Shotcraft
Prompt craft

Image prompts

Subject, framing, optics and the light that decides whether it reads as a photograph.

26 aspects, in the order they work best in a prompt.

Take image prompts with youOne dense page to pin up, or a deck with one section per page. Both built from this page's own content.

Cheatsheet · 1 pageDeck · 31 pages

The order

Working down this list is also the recommended order inside the prompt.

Aspect ratio

Frame shape.

Choose it first. Composition instructions that assume a landscape frame fall apart in a vertical one.

Phrases that work

Medium and style

Photograph, render, illustration or something else.

Ambiguity here produces the uncanny middle ground that looks like neither a photo nor an illustration.

Phrases that work

Model-bound

Brandkit treatment vocabulary is a closed set with defined meanings: flat_vector = solid fills, clean SVG paths, no surface effects; monoline = uniform stroke weight, rounded caps, no fills; vector_gradient = vector-safe linear, radial or duotone gradient with the locked stop count; hand_drawn_vector = intentional stroke variation. "Dimensional/3D treatment is forbidden."

Shot size

How close the camera is.

Same lever as in video and just as reliable. It also controls how much detail the model has to invent.

Phrases that work

Model-bound

Framing is an explicit interview option in product-photoshoot: `[Full body / Three-quarter / Waist up / Closeup on product area]`.

Camera angle

Viewing position.

Phrases that work

Model-bound

game-stylization.md makes the perspective word genre-bound: platformer/runner to side-view; top-down to top-down; puzzle/clicker to flat frontal. For image-to-3D concept images the perspective word is replaced by three-quarter isometric view because it shows top plus two sides.

Composition and framing

Where things sit in the frame.

Explicit composition instructions, including what must stay fully visible, fix it.

Phrases that work

Model-bound

A split layout is used only when the user asks for split, before/after, versus or side by side; "A topical phrase such as `X vs Y` does not itself require a split."

Lens and focal length

Perspective and compression.

Also the cheapest way to signal photographic realism, because a real photograph was taken at some focal length.

Phrases that work

Model-bound

app-cover.md notes the lens block is replaced by style words such as "glossy 3D render / claymation diorama / painterly still" on the graphic route. Vector/logo prompts must never contain lens language.

Depth of field

Focus falloff.

Separates a subject from clutter you did not describe, which is most backgrounds.

Phrases that work

Lighting

Direction, quality and source of light.

Phrases that work

Model-bound

In the thumbnail rig "Only the rim may use a colored accent." Tileable textures need flat lighting so the tile has no focal light. Brandkit mockups explicitly ban "Excessive bloom, haze, depth of field, or cinematic lighting".

Colour and palette

Palette, saturation, contrast.

Keeps a set consistent. Reuse the identical phrase across a batch rather than paraphrasing it each time.

Phrases that work

Model-bound

asset-system.md: a mixed-grade asset kit reads cheaper than no assets; re-generate off-grade pieces "with the hexes named harder".

Two opposite conventions co-exist: website/game prompts NAME the hexes in the prompt; Recraft logo prompts must never contain a colour code and pass hexes as params.

Logo palettes use one, two or three colours, "three is a maximum, not a target or default".

Game palettes are specified BY ROLE, not as one global gamma.

Materials and finish

What surfaces are made of.

Materials are where a render either convinces or does not. Naming the finish is more effective than naming the object.

Phrases that work

Model-bound

Use the generator only when branding must interact with fabric folds, curved packaging, embossing/debossing, foil, print texture, reflections, surface wear, occlusion or realistic perspective; otherwise place the logo deterministically.

Background

What sits behind the subject.

An undescribed background is where models put clutter, text and logos.

Phrases that work

Model-bound

app-cover.md: "One person on an empty backdrop reads as AI slop even when bright, stage a world, not a portrait." For image-to-3D the background is instead "pure flat white, no shadow, no ground".

Mood

The feeling.

One adjective, placed early, quietly resolves dozens of choices in a consistent direction.

Phrases that work

Model-bound

Emotion is one of the allowed surgical-edit scopes; expression changes are made as an edit, not a full regeneration.

Text handling

Whether any lettering appears.

Models add text nobody asked for.

Phrases that work

Model-bound

The dominant convention across the repo is: keep text out of the generation and set type deterministically afterwards. GPT Image 2 is the one model reached for when readable text must be inside the image.

Subject + setting + style (the opening clause)

"Higgsfield models reward concrete, sensory prompts." app-cover.md wants the product's core verb staged as a physical, photographable moment.

Treat it as practice worth copying, not as a count.

Phrases that work

Typography named in the prompt (when text is generated)

Treat it as practice worth copying, not as a count.

Phrases that work

Identity lock / subject fidelity

Up to three referenced identities; face photos are passed first in character order, then the logo, and a manifest line must open the prompt when two or more references are attached.

Treat it as practice worth copying, not as a count.

Phrases that work

Aesthetic / seasonal preset

These are the labelled options the skill offers for restyle and style/mood questions; they become the short intent line handed to the backend enhancer.

Treat it as practice worth copying, not as a count.

Phrases that work

Product-photo genre / mode

For both skills the user-facing prompt is a short intent line only, "Backend assembles the final prompt; never freehand it." `--count 3` makes the backend vary preset, lighting, angle and palette across variants.

Treat it as practice worth copying, not as a count.

Phrases that work

Logo / vector mark specification

One central visual idea per mark, stated in one clause; the exact phrase "no text" must appear in the constraint tail; mark_type routing is evidence-based, defaulting to abstract, never wordmark or combination mark.

Treat it as practice worth copying, not as a count.

Phrases that work

Logo placement on a mockup surface

Full-colour marks go on smooth surfaces; black monochrome on light kraft paper, natural cardboard, pale fabric, stamps, dark-ink screen printing, engraving masks and light uncoated stock; white monochrome on dark paper/boxes/fabric, reverse marks, light-ink screen printing and dark signage.

Treat it as practice worth copying, not as a count.

Phrases that work

Seamless / tileable behaviour

{MATERIAL} in the texture templates is 3-6 concrete words from the reference, e.g. "hand-painted cracked dry earth with small grey stones".

Treat it as practice worth copying, not as a count.

Phrases that work

Key colour for transparency

Models cannot output alpha, so asking for "transparent background" produces a fake checkerboard. Default magenta; switch to green if the subject's own colours are near pink/magenta/purple; blue if both are taken. The chosen colour must appear in both the prompt suffix and the keying script.

Treat it as practice worth copying, not as a count.

Phrases that work

Reference-image role and what it controls

For image-to-image the prompt should describe what changes, not redescribe the input: bad "a man with brown hair in a leather jacket holding coffee, made into anime"; good "transform into anime style, vibrant colors, soft cel shading".

Treat it as practice worth copying, not as a count.

Phrases that work

Design-board direction vocabulary (web/layout imagery)

Commit to one option per category before prompting and hold it across all boards; one board per section, landscape 16:9 or 3:2, never one tall full-page image.

Treat it as practice worth copying, not as a count.

Phrases that work

Reference photos for identity training

5-20 photos, 8-12 the sweet spot, clear face with eyes visible, single person per photo, sharp, ≥1024×1024 ideal.

Treat it as practice worth copying, not as a count.

Phrases that work

Image-to-3D input framing

One figure per input image, a 4-view character sheet reconstructs as "a statuette of 4 fused figures".

Treat it as practice worth copying, not as a count.

Phrases that work

Prompt structure

The shapes these prompts actually take, rather than a description of one. For a complete worked set built on one of them, see hand-drawn VFX fusion.

The base order is subject + setting + style to camera (lens, angle, motion) to lighting to style/medium, kept under about 200 tokens because "Models distort with very long prompts". Specialised skills expand this into numbered block contracts assembled in a fixed order.

The thumbnail contract has eleven blocks: frame to scene brief to text to subjects to key elements to logo to location to composition to background to lighting on people to grade, with per-character IDENTITY LOCK paragraphs and an opening IMAGE REFERENCES manifest when two or more references are attached. The cover doctrine has six: subject + action to world + props to scale + placement to palette to light + lens to and "Always end with: no text, no letters, no logos, no captions, no UI".

Game assets concatenate exactly four parts in order: `<kind template> + <asset description> + <STYLE FORMULA byte-identical> + <kind suffix>`. Recraft logo prompts are "exactly one continuous enhanced Recraft prompt string per candidate, no bullet points inside the prompt" ordered mark type to central subject/mechanism to shape logic to style register to palette behavior to composition to constraint tail, and "Every clause must materially affect the drawing".

Brandkit mockups use bracketed labelled sections, [CREATE ONE FINISHED BRANDED MOCKUP], [AUTHORITATIVE LOGO], [PLACEMENT LOCK], [PHYSICAL APPLICATION], plus a compact [BRAND LOCK, DO NOT DEVIATE] block repeated verbatim across related assets.

Negatives

Grouped by the failure they address. Keep the list short: on a model with no separate negative field, every one of these is just another word competing for attention.

Text and branding

Anatomy and faces

Frame artefacts

Content and safety

Style and clutter

Other

Take it with you

Take image prompts with youThe same reference, as a printable sheet or a presentation.

Cheatsheet · 1 pageDeck · 31 pages

Back to prompt craft · Cinematic language · Downloads