Pose, Shot and Style nodes: a body, a frame and a look added to your prompt

Writing the same body position, camera framing and lighting into every prompt is the slow part of a shoot. On LUVI's workflow canvas, three nodes write those parts for you: Pose adds a body, Shot adds a composition and Style adds a look. They sit on the prompt wire between your own text and a generator, they cost nothing, and the only charge is the generator at the end—for example Grok Imagine Image 2.0 from 120 credits per image, or Nano Banana Pro Edit from 175.

What do the Pose, Shot and Style nodes send to the model?

The Pose, Shot and Style nodes send text, nothing else. Each one takes whatever prompt arrives on its left, adds its own description after a comma, and passes the sum out on the right. So a chain reads in order: Prompt → Pose → Shot → Style → the generator's prompt port. Your words come first, then the body, then the frame, then the look.

Three details matter in practice:

• The text is copied into the node when you pick it. A saved workflow keeps exactly what it sent, even if the library changes later.
• The text stays in English. Names on the tiles are translated into your interface language, but the description that reaches the model is the English original.
• The nodes are free. No model runs inside them, so the workflow's price only counts the generators.

What is in the Pose Atlas?

The Pose Atlas holds 740 poses, and a double-click on a Pose node opens it. They are filed in classes, each with its own row of filter chips:

• Cinematic (150), Fashion & Advertising (100), Action & Game (50) and Creative Editorial (50)
• Ballet & Dance (50), where named positions keep their French names
• Street & Club (40): breaking, popping and locking, house and ballroom, krump, club dance
• 300 working poses: the body doing a job rather than being looked at, in registers such as Load & Carry, Kneel & Floor Work, Push, Pull & Brace and Idle, Wait & Rest

The plates are grey clay mannequins, because they show the shape and not a person. Only the body instruction leaves the node: the mannequin, the black background and the studio light used to draw the plate are stripped, so the generator poses your subject instead of photographing a mannequin. "First Position", for instance, sends a line that begins "Heels touching, both legs fully turned out from the hips so the feet form a straight line".

A pose carries no camera, lens or light. Those belong to the other two nodes.

What does the Shot node add?

The Shot node adds a composition and opens the Shot Lab: 600 frames in nine sections, from The Alphabet — Standard Sizes (extreme close-up to extreme long shot) through angle, depth and lens, to photographic traditions, film movements and The Masters. Every preset describes where the camera sits, how much of the subject it takes and where the subject falls in the frame. None of them moves the camera.

Thirty-one presets wear a lock badge: they change only rendering and tonality—a silhouette, for example—so they can sit on top of a composition you already described.

What is the Resurrected switch?

The Resurrected switch decides whether a filmmaker's name goes into the prompt. Forty-eight presets are named after one of 31 filmmakers—Sergio Leone, Yasujiro Ozu, Roger Deakins and others—but the descriptions themselves name nobody. In Normal mode the node sends the composition alone. In Resurrected mode it puts the full name in front: "Sergio Leone framing — An extremely wide rectangle containing nothing but a horizontal band of the face…".

The switch is off by default for a stated reason: a model handed a name reaches for everything it knows about that person, not just the geometry. It belongs to the node, not the preset, so it stays on while you try other frames. On the 552 presets without a name it is greyed out.

What does the Style node add?

The Style node adds one cue from the Prompt Lab: 1,060 cues in six families—Style (204), Camera (177), Light (140), Lens (99), Stock (128) and Palette (312)—with tag filters and search. Chain several for style, then light, then lens. "Rembrandt" in the Light family sends "A single soft source high and to one side, leaving a small triangle of light on the shadowed cheek…".

One distinction is easy to miss: the Shot node is a still frame, while the Style node's Camera family holds camera moves, such as "Shoulder Cam Follow". A move belongs in front of a video model.

Do a pose and a shot ever contradict each other?

A pose and a shot can contradict each other, and the catalog makes this easy to see. All 740 pose descriptions open with "Full length.", while the Shot Lab also offers close-ups that cut at the chest or the eyes. Chain both and the prompt asks for a full figure and a close-up in the same breath. We did not test which one a model obeys. The safe pairing is a pose with a full or long shot; for a close-up, leave the Pose node out.

How we checked

• What was read: the Pose, Shot and Style node definitions, the code that joins their text into the prompt, and the three catalogs, all on LUVI's main branch on October 20, 2026. Counts were computed from the catalogs themselves.
• Prices: read from the live catalog and from the public model pages on October 20, 2026.
• Results: No outputs were generated for this post, so it makes no claim about output quality.
• Drawbacks: we did not measure how closely any model follows a pose or a composition, or what it does with a pose–shot conflict.

In LUVI

Open Workflows, add a Prompt, then a Pose, a Shot or a Style node, and double-click it to pick. Wire the last one into a generator's prompt port and check the price on the run bar: these nodes add nothing to it. If you work with Claude, the LUVI connector has a search_flow_catalog tool that searches the same libraries, so Claude can pick a pose, a frame or a cue when it builds a workflow for you; building a workflow from Claude walks through it. To try it, create a LUVI account.