op-ed-cover

Imported from okou-ai/vm0-skills (illustration-template/op-ed-cover/SKILL.md). Install upstream with npx skills add okou-ai/vm0-skills --skill op-ed-cover. Copyright stays with the author.

/op-ed-cover — locked editorial magazine cover style

This skill produces a single editorial illustration laid out like a page from a print magazine essay: a flat-vector scene with paper-grain texture, expressive elongated ink-line characters, a muted retro palette, and a serif headline block in the lower-right corner that frames the scene as a witty opinion piece.

The whole point is the editorial tension between the picture and the headline — the picture sets a moment, the headline gives it a contrarian or knowing voice. Headlines should sound like they came from the back pages of a New Yorker / Monocle / Time-Out essay column.

Prompt interpretation

The user will give a short brief, usually some combination of:

• a scene metaphor (e.g. "Tokyo ramen master", "London tube commute", "Italian barista")
• a palette direction (e.g. "mustard + forest green", "dusty rose + ink navy")
• a complexity level (L1 single hero / L2 small ensemble / L3 busy crowd)
• optionally, a headline or a headline voice ("declarative", "punning", "appreciative")

For each invocation:

1. Brainstorm the scene vividly. Picture the time of day, the weather, the specific gesture the main character is mid-doing, the props that signal place. Avoid generic "person in venue" — go for a moment that has tension or charm.
2. Pick a sharp, original headline. Use one of the headline voices below. Do not default to "In Defense of the [Adjective] [Archetype]" — that formula is the most identifiable cliché of the style. Mix declaratives, puns, sensory metaphors, and quiet appreciations.
3. Compose the page with the scene filling the canvas and a solid or translucent color block in the lower-right corner carrying the serif headline + two short lines of small italic body copy (faux lorem ipsum is fine).
4. Lock the style. Every invocation keeps the flat-vector + grain treatment, the elongated ink-line figures, the retro editorial mood, and the headline-block placement.

Locked style axes (NEVER vary)

Rendering

• Flat-vector with subtle paper-grain texture overlay — like a Riso print or a high-quality magazine page reproduction.
• Gentle painterly shading — soft modeling on faces, fabric folds, ceramic, steam. Never airbrushed gradients, never 3D rendering.
• Subtle film grain throughout the canvas, not just on the figures.
• No drop shadows, no glow effects, no digital cleanness.

Character drawing

• Elongated expressive forms. Slightly stretched proportions — long necks, long limbs, slightly tilted heads. Characters lean.
• Confident ink-line outlines with subtle weight variation; lines feel hand-drawn, not vector-uniform.
• Period-appropriate clothing — 1950s–60s editorial-illustration silhouettes (slim suits, structured coats, aprons, hats).
• Expressions are restrained. Characters wear knowing, calm, slightly skeptical looks. Almost never wide-eyed or shouting.

Composition

• Portrait orientation (roughly 7:9 / 4:5), magazine-page aspect ratio.
• Scene fills the canvas edge-to-edge — no white or paper border framing.
• Headline overlay block sits in the lower-right corner, occupying roughly the bottom-right quarter of the canvas.
• The headline block is a solid or translucent rectangle in an accent color from the scene's palette.
• Inside the block: large serif headline (2 lines), then two short lines of small italic body copy beneath.

Mood

• 1950s–60s editorial illustration — think Saul Steinberg, Charley Harper, mid-century Esquire / Punch / Vogue spot illustrations, with a contemporary editorial revival sensibility.
• Wry, observational, gently contrarian. The picture is never angry, never sentimental.

Variable axes (the 5 dials)

1. Palette

Pick three roles: background hue (dominates the canvas), accent (used in the headline block and 1–2 props), figure ink (outlines and key fills). Avoid more than three colors.

Battle-tested palettes:

• Dusty slate-blue · cream · brick-red (original)
• Mustard ochre · cream · forest green
• Dusty rose · warm cream · ink navy
• Sage olive · cream · burnt orange
• Mocha brown · cream · oxblood
• Pale teal · cream · charcoal

When you write the palette into the prompt, be explicit about what each color is for and what colors are off-limits — image-to-image models otherwise drift back to the most-seen palette (dusty blue + brick red).

2. Complexity

• L1 — Single hero. Tight close-up, one character mid-action, 1–3 supporting props. The figure carries the whole frame.
• L2 — Small ensemble. 2–4 characters in a clear shared setting (a train carriage, a counter, a stoop). Spatial relationship between them is part of the story.
• L3 — Busy crowd. 6+ characters, layered vignettes, dense background detail. Reads like a Where's-Waldo of editorial archetypes — but each character still has a specific, observed beat.

L3 is the hardest to keep editorially soft — character softness erodes as count rises. Pull back on character count if the result starts looking like a cluttered children's-book scene.

3. Scene

The scene IS the metaphor. Pick a specific cultural microcosm — a small ritual or social space where a recognizable archetype lives. Examples:

• A Tokyo ramen-ya at closing time
• A London tube carriage at 8am
• A New York Jewish deli at lunch rush
• An Italian café where the barista refuses to be rushed
• A Parisian bakery at dawn
• A Hong Kong dai pai dong after midnight
• A Buenos Aires tango bar mid-set

Avoid "person standing in generic place." The scene must have temporal specificity (rush hour, last call, opening) and textural detail (handrails, salamis on hooks, lacquered chopsticks, brass espresso lever).

4. Headline voice

Pick one — do not default to "In Defense of…".

• Noun-led declarative — The Slowest Bowl in Tokyo, Pastrami at Full Volume, Last Call at the Counter
• Pun / local idiom — Mind the Eye Contact (riff on "Mind the Gap"), Tokyo, by Ladle, Brisket and Brevity
• Quiet appreciation — An Hour at the Ramen-ya, On Patience, By the Bowl, Notes from the Counter
• Contrarian (the original formula — use sparingly) — In Defense of the [Adjective] [Archetype], A Quiet Defense of [Behavior]

Keep headlines to 5–8 words, two visual lines. Use a small italic faux-body line (1–2 short sentences of lorem ipsum is fine) under the title.

5. Headline block style

• Solid opaque rectangle — strongest contrast, reads as a poster block. Use when the scene is busy (L3) or low-contrast.
• Translucent rectangle — softer, lets the scene bleed through. Use for L1/L2 when the figure work should breathe.
• The rectangle's fill color must be the palette's accent, not the background.
• Type is always serif, weight bold-to-medium, color is the palette's lightest tone (cream / ivory).

How to brief an image-to-image model

This style is most reliably reproduced via an image-to-image model that accepts multiple reference images — both reference images in this folder plus a strong textual override of palette, scene, and headline.

Key briefing principles:

• Lead with "in the EXACT style of the reference images" and explicitly call out flat-vector + paper grain, expressive elongated ink-line characters, 1950s–60s editorial mood, portrait magazine composition.
• Write the palette with role tags (background dominates the canvas, accent for the headline block, ink for figure outlines) and explicit negatives (NO blue, NO red).
• Specify complexity level in plain language (Level 1 — single hero close-up).
• Describe the scene with concrete props and gestures — the model will fill in generic decor otherwise.
• For the headline: give the exact text in quotes, specify lower-right corner, name the block fill color, and require serif type.
• Add no logos, no other text outside the headline block to suppress stray watermarks and signage.

Model notes (semantic, not endpoint-specific):

• Multi-reference image-to-image models in the nano-banana / SDXL-IP-Adapter family handle this style well.
• For palette discipline, use explicit negative color callouts in the prompt — these models otherwise regress to the most-seen palette of the reference set.
• For headline legibility, prefer a solid opaque block over a translucent one when text rendering is weak; the model can lose translucent serif text.

Reference outputs

Three reference outputs are committed alongside this skill to anchor the style. Each demonstrates a different point on the palette × complexity × scene matrix:

• ref-slowest-bowl.png — L1 single hero, mustard ochre + forest green, Tokyo ramen master, declarative headline ("The Slowest Bowl in Tokyo").
• ref-mind-eye-contact.png — L2 ensemble, dusty rose + ink navy, London tube carriage, pun headline ("Mind the Eye Contact").
• ref-pastrami-full-volume.png — L3 crowd, sage olive + burnt orange, NYC Jewish deli, sensory headline ("Pastrami at Full Volume").

When briefing a new piece, treat these three as the style anchor and override only the variable axes.

Anti-patterns

• ❌ Defaulting every headline to "In Defense of [Adjective] [Archetype]". One per series, max.
• ❌ Generic "person standing in venue" — always anchor to a specific cultural microcosm with temporal and textural detail.
• ❌ Bright pure white background. The background should be a saturated muted hue from the palette.
• ❌ Photographic shading, lens-blur depth-of-field, or 3D lighting. Keep shading flat and painterly.
• ❌ Centered headline blocks. The block always sits in the lower-right.
• ❌ Modern sans-serif headlines. Always serif.
• ❌ Logos, brand marks, URLs, or stray decorative text outside the headline block.
• ❌ Cluttering L3 scenes past ~10 characters — the editorial softness collapses.