No camera tag | Flat, generic eye level | Always set shot type and angle |
| Full body too often | More anatomy errors | Use medium or cowboy when possible |
| Conflicting angles | Confused perspective | One angle token per image |
| No depth cue | Busy, distracting background | Add shallow depth of field |
| Extreme low angle | Distorted lower body | Soften the angle or reframe |
| POV without iteration | Broken framing | Expect several attempts on POV |
Putting it together with the rest of the prompt
Camera choices must agree with pose and outfit. A full body shot needs a pose that reads at distance and an outfit that displays well head to toe. A close-up wants a strong facial expression and flattering light. An over-the-shoulder shot needs a believable setting behind the subject. Coordinate with pose prompts, outfit prompts, and lighting prompts. When camera, pose, light, and setting all support one intent, the image stops looking generated and starts looking shot.
The creators who get cinematic results are not using secret tokens. They are simply filling the camera slot deliberately on every render instead of leaving it empty. A small, reusable set of three or four go to camera recipes (a flattering 85mm portrait, a 50mm medium, a 35mm full body, a 28mm POV) covers almost everything you will ever make.
A reusable camera swipe file
The fastest workflow is a small set of trusted camera recipes you paste and tweak. Keep these as named blocks and append your safety negative to each.
- Flattering portrait:
close-up, looking at viewer, 85mm, shallow depth of field, bokeh, soft key light, rim light - Balanced medium:
medium shot, looking at viewer, 50mm, depth of field, soft lighting - Outfit display:
cowboy shot, standing, 50mm, depth of field - Full body environmental:
full body shot, 35mm, deep focus - Intimate POV:
pov, looking at viewer, 28mm, shallow depth of field - Dramatic low angle:
from below, looking down at viewer, low angle, dramatic lighting, depth of field - Overhead intimate:
from above, lying pose, 35mm, shallow depth of field
These seven cover the overwhelming majority of NSFW compositions. Because each one bundles a shot type, an angle, a lens, and a depth cue, you never have to think about the camera from scratch again. You just pick the mood and paste.
Why deliberate camera work matters for a creator brand
A consistent camera language is part of what gives a creator set a recognizable look. If every image in your gallery uses the same flattering 85mm portrait recipe, the set feels like one photographer’s body of work rather than random outputs. Many top creators deliberately limit themselves to two or three camera recipes precisely to build that signature consistency. The lens, the angle, and the depth become a visual fingerprint.
This ties back to the broader character consistency techniques: a stable face, a stable wardrobe, a stable setting, and a stable camera language together create a persona that reads as real and coherent across dozens of images. The camera is the slot most people ignore, which is exactly why mastering it sets your output apart.
Combining angle and shot type deliberately
The real power comes from pairing a shot type with an angle that reinforces the same intent. A close-up from slightly above feels intimate and tender. A full body from below feels powerful and statuesque. A medium shot at eye level feels honest and direct. The wrong combination undercuts itself: a powerful low angle paired with a tiny close-up wastes both. Think of shot type as the distance and angle as the attitude, and choose a pair that agree.
| Shot type plus angle | Feel | Use when |
|---|---|---|
| Close-up from above | Intimate, tender | Emotional, soft mood |
| Full body from below | Powerful, statuesque | Confident, dominant |
| Medium at eye level | Honest, direct | Natural, candid |
| Cowboy with slight low angle | Strong, flattering | Outfit and pose display |
| Wide at eye level | Environmental, cinematic | Setting is the story |
Once you internalize a few of these pairings, you stop guessing. You decide the feeling you want and the camera tokens follow automatically. This is exactly how working photographers think on set, and it translates directly to prompting.
Aspect ratio is part of the camera decision
Framing does not end at shot type and angle. The aspect ratio of the canvas decides what fits. A portrait ratio like 2:3 or 3:4 suits standing figures and close-ups and crops cleanly to social feeds. A landscape ratio suits wide environmental shots and reclining poses. A square is a safe general default. Set the aspect ratio to match the shot type: forcing a full body standing pose into a square crops the legs, while a portrait ratio gives it room. Decide the ratio at the same time you decide the shot, not as an afterthought, because no prompt token can recover detail that was cropped out of the frame.
This matters even more for creator sets destined for a specific platform. Choose the ratio for the destination first, then pick a shot type and angle that compose well inside it. Coordinating ratio, shot, and angle from the start is what gives a set its polished, intentional, platform native look.
A reusable camera swipe file
The fastest workflow is a small set of trusted camera recipes you paste and tweak. Keep these as named blocks and append your safety negative to each.
- Flattering portrait:
close-up, looking at viewer, 85mm, shallow depth of field, bokeh, soft key light, rim light - Balanced medium:
medium shot, looking at viewer, 50mm, depth of field, soft lighting - Outfit display:
cowboy shot, standing, 50mm, depth of field - Full body environmental:
full body shot, 35mm, deep focus - Intimate POV:
pov, looking at viewer, 28mm, shallow depth of field - Dramatic low angle:
from below, looking down at viewer, low angle, dramatic lighting, depth of field - Overhead intimate:
from above, lying pose, 35mm, shallow depth of field
These seven cover the overwhelming majority of NSFW compositions. Because each one bundles a shot type, an angle, a lens, and a depth cue, you never have to think about the camera from scratch again. You just pick the mood and paste.
Why deliberate camera work matters for a creator brand
A consistent camera language is part of what gives a creator set a recognizable look. If every image in your gallery uses the same flattering 85mm portrait recipe, the set feels like one photographer’s body of work rather than random outputs. Many top creators deliberately limit themselves to two or three camera recipes precisely to build that signature consistency. The lens, the angle, and the depth become a visual fingerprint.
This ties back to the broader character consistency techniques: a stable face, a stable wardrobe, a stable setting, and a stable camera language together create a persona that reads as real and coherent across dozens of images. The camera is the slot most people ignore, which is exactly why mastering it sets your output apart.
When your framing is set, open the generator and shoot it. Keep every subject adult, fictional, and AI-generated, never a real person’s likeness, and keep your safety negatives locked on.
Frequently asked questions
How do I control the camera angle in an AI image prompt?
Use two layers of tokens: a shot type such as close-up, medium shot, or full body, and an angle such as from above, from below, eye level, or dutch angle. Then add a focal length like 85mm and a depth cue like shallow depth of field. Booru models know these as exact tags, while Flux prefers a descriptive sentence.
What does focal length like 85mm actually do?
Focal length changes facial proportions and background compression and nudges the model toward photographic realism. 85mm gives flattering compression and a soft background, ideal for portraits. 35mm looks natural and documentary for full body shots. 135mm compresses heavily for tight beauty shots. Naming a real lens is one of the most underused realism levers in prompting.
Why does from below make my subject look distorted?
Low angle shots looking up exaggerate whatever is closest to the camera, which often distorts the lower body or legs. The effect is dramatic and powerful, but extreme angles amplify perspective errors. Soften the angle, reframe slightly higher, or add depth cues to manage it. A moderate low angle gives the dominant feel without the worst distortion.
What is a cowboy shot and when should I use it?
A cowboy shot frames the subject from roughly mid thigh upward. It is a film and danbooru term. It shows more of the outfit and pose than a medium shot while keeping the face readable, which makes it ideal for displaying a full outfit without committing to a riskier full body render. It is one of the most reliable framings overall.
How do I get a blurry background like a real camera?
Add a shallow depth of field and bokeh to your prompt, ideally paired with a longer lens like 85mm or 135mm and a simple background. Shallow depth of field blurs everything except the subject, isolating it the way a fast portrait lens does. This also hides messy background artifacts the model might otherwise generate, so it doubles as a cleanup trick.
Do camera tags work the same on Flux and Pony?
The intent is the same but the dialect differs. Booru models like Pony and Illustrious recognize exact danbooru framing tags such as cowboy shot, from above, and pov. Flux prefers a full descriptive sentence like shot from a low angle on an 85mm lens. SDXL understands both. Match the phrasing to the model family for the most reliable framing.
Why does my POV shot keep breaking?
Point of view framing is one of the hardest things to get clean from text because the model must imagine the camera as a participant. Phrase it as pov, looking at viewer and expect to iterate across several seeds. A wider lens like 28mm helps. If you need it reliable, combine the prompt with ControlNet to constrain the composition.
Can I use camera angles to recreate a real photo of someone?
You can study real photography for framing technique, but you must never recreate NSFW images of a real identifiable person, even by matching their photo’s angle and pose. Every subject here is an adult, fictional, AI-generated character. Use original or owned personas only, and keep the likeness and minor safety negatives active on every render without exception.
