Video asset library

on navy

on gold

on cream

Master style prompt — prepend to every engraving subject (swap metallic gold #C2A15A for cream #F4EFE4 or deep navy #0A1729 to make the other two inks):

Antique copperplate engraving illustration of {SUBJECT}, rendered entirely in
fine cross-hatched linework in a single metallic gold color, hex #C2A15A, on a
fully transparent background. Style of a 19th-century steel-plate etching from
a law-book frontispiece: precise parallel hatching, controlled line weight, no
solid fills. Dignified heritage law-firm aesthetic. Subject centered, generous
negative space, no border, no frame, no text, no watermark, no background.
Square, 2048 x 2048.

All eighteen motifs are installed, each as <name>-gold.webp, <name>-ivory.webp, and <name>-navy.webp in public/design-system/illustrations/.

Batch generation prompt — the recipe these were made with; rerun to extend the set or add an ink (regenerate a batch with the color swapped in the HOUSE STYLE line):

You are generating the complete engraving asset library for Pierce Law
Group's video design system. Follow these rules exactly for every image in
this session.

HOUSE STYLE — applies to every image:
Antique copperplate engraving illustration rendered entirely in fine
cross-hatched linework in a single metallic gold color, hex #C2A15A, on a
fully transparent background. Style of a 19th-century steel-plate etching
from a law-book frontispiece: precise parallel hatching, controlled line
weight, no solid fills. Dignified heritage law-firm aesthetic. Subject
centered, generous negative space. Square, 2048 x 2048, PNG with
transparency.

HARD RULES — check every image against these before moving on; if one
fails, regenerate it before continuing:
- One color only: gold #C2A15A. No gradients, glows, sheen, drop shadows,
  or secondary colors. Gold is ink, not light.
- Fully transparent background. No paper texture, vignette, or backdrop.
- No text, letters, numbers, watermarks, borders, or frames.
- Identical stroke weight and hatching density across all images. The first
  approved image is the style anchor — match every later image to it.
- Linework only. No photorealism, no 3D shading.

PROCESS:
Work through the batches in order. Generate ONE image per asset, captioned
with its filename. After each image passes the hard rules, continue to the
next asset automatically without asking me.

BATCH 1 — needed now (these replace placeholders in the current video):
1. will-gold.png — a folded last will and testament document tied with a
   ribbon and a round wax seal, a quill pen resting across it
2. letters-gold.png — an unfolded formal court letter bearing a large round
   embossed wax seal and a hanging ribbon
3. hourglass-gold.png — a classical hourglass on a small plinth, sand
   caught mid-fall

BATCH 2 — property and assets:
4. house-gold.png — a modest two-story American family home with a front
   porch, three-quarter view
5. car-gold.png — a classic family sedan in three-quarter view, parked
6. ledger-gold.png — an open leather-bound bank ledger with ruled columns
   and a fountain pen

BATCH 3 — notices and venue:
7. envelope-gold.png — a formal notice envelope closed with a wax seal,
   slightly angled
8. nc-map-gold.png — the outline map of the state of North Carolina with
   fine county hairlines, no labels

When all 8 are done, list any that needed regeneration and why, so I can
double-check them before export.

Which background — and how often? A video isn't one flat color; it's a navy body with gold and ivory used as punctuation. Each surface has a job and a budget.

The arc. Open and argue on navy, drop to ivory for a calm/definition beat, spike to gold for the one line that matters, resolve on ivory or a navy photo CTA. Gold and ivory are opposites — one raises the voice, one lowers it — so they balance each other across the runtime. Here's how the three "Probate, Plainly" videos spend their surfaces:

Which ink is legible on it? Measured WCAG contrast of each engraving ink against each surface — a fine-line engraving needs roughly 3:1 or better to hold its hatching.

The four you'll actually use:

Every default scene.

When gold competes with a gold headline.

The one gold moment.

The quiet, editorial interlude.

Master photo prompt — for new photography, splice a subject the same way:

Warm documentary photograph, editorial style, of {SUBJECT}. Natural
golden-hour light, shallow depth of field, muted heritage palette of deep
navy, cream, and warm gold. No faces in sharp focus. Quiet, dignified mood —
a North Carolina family setting. Photorealistic, 16:9, high resolution,
no text, no watermark.

Batch generation prompt — the recipe these were made with; rerun to extend the photo set:

You are generating the photography library for Pierce Law Group's video
design system. These are emotional backdrop plates for law-firm explainer
videos — they sit behind a dark navy overlay with text on top, so mood and
composition matter more than detail. Follow these rules exactly for every
image in this session.

HOUSE STYLE — applies to every image:
Warm documentary photograph, editorial style. Natural light only — golden
hour or soft window light. Muted heritage palette: deep navy shadows, cream,
warm gold highlights; nothing saturated. Quiet, dignified, middle-class
North Carolina family setting. Photorealistic, shallow depth of field,
16:9 landscape, high resolution.

HARD RULES — check every image against these before moving on; if one
fails, regenerate it before continuing:
- No faces in sharp focus: subjects turned away, softly out of focus, or
  framed below the shoulders. Hands may carry the story.
- No posed stock-photo smiles at camera, no studio lighting.
- No text, signage with readable words, logos, or watermarks anywhere.
- Leave breathing room: at least one third of the frame calm and
  uncluttered, since a headline or caption will sit over it.
- Consistent color grade across all images. The first approved image is
  the grade anchor — match every later image to it.

PROCESS:
Work through the batches in order. Generate ONE image per asset, captioned
with its filename. After each image passes the hard rules, continue to the
next asset automatically without asking me.

BATCH 1 — loss and the estate (video hooks):
1. empty-armchair.png — a quiet living room in morning light, a worn
   empty armchair with a folded blanket, a side table with reading glasses
2. kitchen-table-papers.png — an older kitchen table covered with folders,
   envelopes, and handwritten papers, two people's hands sorting them
3. farmhouse-drive.png — a modest North Carolina family house seen from
   the end of its driveway at golden hour, trees framing the view

BATCH 2 — the process:
4. courthouse-steps.png — the stone steps and columns of a small county
   courthouse in soft morning light, one person ascending, seen from behind
5. mailbox-notice.png — a hand taking a formal envelope from a rural
   mailbox at the end of a driveway, house soft in the background
6. attorney-consult.png — an across-the-desk consultation seen from the
   side: hands, documents, and a pen in focus, faces out of frame

BATCH 3 — resolution (video outros):
7. house-keys-handover.png — a close-up of house keys passing from an
   older hand to a younger one on a front porch, warm evening light
8. family-embrace.png — a multi-generation family embracing in a backyard
   at dusk, seen from behind, string lights soft in the distance

When all 8 are done, list any that needed regeneration and why, so I can
double-check them before export.

Type hierarchy — always contrast adjacent levels on more than one axis. Every scene stacks at most three type levels (eyebrow/kicker, headline or numeral, body/unit). Two levels that only differ in font-size read as the same level once they're both small caps at a glance — a caption and a unit label can end up competing instead of ranking. Vary at least two of size, weight, color, and case between any two adjacent levels: e.g. a bold gold all-caps eyebrow above a lighter-weight, muted-color unit tag, not two same-weight gold caps blocks at different sizes.

Numerals are always lining (modern), never old-style. Cormorant Garamond defaults to old-style figures (varying baseline, text-like), which read as archaic at the scale a giant stat numeral renders. The TimDS composition applies font-variant-numeric: lining-nums at its root so every digit uses an even baseline.

An icon must never crowd its own label. A trimmed engraving's ink can sit flush against its own bounding box — a pen tip, a roofline — and different motifs are width- or height-bound at random depending on their aspect ratio, so sizing an icon by percentage of its box gives inconsistent real clearance from asset to asset. Cap the icon in fixed pixels well under its box (e.g. a 300 × 176 px box holding an image capped at 260 × 140), not 100% — that guarantees the same margin regardless of what the specific engraving's ink reaches. A flex gap between icon and label is not sufficient on its own; it only spaces two boxes, not the ink inside them.

A manual line break doesn't guarantee a line. A hard <br/> starts a new line, but if the clause before it is still too wide for the frame, the browser auto-wraps that clause again — stranding a short orphan between two full lines (e.g. a 6-word clause wraps to "…an account" / "in days." before the second clause even starts, reading as three ragged lines instead of two clean ones). Two defenses, used together: set textWrap: "balance" on any block that might wrap, so the browser distributes words evenly instead of leaving a short last line; and size the type (or shorten the copy) so each intended clause actually fits on one line — verify by rendering the frame, don't assume from the JSX.