If you just type "warrior" in Midjourney, you get a generic warrior. But if you type "low-angle cowboy shot of a cyberpunk samurai, 85mm lens, shallow depth of field, neon reflections flickering across her face"? Suddenly it looks like a scene from a movie.
What's the difference? It's not computing power—it's precision in language.
Models like Midjourney v6, Sora, and Veo don't just parse subjects—they understand cinematic grammar. They recognize cinematography terms that have evolved over 100+ years. When you know these terms, your prompts become dramatically different.
I used to just write "cool angle" and kept getting similar results every time. Once I started using cinematography terminology, I got much closer to what I was envisioning.
Part 1: Shot Types (How Much of the Frame the Subject Occupies)
Extreme Wide Shot (ELS)
The subject appears very small while the environment dominates. Used for world-building or showing scale.
Prompt example:
ELS of a lone archaeologist approaching a pyramid on Mars,
dust storm on horizon, subject is thumbnail-sized silhouette
against epic rust-colored sky, cinematic establishing shot --ar 16:9
Technical tip: 14-24mm ultra-wide lens, deep focus (f/8-f/11), 16:9 or wider aspect ratio
Long Shot (LS) vs Full Shot (FS)
Long Shot: Full body visible with significant background. For placing characters within their environment.
LS of a wizard atop a castle tower, full body visible,
storm clouds swirling behind, balances character presence with epic scale --ar 2:3
Full Shot: Head to toe fills the frame. Essential for character design sheets.
FS of a Victorian vampire hunter, standing pose,
every detail of leather coat and silver weaponry visible,
character design sheet --ar 2:3
When designing characters for games or animation, you must use "full shot". Otherwise feet get cut off or proportions come out weird.
Medium Distance Shots
| Shot Type | Framing | Purpose |
|---|---|---|
| Medium Long Shot (MLS) | Above the knees | Capturing body language during dialogue |
| Cowboy Shot | Mid-thigh up | Emphasizing weapons, tools, hand gestures |
| Medium Shot (MS) | Waist up | Corporate portraits, standard dialogue scenes |
Cowboy shot example:
Cowboy shot of a mecha pilot gripping throttle controls,
HUD visible in background, mid-thigh framing emphasizes gauntlets --ar 2:3
The Cowboy Shot got its name from Western films where they framed characters down to mid-thigh to show their holsters. Great for emphasizing weapons or tools.
Close-Up Territory
| Shot Type | Framing | Emotional Intensity |
|---|---|---|
| Medium Close-Up (MCU) | Chest up | Reactions |
| Close-Up (CU) | Face and neck | Intimacy |
| Extreme Close-Up (ECU) | Single eye, lips, details | Maximum focus |
ECU example:
ECU of a dragon's eye with vertical slit pupil,
scales around eyelid, reflection of approaching hero,
macro photography detail --ar 1:1
Important: When using ECU, you need to specify "macro photography" or macro lens (100mm+). Otherwise the AI will just make a blurry crop.
Part 2: Camera Angles (The Psychology of Emotion)
Angles determine how the audience feels about the subject. They're a powerful tool for manipulating emotions even without context.
Low Angle vs High Angle
Low Angle: Camera looks up at the subject from below. Power, dominance, heroism, intimidation.
Low angle shot of a queen addressing her army,
camera at ground level, silhouette against sunrise,
24mm lens exaggerates stature, inspiring and formidable --ar 2:3
High Angle: Looking down from above. Vulnerability, submission, isolation.
High angle shot of a lost dog in Times Square,
subject appears small and overwhelmed,
overhead perspective captures crowds parting around it --ar 4:5
When I want a character to look impressive, I almost always use low angle. Conversely, when I want them to look pitiful or weak, I use high angle. Just knowing this makes a huge difference in the mood.
Extreme Perspectives
| Angle | Description | Effect |
|---|---|---|
| Ground Level | A few inches off the floor | Exaggerated scale |
| Hip Level | Hip height | Dynamic movement (skateboarding, swordplay) |
| Bird's Eye View | 90-degree vertical overhead | Patterns and geometric composition |
| Aerial Shot | High but not vertical (drone feel) | Vast environments |
Bird's Eye example:
Bird's-eye view directly overhead of a circular crop circle,
geometric precision, tiny investigator at center, alien mystery --ar 1:1
Dutch Angle
A shot where the camera is tilted on its axis so the horizon appears diagonal. Creates subconscious discomfort, instability, and tension.
Dutch angle shot of a conspiracy theorist in cluttered basement,
30-degree tilt, papers sliding off desk, unsettling paranoia --ar 4:5
Tip: Specifying the tilt angle makes it more precise (15°, 30°, 45°). 15 degrees for subtle unease, 45 degrees for extreme chaos.
When going for horror, thriller, or crime vibes, adding a dutch angle definitely changes the mood. But overuse it and things get dizzy, so use it sparingly.
Part 3: Camera Movement (For Video Generation)
These are terms you need when using video generation AI like Sora, Veo, or Kling. They're also useful for adding dynamic feels to still images.
Basic Movements: Pan & Tilt
| Movement | Direction | Purpose |
|---|---|---|
| Pan | Horizontal (left↔right) | Revealing information, following action |
| Tilt | Vertical (up↔down) | Revealing scale, guiding the eye |
Pan example:
Slow pan right across ancient library, camera fixed on pedestal,
revealing towering shelves of scrolls, dust particles catching light,
4-second duration --ar 16:9
Dolly Zoom
A technique where the camera moves while simultaneously zooming in the opposite direction. It creates a disorienting scale-shift effect where the background suddenly compresses or expands. Made famous by Hitchcock's "Vertigo."
Dolly zoom in on astronaut seeing Earth from space station viewport,
background stars compress, emotional weight of realization, 2-second effect --ar 4:5
Spatial Movement: Push, Pull, Truck
| Movement | Direction | Emotional Effect |
|---|---|---|
| Push In | Camera advances toward subject | Intensifying emotion, building tension |
| Pull Out | Camera retreats | Revealing context, isolation |
| Truck | Lateral movement (parallel to subject) | Tracking action, parallax |
Pull Out example:
Pull out from crying child to reveal entire empty playground,
isolation emphasized, context builds sadness --ar 16:9
Advanced Techniques
| Technique | Description |
|---|---|
| 360° Rotation | Orbiting around the subject |
| POV (Point of View) | First-person view through character's eyes |
| Crane Shot | Vertical up or down movement |
| Handheld | Slight shakiness as if hand-held |
| Whip Pan | Lightning-fast pan with motion blur |
POV example:
POV shot of rock climber reaching for next hold,
hands visible, vertigo-inducing drop below,
helmet cam perspective, slight head movement --ar 16:9
Handheld example:
Handheld footage of journalist running through protest,
slight camera shake, documentary realism, urgent and immersive --ar 16:9
Handheld is great for documentaries or urgent scenes when you want realism. Even for still images, adding "slight motion blur, handheld feel" makes it less pristine and more dynamic.
Practical Prompt Combinations
Here are combinations I use frequently.
Heroic Character
low angle cowboy shot, 24mm wide lens, dramatic rim lighting,
heroic pose, cape flowing, sunset sky --ar 2:3
Intimate Emotional Scene
MCU, eye level, 85mm lens, shallow depth of field,
soft window light, subtle emotion, intimate moment --ar 4:5
Thriller/Horror Atmosphere
dutch angle 25 degrees, high contrast lighting,
close-up, deep shadows, unsettling atmosphere --ar 4:5
World-Building (Establishing)
ELS aerial shot, drone perspective, golden hour,
epic fantasy landscape, tiny travelers on road,
cinematic establishing shot --ar 21:9
Action Scene
hip level tracking shot, motion blur,
dynamic pose mid-action, 16mm wide angle,
high shutter speed freeze --ar 16:9
Lens Recommendations Summary
| Lens | Focal Length | Use Case |
|---|---|---|
| Ultra Wide | 14-24mm | ELS, world-building, exaggerated perspective |
| Wide | 24-35mm | LS, balancing environment + character |
| Normal | 50mm | Natural perspective, documentary |
| Portrait | 85mm | Portraits, emotional scenes, bokeh |
| Telephoto | 135mm+ | Compressed background, subject isolation |
| Macro | 100mm+ | ECU, details |
Conclusion
"Image generation" and "frame direction" are different things.
Every prompt is a storyboard, every term is a director's note. When you know cinematography terminology, you're not just "generating" with AI—you're collaborating like a cinematographer.
Subject matters. Lighting matters. But where and how you place the virtual camera determines what the audience will feel.
Start by memorizing just a few and expand from there. I still can't remember all of them and look things up when I need to.
