Nodaro Docs
DocumentationNode ReferenceModelsAI Agents (MCP)DevelopersSelf-hostingResearchCompare
Guides

Ten camera angles from one frame

Turn one AI image into ten camera angles of the same scene in Nodaro. An LLM writes the shot list, and one Generate Image node renders every angle.

View as MarkdownOpen in ChatGPTOpen in Claude

The Camera Coverage tutorial turns one image into ten shots of the same scene: wide, medium, close-up, over-the-shoulder, reverse, insert, low angle and profile. An LLM reads the image and writes the shot list, and a List runs one Generate Image node once per shot. The character, the clothes, the place and the light stay the same; only the camera changes.

The reference frame

The scene starts as one image, made by GPT Image 2 from a short prompt:

The prompt of the reference frame
a young woman in a worn olive field jacket sitting alone at a rain-streaked cafe window at night, warm interior lamp light on her face, cold neon glow outside, wet street reflections, shallow depth of field, 35mm film grain, cinematic still, photorealistic, high detail
A young woman in an olive field jacket at a rain-streaked café window at night, with neon signs and wet street reflections outside
The reference frame, made with GPT Image 2.

How the workflow plans the shots

promptreferencespromptpromptreferencesTextThe sceneGenerate ImageReference frameTextCoverage briefPromptGPT-5.2 writes 10 linesSplit TextOne line per shotListShot list, editableGenerate ImageRuns once per row
The reference frame and a coverage brief go to an LLM, its ten lines become a List, and the List runs one Generate Image node ten times with the reference frame as a reference.
  1. A Text node holds the scene, and Generate Image makes the reference frame from it.
  2. A second Text node holds the coverage brief: the rules of real film coverage, shown below.
  3. A Prompt node reads the reference frame and the brief, and GPT-5.2 writes ten camera setups, one per line.
  4. Split Text turns the ten lines into ten items, and the List shows them as rows you can edit.
  5. One Generate Image node takes the List as its prompt and the reference frame as a reference. It runs once for each row, so ten rows make ten images.

Edit any row of the List before you run the shots, so you spend credits only on the plan you want.

The brief is the part to change. This is the brief the LLM received, word for word:

The coverage brief
You are given a reference frame from a scene. Write 10 image prompts that COVER THE SAME SCENE from different camera setups, so the shots can be cut together into one sequence. Character, wardrobe, location, time of day and lighting stay identical. Only the camera setup changes. Rules: - Build real coverage, in this order: 1 wide establishing, 2 medium, 2 close-up on the character, 1 over-the-shoulder, 1 reverse angle showing what she faces, 1 insert / cutaway detail, 1 low or high dynamic angle, 1 profile. - The character must stay recognizable in at least 7 of the 10. Pure detail shots: maximum 2. - ONE camera angle per prompt. Never stack two angle terms in the same line. - Describe the camera position, NOT camera movement. No pans, no zooms, no dolly - motion is added later. - Keep screen direction consistent across all shots: she stays on the same side of the frame line, so the cuts do not jump. - Every frame must be a usable start frame for an animated clip: leave breathing room around the subject, never crop tight against the frame edge. - No two prompts may describe the same framing. Reject near-duplicates before writing. - Each prompt under 30 words, phrased as a camera setup, not as a story. - Output exactly 10 lines, one prompt per line. No numbering, no bullets, no blank lines, no preamble.

The ten shots

Under each shot is the line the LLM wrote, which was the whole prompt of that image.

A wide shot of the café interior, the woman small at a table by the rainy window
1. Wide establishingWide establishing shot from inside the cozy café, woman seated left of frame beside rain-streaked window, neon-lit street outside, warm tungsten lighting, ample headroom.
A medium shot of the woman from across the table, a coffee mug and books in front of her
2. Medium, across the tableMedium waist-up from across the table, woman left of frame gazing through rain-covered glass, coffee mug and books foreground, identical night lighting.
A medium three-quarter view of the woman leaning on folded arms beside the window
3. Medium, three-quarter sideMedium three-quarter side view from café aisle, woman left of frame leaning on folded arms beside window, reflections shimmering outside, warm interior glow.
A close-up of the woman's face in three-quarter profile, neon bokeh behind her
4. Close-upClose-up on her face in profile-three-quarter, eyes fixed outside, raindrops and neon bokeh behind, soft warm café lighting, breathing room around head.
A tight close-up of the woman's thoughtful face with window reflections
5. Tight close-upTight close-up of her thoughtful expression, window reflections crossing her face, identical wardrobe and lighting, negative space ahead of her gaze.
An over-the-shoulder shot from behind the woman toward the rainy street and neon signs
6. Over-the-shoulderOver-the-shoulder from behind her left shoulder toward the rain-soaked street, her silhouette and jacket foreground, neon storefronts beyond, consistent screen direction.
A reverse angle from outside, looking through the wet window at the woman inside the café
7. Reverse angleReverse angle from outside the window toward the café interior, rain-speckled glass foreground, woman visible inside watching the street, warm interior contrasting cool night.
The woman at the window with a steaming mug and books, instead of a detail shot of the mug alone
8. Insert: the woman stayed in the frameInsert detail of steaming coffee mug, worn books, and raindrops on the windowsill, warm café light reflecting across wet surfaces.
A medium shot of the woman from table height, the glowing window and neon behind her
9. Low angleLow-angle medium shot from table height, woman left of frame against glowing window and neon reflections, same wardrobe, location, and nighttime lighting.
A side view of the woman facing the rainy window, the café blurred behind her
10. ProfileClean side-profile shot parallel to the window, woman left of frame facing right toward the rain, café interior softly receding behind her.

What held, and what did not

  • The scene held. The woman, her olive jacket, the café, the rain and the night light match in all ten shots. Each shot used the reference frame as a reference image, which is what keeps them consistent.
  • One shot missed its plan. The insert was planned as a detail of the mug and the books alone. GPT Image 2 kept the woman in the frame, so shot 8 is another medium shot. Rewrite that row, or run it again, to get a true insert.
  • The neon signs are invented. The model wrote made-up words on the signs outside, so check any lettering before you use a shot.

Make it yours

  • A new scene: change the scene in the first Text node and run the workflow from the top. The coverage plan stays the same.
  • A different plan: change the brief to ask for six shots, another order, or another kind of coverage, such as documentary, action or interview.
  • One angle at a time: edit one row of the List before you run Generate Image.
  • Motion next: the brief asks for room around the subject, so the shots can serve as start frames for Generate Video.

What it costs

Prices are today's, for the same settings:

StepModelCredits
The reference frameGPT Image 221
The shot listThe LLM of the Prompt node2–7, depending on the model
The ten shotsGPT Image 2210

Try it

Open the Camera Coverage tutorial on app.nodaro.ai. It shows this run, with every node and image, without signing in. Run tutorial copies the workflow into your account, and copying is free. See Tutorials and templates for where tutorials live in the app.

Frequently asked questions

Generate Image

Create an image from a text prompt with one of Nodaro's image models, keep characters consistent with references, and edit results in place with a mask.

List

Type or paste a list of prompts, or build a table with several columns, and every connected node runs once per row. The List node itself is free.

Prompt

Generate, rewrite or summarize text with Gemini, Claude, GPT or Grok. Add instructions, images, video or audio, and split the answer into items.

Consistent characters

The complete method to keep the same person, pet, product or place in every image and video, with assets, approved pictures, mentions and identity lock.

Pickers in action

Nodaro pickers turn a choice from a grid of pictures into tested prompt words for light, style, lens, camera and music. Real runs show what each choice added.

Last updated on

Seven songs from music pickers

Seven Suno songs made in Nodaro from music pickers. Choose a genre, a mood, instruments and a voice, and the pickers write the description Suno receives.

Import and export

Export a workflow as one JSON file, with or without its characters, objects and locations, and import it into another account, project or self-hosted install.

On this page

The reference frame
How the workflow plans the shots
The ten shots
What held, and what did not
Make it yours
What it costs
Try it