Camera Angle Prompts for AI Images: Wide, Macro, 85mm, Drone

Camera angle prompts name where the camera sits, how high it looks, which lens it uses, and how close it stands so Midjourney v7, FLUX, Leonardo, SDXL, or GPT Image draw the frame you meant. Searchers want paste-ready language for wide establishing shots, 85mm portraits, macro product planes, drone overheads, and low-angle hero frames. You leave with a camera stack, kits for those five shot types, dialect notes as of mid-2026, and a photo-first path when a still shows the viewpoint. This page stays on viewpoint and optics. Keep lighting as its own layer. Soft path when a reference photo holds the angle: PromptMake /image, then confirm the camera line before you generate.

Who needs camera angle prompts

You type camera angle prompts because a model with no viewpoint instruction picks a default: eye-level, medium distance, mixed lens. That default can serve a standing portrait. It fails a watch crown and a hero looking up at the sky. Product teams need a parallel macro plane so labels stay readable. Photographers need 85mm compression on faces. Architects need wide glass that shows floor and street. Marketers need drone plates from 200 to 400 feet. The prompt must name shot size, camera height, lens, and subject distance. Vague "cinematic camera" returns a random tilt and a mystery focal length.

Text generation invents a viewpoint from nouns. Image-to-prompt tools that read a still own the "keep this height and lens, rewrite the rest" job. This article covers writing camera language: a four-part stack, kits for wide / 85mm / low angle / macro / drone, Midjourney v7 and FLUX paste dialects, and photo-first Recreate or Create Variation when you upload a frame. Soft product path for the photo lane: https://promptmake.net/image. Lighting keywords (key, fill, rim, Kelvin) belong in a different clause; a sister post on this blog owns that layer. Aspect ratio is canvas shape. Lock --ar or pixels after you lock the camera line.

Treat generator stills as composition tests and concept frames. Clear location and likeness rights before you publish.

Strong fit for:

• Product and catalog teams who need a locked macro plane or a three-quarter pack shot across a SKU set
• Portrait and lookbook work that must stay on 85mm compression instead of 24mm face stretch
• Location, architecture, and campaign teams who need wide establishing plates or drone overheads
• Anyone who ports the same camera line into Midjourney v7, FLUX, SDXL, or GPT Image without guessing the lens

The camera stack: shot size, height, lens, distance

Treat camera angle prompts as a short shot order. Models fill a missing viewpoint with stock tropes: eye-level beauty, Dutch tilt on a "dramatic" adjective, 35mm on a scene that needed 14mm or 100mm macro. Write four facts in a stable order. Shot size says how much of the world sits in frame: extreme wide, wide, medium, close-up, extreme close-up, macro. Camera height says where the lens sits relative to the subject: eye level, low angle looking up, high angle looking down, bird's eye, worm's eye. Lens or focal length names the glass: 14mm, 24mm, 35mm, 50mm, 85mm, 100mm macro, 135mm. Subject distance closes the ask: chest height three meters back, camera one meter from the watch, drone at 300 feet. Write the stack once as labeled facts. Format twice: compact phrases plus trailing parameters for Midjourney v7 (and V8 Alpha where you have access), flowing sentences for FLUX family builds. SDXL and Leonardo take denser camera tokens. GPT Image prefers a clean paragraph you can revise in chat. Leave the lamp in a different clause so a lens swap does not rewrite the key.

Shot size and camera height

Shot size is crop intent. Extreme wide and wide put the environment in charge: a person is small, architecture or landscape reads first. Medium shot holds torso and head. Close-up holds the face or a product hero. Extreme close-up and macro hold a plane: an eye, a watch crown, fabric weave. Name the size in the first camera clause so the model does not invent a bust crop on an establishing brief. Camera height is the second clause. Eye level matches a standing conversation and most catalog faces. Low angle (camera below the subject, looking up) adds stature: athletes, monuments. High angle (camera above, looking down) shrinks the subject or shows a tabletop layout. Bird's eye and nadir look straight down. Worm's eye looks straight up from the ground. Over-the-shoulder puts a foreground silhouette in the near edge. Dutch tilt (canted horizon) is a spice for unease; leave it out of catalog and headshot work. For Midjourney v7, pair height with shot size: "low angle medium shot, camera at knee height looking up." For FLUX, put the same facts in a sentence: shot from knee height looking up, medium framing, sky in the upper third.

Lens and subject distance

Focal length changes geometry. Wide glass (14mm to 24mm) stretches near objects and shows more of the room or street; faces at the edge warp. Normal glass (35mm to 50mm) sits close to how you see a scene with your eyes. Portrait glass (85mm to 135mm) compresses features and throws the background out of focus when you also ask for shallow depth of field. Macro glass (90mm to 105mm class) holds a tiny plane in focus. Write the millimeter number. "Portrait lens" is weaker than "85mm lens, shallow depth of field, head and shoulders." Subject distance must agree with the glass. An 85mm headshot wants the camera a few meters back. A 24mm interior wants the camera in the corner so floor and ceiling lines read. A 100mm macro wants the camera parallel to the label, centimeters from the subject. Conflicting pairs waste a generate: "14mm extreme wide, creamy 85mm bokeh, face filling the frame" asks for three cameras. Pick one glass per run. Change focal length on the next round if the first still feels cramped or distant. Aperture feel belongs with the lens: "shallow depth of field" on 85mm portraits, "deep focus" on 24mm architecture. Midjourney users add --style raw for photographic lens behavior. FLUX users spell the same optics in prose.

Camera angle prompts for wide, 85mm, and low-angle shots

Wide, 85mm, and low-angle kits share a habit: you lock height and glass before you add wardrobe or grade. Wide establishing shots fail when you also demand a tight face and creamy bokeh. 85mm portraits fail when you also demand a full lobby in focus. Low-angle hero shots fail when the camera sits at eye level and you added the word "epic." Write a paste template per kit and reuse it across subjects so the camera line stays stable while the person or building changes. Midjourney v7 wants compact phrases and trailing --ar plus --style raw for photos. FLUX.1.x and Flux 2 family builds (name the variant your host shows) want a full sentence that embeds height, millimeters, and distance. Keep --ar as canvas shape: 16:9 for establishing plates, 4:5 or 3:4 for portraits, 2:3 for full-body hero frames. Aspect ratio is the box. "Low angle" and "24mm" are camera facts. The subsections below give paste-ready wide, 85mm, and low-angle cores you can drop a subject into. Lighting stays a short, stable clause ("soft window light from camera left") so this page stays on viewpoint.

Wide establishing frames

Wide establishing (Midjourney v7): empty coastal highway at dawn, extreme wide establishing shot, camera at chest height, 24mm lens, subject small in frame, deep focus, environment dominates --ar 16:9 --style raw --v 7

Wide interior (FLUX prose): Photograph of a hotel lobby from the far corner, wide 24mm lens, camera at chest height, eye-level, deep focus so floor tiles and the mezzanine rail both read, a single person small near the desk.

Architecture exterior: "wide angle 16mm, camera at street level, three-quarter view of the facade, vertical lines held straight, full building in frame, deep focus." Add "straight horizon" when you need a listing plate. 14mm to 16mm shows more street and sky; it also bends faces at the edge, so keep people small or toward center. 24mm is the safer establishing default for mixed subject-plus-place frames. Pair with --ar 16:9 or --ar 21:9 on Midjourney when the delivery is a site hero. Change the location nouns on reuse so a campaign set shares one camera. Street work: 35mm for a documentary walk, 24mm if the street itself is the subject. Leave drone language out of this kit so you do not mix a street-level 24mm with a 400-foot nadir in one line.

85mm portraits and low-angle hero shots

85mm portrait (Midjourney v7): editorial portrait, woman in a linen jacket, 85mm lens, shallow depth of field, eye-level camera, head and shoulders crop, background falloff --ar 4:5 --style raw --v 7

85mm portrait (FLUX): Editorial photograph of a man in a wool coat, shot on an 85mm lens at eye level, camera a few meters back, head and shoulders, shallow depth of field, background lights as soft discs.

Head-and-shoulders 85mm is the default for client portraits and lookbook faces. 135mm compresses more and needs more distance; use it when you have a long studio. 50mm is a full-body or environmental portrait with more setting around the head. Keep "85mm" and "14mm interior" in separate prompts. Low-angle hero (Midjourney): low angle hero shot, athlete on wet asphalt, camera at knee height looking up, 35mm lens, sky filling the upper frame --ar 2:3 --style raw --v 7. Low-angle hero (FLUX): Photograph from knee height looking up at a runner, 35mm lens, medium-wide framing, clouds in the upper third, the figure tall in the frame. Low angle plus 24mm or 35mm reads as stature. Low angle plus 85mm is tighter, so the sky may vanish. Monument looking-up shots use 16mm to 24mm from near the base. Keep Dutch tilt out of hero kits unless unease is the brief.

Macro and drone camera angle prompts

Macro wants a parallel close plane and a 90mm to 105mm class lens. Drone wants altitude, a nadir or oblique angle, and wide-to-normal aerial glass. Mixing "macro of the river from 400 feet" produces mush. Write altitude in feet or meters for aerials. Write working distance in centimeters for macro. Name the focus plane (engraving on the crown, roof grid) so the model knows what must be sharp. Midjourney v7 responds to "macro photograph" plus a millimeter number plus --style raw. FLUX responds to a sentence that includes macro lens, parallel to the subject, thin focus plane. Aerials respond to "drone photograph," "nadir," or "oblique 30 degrees," plus altitude. Bird's-eye from a building roof is a rooftop still; say rooftop if the camera sits on a parapet. The subsections below split close planes from aerials so you can paste one kit at a time. Soft path if you shot the watch or the site from a phone: upload to PromptMake /image, Recreate Exactly to recover the plane or the altitude language, then tighten millimeters by hand.

Macro close-ups and product planes

Macro product (Midjourney v7): macro photograph of a mechanical watch crown, 100mm macro lens, camera parallel to the subject, razor-thin focus on the engraving, smooth background falloff --ar 1:1 --style raw --v 7

Macro product (FLUX): Macro photograph of a ceramic glaze droplet, 100mm macro lens, camera parallel to the table, focus on the droplet edge, the rest of the plate falling off, no wide environment.

Food and craft macros use the same plane rule: camera parallel to the hero surface, 90mm to 105mm class, one focus noun (crumb crust, needle eye, circuit solder). "Close-up" without "macro" may return a 50mm product shot with too much of the room. Add "macro" and the millimeter number when the job is texture. For jewelry and watches, say "camera parallel to the face" so the model does not shoot from 45 degrees and hide the hallmark. Tabletop product three-quarter is a different kit: 50mm or 85mm, camera at 30 degrees above the sweep, full object in frame. Keep that kit out of macro runs. SDXL users can add a short negative against "wide angle interior, full room, tiny product." GPT Image lets you follow up with "move the camera closer, keep 100mm macro."

Drone, overhead, and aerial plates

Drone oblique (Midjourney v7): aerial drone photograph of a winding river through pine forest, 400 feet altitude, 30-degree oblique angle, 24mm equivalent, no people --ar 16:9 --style raw --v 7

Drone nadir (FLUX): Overhead drone photograph looking straight down at a courtyard, 200 feet altitude, nadir view, 24mm equivalent, paths and roof edges sharp, late-morning sun.

Oblique (20 to 45 degrees off nadir) shows the side of buildings and the lay of a valley. Nadir (straight down) maps roofs, fields, and parking grids. Pick one. Altitude language matters: 80 to 120 feet for a low site inspect, 200 to 400 feet for a neighborhood plate, higher for a river or coastline. "Bird's eye view" without altitude returns a 3D-game overlook. Write "drone photograph" plus feet. Coastal and farm plates pair with --ar 16:9. Keep 85mm portrait language and "macro" out of drone prompts. Rooftop bird's eye (camera on a building, looking down at a street) is a high-angle still; say "shot from a rooftop, looking down, 35mm" when that is the job. Follow local rules for real drone photography; generator aerials are stills.

Midjourney, FLUX, and GPT Image dialects

Porting a winner means translating the camera line. Midjourney v7 rewards a compact stack: subject, then shot size, height, lens, distance, then --ar, --style raw for photos, --v 7. Style and character references (--sref, --cref, --oref) lock look and identity inside the viewpoint you chose; you still write "85mm" or "knee height" yourself. FLUX.1.x and Flux 2 family builds take natural-language prompts. Put optics in the sentence: shot on an 85mm lens at eye level, camera a few meters back. Size sits in the host UI or API. SDXL wants camera tokens in the positive prompt and a short negative when the wrong glass keeps appearing ("fisheye, ultra wide, Dutch tilt" on a headshot). Leonardo remains a practical free-tier and game-style target; use the same height and millimeter nouns. DALL·E / GPT Image inside ChatGPT stay conversational: "make this a low-angle 35mm hero shot, camera at knee height" as a follow-up works because the thread holds the subject. Ideogram v3 helps when a title must sit in a wide plate; still name the camera so type does not float on a mystery crop.

Language models can draft the stack before you spend a generate. As of mid-2026, GPT-5.6 Sol (API gpt-5.6) handles hard brief writing; ChatGPT defaults to GPT-5.5 Instant for fast chat. Claude Fable 5 and Claude Opus 5 tidy millimeter lists. Ask for one prompt plus a separate line that states shot size, height, lens, and distance.

Step-by-step: lock the camera before you generate

A fixed order beats lucky tilts. Decide the shot type (wide, 85mm, macro, drone, low angle), write the four-part camera stack, format for the model, generate once, then change one layer (wardrobe, grade, or pose) without touching height or glass unless the brief changed. Two to five rounds with a locked camera beat ten rounds that hop from 24mm to 85mm. If you hold a photo or a prior AI still with the correct viewpoint, run image-to-prompt so the first draft inherits height and distance, then enforce the matching millimeters on paste. PromptMake /image formats for Midjourney, FLUX, DALL·E, Stable Diffusion, and Leonardo, with goal modes when the still is a reference. Guests get about 3 image runs per day; a free account raises that to about 5. Soft start: https://promptmake.net/image. The steps below cover blank briefs and photo-first jobs. Keep lighting words stable across those rounds so you can see what the camera change did.

From a blank brief to first paste

1. Write the job in one sentence: "85mm headshot, eye level," or "24mm lobby plate," or "100mm watch macro," or "drone 300 feet oblique," or "low-angle 35mm athlete."
2. Expand into the stack: shot size, camera height, lens, subject distance. Add one short light clause you will not edit this round.
3. Pick canvas: --ar 4:5 for portraits, --ar 16:9 for establishing and drone, --ar 1:1 for many macros, --ar 2:3 for full-body hero. FLUX and SDXL get the matching preset or pixels.
4. Format: Midjourney phrases plus --style raw --v 7. FLUX full sentences. SDXL tokens plus a short negative against the wrong glass. GPT Image a short paragraph.
5. Generate once. Check height and lens first. If the face stretches, you are too wide or too close. If the product is lost in a room, you are not in macro. Change one fact: millimeters, or height, or distance. Leave the subject nouns alone until the camera reads.

Save the winning prompt beside the PNG with the kit in the filename (watch_macro_100mm_v3.png) so the next teammate does not reopen a wide plate and wonder why the engraving vanished.

From a reference photo (Recreate, Variation, Change Style)

Upload a still that shows the viewpoint you need: a low-angle phone frame of a building, a true macro of a product, a drone plate from a past shoot. On PromptMake /image, pick Midjourney, FLUX, DALL·E, Stable Diffusion, or Leonardo as the paste target. Recreate Exactly when you want near-match language and will confirm height and millimeters. Create Variation when you need sibling moods inside the same camera. Change Style when the viewpoint stays but the medium moves. Adjust Lighting when framing is fine and the key is wrong; that mode is for lamps. After copy-out, read the camera clause. Insert "85mm" or "100mm macro" or "400 feet drone" if vision wrote "close-up" or "aerial" with no glass. Split a lens rewrite and a lighting rewrite into separate runs; one overloaded pass muddies the draft.

Common mistakes with camera angle prompts

Cinematic camera with no height or glass. The model guesses. Write shot size, height, millimeters, and distance.

Two lenses in one line. 14mm extreme wide plus creamy 85mm bokeh plus a face-filling crop is three cameras. Pick one kit per generate.

Treating --ar as a viewpoint. Aspect ratio is the box. Low angle and 24mm are camera facts. Set both.

Confusing camera left with a low angle. "Key from camera left" is lighting. "Camera at knee height looking up" is viewpoint. Keep those clauses apart.

Macro without a parallel plane. "Close-up of the watch" from 45 degrees hides the hallmark. Say macro, 100mm, camera parallel to the face.

Bird's eye with no altitude. Write drone plus feet, or rooftop plus looking down. Bare "bird's eye" drifts into illustration overlooks.

Dutch tilt on catalog work. Canted horizons suit unease. Leave them out of pack shots and headshots.

Changing camera and lighting in one rewrite. You cannot tell which edit worked. Lock the lamp while you hunt for the right glass.

When PromptMake /image helps camera-aware prompts

Blank-page prompts work when you know the kit and the stack. Photo-first workflows win when a still defines the viewpoint: a knee-height phone frame, a true macro, a drone still from last year's site visit. PromptMake /image runs vision with model-specific formatting and goal modes: Recreate Exactly, Change Style, Adjust Lighting, Create Variation. Same upload, Midjourney dialect one run, FLUX the next. Soft path: https://promptmake.net/image. After the draft lands, you still own the camera line: insert millimeters, altitude, and height before you spend generator credits.

Use chat when you explore GPT Image wording and need a fast "move the camera lower" tweak in one thread. Use /image when a still exists and you need recreate or variation text aimed at Midjourney, FLUX, DALL·E, Stable Diffusion, or Leonardo. Keep this article on camera angle prompts so your first grid matches the shot you meant.

FAQ

What are camera angle prompts?

Camera angle prompts are text that tells an image model the viewpoint: shot size, camera height, lens, and subject distance. You paste them into Midjourney v7, FLUX, Leonardo, SDXL, or GPT Image so the frame matches a wide establishing plate, an 85mm portrait, a macro plane, a drone overhead, or a low-angle hero shot. Strong prompts name millimeters and height; weak ones stop at cinematic camera. PromptMake /image can recover that language from a reference still, then you confirm the camera line before you generate.

How do I write camera angle prompts for Midjourney v7?

Lead with subject, then add shot size, height, lens, and distance in compact phrases. End with --ar for the canvas, --style raw for photographic lens behavior, and --v 7. Example core: 85mm lens, eye-level, head and shoulders, shallow depth of field --ar 4:5 --style raw --v 7. References such as --sref or --cref hold look and identity, so you still write the camera clause by hand.

Which lens should I prompt for portraits versus landscapes?

Use 85mm (or 135mm) with shallow depth of field and a few meters of distance for head-and-shoulders portraits. Use 24mm or 16mm with deep focus and the camera pulled back for landscapes, interiors, and establishing plates. 50mm sits in the middle for environmental portraits and product three-quarters; macro work wants 90mm to 105mm and a parallel camera. Keep one millimeter number per prompt so the model does not blend wide stretch with portrait compression.

How do I prompt a drone or aerial shot?

Write drone photograph or aerial, then altitude in feet, then nadir (straight down) or an oblique angle in degrees, then a wide-to-normal equivalent such as 24mm. Example: aerial drone photograph, 400 feet, 30-degree oblique, 24mm equivalent --ar 16:9 --style raw --v 7. Bare bird's eye returns a game-like overlook; rooftop looking-down shots are high-angle stills, so say rooftop if the camera sits on a building. Generator aerials are stills.

How do I prompt a macro product shot?

Name macro photograph, a 90mm to 105mm class lens, camera parallel to the hero plane, and one focus noun (engraving, glaze droplet, fabric weave). Square --ar 1:1 fits many catalog macros; keep the room out of the brief. Close-up without the word macro may return a 50mm product shot with too much sweep in frame. Tabletop three-quarter at 50mm or 85mm is a separate kit, so keep it out of the macro run.

What is the difference between camera angle and lighting in a prompt?

Camera angle is where the lens sits and which glass it uses: height, tilt, millimeters, distance. Lighting is the lamp: source, direction, quality, color temperature. Change one layer per round: "camera at knee height looking up, 35mm" is viewpoint, and "soft key from camera left" is light. Use Adjust Lighting on PromptMake /image when the frame is right and the key is wrong; rewrite the camera clause when the still uses the wrong height or the wrong lens.

How do I start for free today?

Write one sentence with kit and glass ("85mm eye-level headshot" or "24mm lobby plate"), expand it with shot size, height, lens, and distance, and paste a Midjourney v7 or FLUX template from this guide. If you hold a photo that shows the angle, open https://promptmake.net/image, pick your model, run Recreate Exactly or Create Variation on the free daily quota (about 3 guest runs, about 5 registered), then insert missing millimeters before you generate. Save the winning prompt beside the PNG as the kit seed for the next subject.