/* ============================================================================
   HOME OPENING  —  .sfop
   ----------------------------------------------------------------------------
   Section 1 of the home page, built 2026-11-10 to the arrangement on
   wolverineworldwide.com. The full study of the reference, with the measured
   intro timeline, is docs/research/2026-11-10-wolverine-hero-section.md —
   read it before changing any number in here.

   THIS IS NOT .sfv. The live hero (page 42, container sfvideo1, widget
   sfvideow1) is untouched; this is a separate section with its own namespace
   so the two can be compared before anything is swapped. Nothing enqueues
   this file yet — see home-opening.js for why.

   The namespace .sfop was free as of 2026-11-10. Grep assets/css before
   picking another one: sf-hero, sf-intro, sf-net, sf-pdp, sfasr, sfb, sfc,
   sfcl, sfe, sffw, sfidx, sfl, sfmq, sfp, sfrail, sfrdr, sfrev, sfs, sfsr,
   sft and sfv are all taken.

   THE THREE THINGS THAT MAKE THE LOOK, in order of how easily they are lost:

   1. THE PANEL IS INSET, NOT FULL BLEED. The page shows down all four sides —
      8px on phones, 16px from 768px up. The rounded corner is only legible
      because of that margin: take the inset away and you are rounding a
      corner against the viewport edge, which draws nothing. On this site the
      gutter shows the house carbon weave rather than the reference's white.

   2. THE PANEL FITS THE VIEWPORT EXACTLY. Its height is 100svh minus both
      insets, so the section totals one screen and all four corners are
      visible.

      THIS REVERSED ON 2026-11-10 and the old reasoning is worth knowing,
      because it was not arbitrary. The panel used to be a full 100svh with
      no bottom padding, so its bottom edge sat one inset BELOW the fold —
      which is what the reference does, and it makes the panel read as
      tucked under the next section rather than floating in a frame. The
      operator asked for it to fit instead, on every screen size. If anyone
      ever wants the tucked look back, it is the `padding` on .sfop and the
      `height` on .sfop__panel, together.

   3. LINE-HEIGHT 0.79 ON THE TITLE. Smaller than 1, so the three lines
      interlock into one block of type. At a normal 1.1 the same three words
      read as a paragraph and the whole thing falls apart.

      (This one has moved too: the title is uppercase now and runs at .9 —
      all caps has no descenders to interlock. See the title block below.)

   Phones had neither inset nor radius until 2026-11-10, following the
   reference, which switches both off below its md breakpoint. They have both
   now, at smaller values.
   ============================================================================ */

/* BORDER-BOX ON THE ROOT AS WELL AS THE DESCENDANTS. The panel is
   height:100svh AND carries bottom padding, so under content-box it comes out
   100svh + 48px — eight hundred and ninety-eight pixels of panel in an 900px
   window, which pushes the whole title below the fold and leaves the bottom
   corners nowhere near where this file says they are. Measured before the
   fix: 678.4 against a 630px viewport. Every per-page stylesheet in this
   theme sets box-sizing on `.superfly-X *` and not on the root, which is the
   same trap from the other direction (see terms.css). State both. */
/* ----------------------------------------------------------------------------
   THE DNA SECTION'S PULL-UP MUST BE CANCELLED, AND THIS IS THE FILE THAT OWNS
   THAT CONTRACT.

   The band below this one, .sfd, is written to slide up OVER the hero:

       .sfd{ margin-top: calc(-1 * var(--sfv-overlap, 100svh)); z-index: 2; }
       .sfd{ min-height: var(--sfv-overlap, 100svh); }

   That made sense when section 1 was .sfv — a tall sticky stage that PUBLISHED
   --sfv-overlap from its own script, so the DNA band knew how far to climb.
   This section is one screen tall with nothing sticky in it and no overlap to
   give, and the old script is gone with the old hero, so nothing sets the
   variable any more and .sfd falls back to its 100svh default: it climbs a
   full viewport and lands squarely on top of this section.

   Caught on the live page during the 2026-11-10 deploy. Everything measured
   correctly — section top 0, panel 8/8/1414/900 — while elementFromPoint at
   the middle of the screen returned .sfd__stage-n, because the geometry WAS
   right and something else was simply painting over it. Measuring the section
   you changed cannot catch this; ask what is on top.

   Publishing 0px here is the same contract the old hero had, with the value
   this hero actually wants. It is a :root declaration from a section
   stylesheet, which is normally the wrong thing to do — it is deliberate, the
   file only loads on the front page, and the alternative is editing a second
   widget's markup in the database.

   IF THE DNA BAND IS EVER MEANT TO OVERLAP AGAIN, this is the line to change,
   not .sfd's own markup.
   ---------------------------------------------------------------------------- */
:root{
	--sfv-overlap: 0px;
}

/* BORDER-BOX ON THE ROOT AS WELL AS THE DESCENDANTS. (continued below) */
.sfop,
.sfop *,
.sfop *::before,
.sfop *::after{
	box-sizing: border-box;
}

.sfop{
	/* PHONES GET THE INSET AND THE CURVE TOO, since 2026-11-10 on the
	   operator's call. The reference switches both off below its md
	   breakpoint and this followed it until now — full bleed, square
	   corners. Smaller than desktop's 16/20 because the proportion is what
	   reads, not the pixel count: 8px of gutter on a 375 screen is the same
	   share of the width as 16px on a 1440 one.

	   ROOMIER TITLE (operator, 2026-11-10): the title's own margins went up,
	   12 -> 20 on the left and 16 -> 28 at the foot, and the ramp below was
	   re-fitted to pay for it — both come out of the same line of measure.

	   The gutter went 8 -> 6 in the same pass and straight back to 8 when
	   the operator saw it. That round trip cost nothing but it ate half the
	   slack at 320px: 4px now, not 8. The ramp was NOT reduced again for
	   those 4px, so this is as tight as it goes. Anything that widens the
	   gutter, the left margin or the type from here overflows a 320 screen
	   — check the sum under SIZING RULE below before touching any of
	   them. */
	--sfop-inset: 10px;     /* gutter around the video panel */
	--sfop-radius: 12px;    /* SHARED WITH THE NAVBAR - see the note below */
	--sfop-gutter: 20px;    /* title's left margin, inside the panel */
	--sfop-foot: 28px;      /* title's bottom margin, inside the panel */

	/* THE TITLE SIZE, in one place so it can be dialled without going near
	   the ramp. Brought down about 30% on 2026-11-10 — the operator's call;
	   the first build matched the reference's ramp literally and at 1440 that
	   put "Research." at 764px, nearly the full measure.

	   "Research." is 5.03 times the font size wide in Archivo Black, so the
	   line's width is roughly font-size x 5. That is the number to check
	   against the panel's inner width (viewport - 16px inset - 80px gutters)
	   before changing this. Now: 107px at 1440, so the line is ~540 of a
	   1329 measure.

	   Mobile floor is 40px, desktop ceiling 120px. */
	/* Reduced twice on the operator's call, 2026-11-10: the reference's own
	   ramp put "Research." at 764px of a 1329 measure, then 107px was still
	   too big once the title went uppercase — and uppercase is ~24% wider
	   again. Now 84px at 1440, so "RESEARCH." is ~520 of 1329.

	   The multiplier to redo this arithmetic with: uppercase Archivo Black
	   measures about 6.2x the font size for "RESEARCH.". Check that against
	   the panel's inner width (viewport - 32px inset - 80px gutters) before
	   changing it. */
	/* Re-pitched 2026-11-10 (third time): the previous ramp was tuned from
	   the desktop end and collapsed to its 36px floor across every phone
	   width, so a 320 and a 430 screen got identical type and both looked
	   undersized.

	   This one is a straight line through the two sizes that actually
	   matter — 48px at 375 and 84px at 1440 — so it is genuinely responsive
	   between them instead of flat at one end:

	       320   41px      768   58px
	       375   43px     1024   67px
	       430   45px     1440   84px

	   Re-fitted twice more: once when phones gained the inset, and again
	   when the title's own left margin went 12 -> 20px. A wider margin is
	   measure spent, and at 320px there is very little of it.

	   The 6rem cap binds from ~1795px up; the floor effectively never does.
	   No breakpoint jump: one expression covers phone through desktop.

	   SIZING RULE: uppercase "RESEARCH." in Archivo Black measures 6.2x the
	   font size. Check it against the panel's inner width — viewport minus
	   2x inset minus 2x gutter. The binding case is 320px:

	       available  320 - 2x10 - 2x20 = 260
	       needed     41 x 6.2          = 254
	       slack                          6px

	   SIX PIXELS. Four things share them — the inset, the left margin, this
	   ramp and the length of the word. Move any one and redo the sum; a
	   headline longer than "RESEARCH." needs it redone from scratch.

	   The gutter went 8 -> 10 on 2026-11-10 and the ramp came down with it,
	   41px at 320 instead of 42. At 42 the line measured 260.4 against 260
	   available — it would have been clipped by the panel rather than
	   overflowing, so it would have LOOKED fine while being a fraction
	   short. Each gutter pixel costs two of measure, and at 320 there is
	   nowhere left to take it from but the type. */
	--sfop-title-size: clamp(2.25rem, 1.794rem + 3.84vw, 6rem);

	/* The scrim is the one deliberate departure from the reference, which
	   lays white type straight on the footage with nothing over it. That
	   works because their footage is graded for it. Ours is the teardown
	   clip and it is dark and busy at the bottom, where both the title and
	   the button sit. Set --sfop-scrim to 0 to match the reference exactly
	   once the footage is final. */
	--sfop-scrim: 1;

	/* THE HEADER IS STICKY AND IN FLOW, so it takes 66px off the top of the
	   page and everything below starts under it. This section wants to begin
	   at the top of the VIEWPORT — the glass navbar is meant to sit over the
	   footage, as it does on the reference — so it pulls itself up by exactly
	   that much.

	   home-opening.js measures the gap and writes this property; the 0px
	   default means that with no JavaScript the section simply sits below the
	   bar, which is untidy but not broken. The hero this replaced did the
	   same thing with --sfv-under, and for the same reason.

	   Do NOT hardcode 66px. It is the logo height plus the bar's padding and
	   it has moved before. */
	--sfop-under: 0px;

	position: relative;
	z-index: 0;
	margin-top: calc(-1 * var(--sfop-under));

	/* PADDING ON ALL FOUR SIDES, so the panel fits the viewport exactly.

	   It used to be `var(--sfop-inset) var(--sfop-inset) 0` — no bottom
	   padding, the panel a full 100svh starting one inset down, so its
	   bottom edge and both bottom corners sat below the fold. That is what
	   the reference does and the note at the top of this file argued for it:
	   the panel reads as tucked under the next section rather than floating.

	   The operator asked for the opposite on 2026-11-10 — the video should
	   fit the screen at every size, with the gutter visible all the way
	   round. So the panel is now 100svh MINUS both insets and the section
	   totals exactly 100svh. All four corners are visible. */
	padding: var(--sfop-inset);

	/* THE HOUSE SURFACE. Every band on this page carries both diagonals —
	   the light one is the sheen on the tow, the dark one the shadow under
	   it, and with one alone it reads as corduroy. Settled 2026-10-20. */
	background-color: var(--wp--preset--color--base, #171717);
	background-image:
		repeating-linear-gradient( 45deg, rgba(255,255,255,.025) 0 2px, transparent 2px 4px),
		repeating-linear-gradient(-45deg, rgba(0,0,0,.35)        0 2px, transparent 2px 4px);
}

/* ---------------------------------------------------------------------------
   1 · the panel
   --------------------------------------------------------------------------- */

.sfop__panel{
	position: relative;

	/* The two insets come off the height, so section = 2x inset + panel =
	   exactly 100svh and the whole thing fits one screen.

	   svh, not vh or dvh, on purpose: svh is the SMALL viewport height, the
	   one you get with the phone's toolbars showing. It is the only unit
	   that guarantees the panel fits at the worst moment. The cost is that
	   when the toolbars hide and the viewport grows, a little page shows
	   below the panel — the right trade for "fits every screen". */
	height: calc(100svh - 2 * var(--sfop-inset));

	display: flex;
	flex-direction: column;

	/* BOTTOM-ALIGNED ON PHONES TOO, since 2026-11-10. It was
	   space-between with the title at 40svh and the SHOP NOW button at the
	   foot. With the button removed that left the title stranded mid-screen
	   over an empty bottom half, so the phone layout now matches desktop:
	   the title sits on the foot padding.

	   The top padding is no longer what positions anything — flex-end does
	   that. It stays as a guard: if the title ever grows (longer words, a
	   fourth line) this is what stops it climbing into the navbar. */
	justify-content: flex-end;
	padding: 22svh var(--sfop-gutter) var(--sfop-foot);

	color: #fff;
	isolation: isolate;

	/* REQUIRED BY THE INTRO, not by the layout. The media starts 60% of its
	   own height below where it lands; without this it hangs out over the
	   next section for the first second. The reference clips it the same
	   way. Safe here because nothing in this section is sticky — see the
	   note in smooth-scroll.css about pins dying on a clipped ancestor. */
	overflow: hidden;
}

/* ---------------------------------------------------------------------------
   2 · the video, and the curve
   --------------------------------------------------------------------------- */

.sfop__media{
	position: absolute;
	inset: 0;
	z-index: 0;
	overflow: hidden;
	border-radius: var(--sfop-radius);
	pointer-events: none;

	/* No will-change: nothing animates this element any more. The intro used
	   to scale and move it — and scaling it scaled the rendered corner too,
	   which was the "zooms to fit the curve" read — but the operator asked
	   for the video to display at full size straight away (2026-11-10), so
	   the panel is static and holding a compositor layer for it would be
	   pure cost. */
}

.sfop__video{
	position: absolute;
	inset: 0;
	width: 100%;
	height: 100%;
	object-fit: cover;
	display: block;
	/* will-change dropped with the counter-zoom — see .sfop__media above. */
}

.sfop__scrim{
	position: absolute;
	inset: 0;
	pointer-events: none;
	opacity: var(--sfop-scrim);
	background: linear-gradient(
		180deg,
		rgba(10,10,12,.38) 0%,
		rgba(10,10,12,0)  20%,
		rgba(10,10,12,0)  52%,
		rgba(10,10,12,.62) 100%
	);
}

/* ---------------------------------------------------------------------------
   3 · the title
   --------------------------------------------------------------------------- */

.sfop__title{
	position: relative;
	z-index: 1;
	margin: 0;

	/* ─────────────────────────────────────────────────────────────
	   THE SAME TYPE AS SECTION 2's TITLE, VALUE FOR VALUE — operator,
	   2026-11-10: "for section 1 homepage also i want the title is
	   same font style and size font like section 2". The two titles
	   are the first two things on the page and they now read as one
	   voice.

	   THESE FIVE VALUES ARE DUPLICATED FROM category-grid.css ON
	   PURPOSE, as two literals rather than one shared token — the
	   same decision, for the same reason, that --sf-navbar-bar-radius
	   and --sfop-radius are kept apart. A custom property substitutes
	   at computed-value time on the element where it is DECLARED, so
	   a :root token declared in one stylesheet and read from another
	   depends on load order and fails silently. Both files happen to
	   load only on the front page today, which is exactly what would
	   make such a bug survive unnoticed.

	       CHANGE BOTH FILES OR NEITHER.
	       category-grid.css → .sfcg__title and the ladder under it.

	   What came off with Archivo Black:

	   - `--sfop-title-size`, the one fluid clamp this section used from
	     phone to desktop. It is STILL DECLARED above and nothing reads
	     it now; it is left in place with its whole width-budget note,
	     because that note is the only record of the 320px arithmetic
	     and it applies again the moment this goes back to a heavy face.
	   - `.79 / -.05em`, whose justification was specific to Archivo
	     Black: lowercase descenders riding into the next line's caps
	     is what made THAT face interlock at .79, and the note is still
	     above the old clamp. Inter's descenders are shorter and its
	     caps are a different height, so carrying .79 across would have
	     been carrying a number whose stated reason no longer held.

	   Section 2's values are sohub.digital's own, and so these are too:
	   `leading-none` and -0.075em, on their size ladder. They were
	   briefly .95 / -.03em and ~11% smaller, and were put back to the
	   reference's the same day — see category-grid.css, which carries
	   the measured table.

	   The title is three one-word lines, so the size ladder's job here
	   is different from section 2's — nothing wraps, and the binding
	   case is the 320px width budget in the note above rather than a
	   line count. Inter 600 is far narrower than Archivo Black, so
	   that budget stopped being tight: see the measured figures in
	   CLAUDE.md.
	   ───────────────────────────────────────────────────────────── */
	font-family: 'Inter', system-ui, -apple-system, sans-serif;

	/* **700, NOT THE 600 THE PROSE ABOVE DESCRIBES** (operator,
	   2026-11-12: "now i want you make section 1 , 2 and section 4
	   using weight 700"). It followed an audit of the whole home
	   page: section 3 alone was at 700 and every other band at 600,
	   and the answer was to bring the others UP rather than take
	   that one down. Everything above about sohub's own 600 is the
	   older state and is kept because the rest of it — the family,
	   the ladder, the leading and the tracking — still holds.

	   ALL FOUR FILES THAT CARRY THESE VALUES AS LITERALS ARE IN STEP
	   AGAIN: this one, category-grid.css, bike-ribbon.css and
	   video-panel.css. **Change them together or not at all** — the
	   note above says why they are literals rather than one shared
	   token. */
	font-weight: 700;

	/* ─────────────────────────────────────────────────────────────
	   ONE FLUID SIZE, NO STEPS — AND IT IS BOUND ON BOTH AXES
	   (2026-11-11, operator: "if height is difference the text is
	   too big in difference screen size .. i want to make the text
	   title is smoothly follow the responsive screen size").

	   It replaces a six-step em ladder. Three terms, each with its
	   own job, and the smallest one wins:

	     5.5vw + 5svh the size itself. BOTH axes on purpose — a
	                  vw-only ramp under-sizes a tall narrow phone
	                  and over-sizes a short wide laptop, which was
	                  the original complaint.

	                  THE vw HALF CAME DOWN 8 -> 5.5 ON 2026-11-11
	                  (operator, against a screenshot of a ~1780px
	                  laptop: "but in desktop size ... too bigger").
	                  At 8vw the same window gave ~180px. The svh
	                  half was left at 5 on purpose: it is what
	                  carries a phone, where nothing was wrong, so
	                  this is a desktop-only reduction — 390x844
	                  moves 65.8 -> 63.7, under two pixels.

	     23vw - 24px  the NO-CLIP ceiling. "Research." measures
	                  4.131x the font size in Inter 600 (measured
	                  live, not the 4.0 this file used to quote), and
	                  the phone panel's inner measure is vw less the
	                  10px inset, the 20px gutter and the 23px indent,
	                  both sides. The old 5em CLIPPED it: at 390px the
	                  word ran to x=400 against a panel edge at 380,
	                  and at 320px to 330 against 310. The panel is
	                  overflow:hidden, so nothing reported an overflow
	                  and the page quietly lost the last glyphs —
	                  measure a RANGE over the line, never the box.
	                  Derived from the phone gutters because those are
	                  the tight ones; on a desktop it is never the
	                  smallest term.

	     17svh        the HEIGHT cap, and the answer to the first
	                  complaint. Three lines at 0.9 leading stand
	                  2.7x the font size, so 17svh holds the block at
	                  about 46% of the panel at every height. At
	                  1440x620 the old em ladder gave 159.6px and the
	                  title filled 72.3% of the panel, starting 117px
	                  under the navbar. It was 20svh for a few hours
	                  and came down with the vw term.

	   Floor 30px and cap 200px are guards; neither binds at any
	   real viewport now.
	   ───────────────────────────────────────────────────────────── */
	/* ─────────────────────────────────────────────────────────────
	   THE NO-CLIP CEILING CAME DOWN WITH THE CASE, 23vw -> 19vw
	   (2026-11-12, operator: "section 1 title is upper case").

	   That term is derived from the width of the longest word, and
	   "RESEARCH." is 20.8% wider than "Research." — measured live on
	   this page, 4.992x the font size against 4.131x. Leaving the
	   ceiling where it was would hand a narrow, tall phone a size the
	   panel cannot hold, and the panel is overflow:hidden, so NOTHING
	   WOULD REPORT IT: no overflow, no console, scrollWidth ===
	   clientWidth, and the last glyphs simply gone. That is exactly
	   how the old 5em shipped clipped on every phone.

	   19vw - 20px is the old term scaled by 1/1.208 and rounded in the
	   safe direction. MEASURE A RANGE OVER THE LINE, never the
	   element box, if the case or the headline changes again. */
	font-size: clamp(30px, min(calc(5.5vw + 5svh), calc(19vw - 20px), 17svh), 200px);
	line-height: 0.9em;
	letter-spacing: -.075em;

	/* UPPERCASE — operator, 2026-11-12, with section 2 in the same
	   breath. It reverses the 2026-11-10 "titlecase not uppercase"
	   ruling; category-grid.css carries the measured width ratio, and
	   the two files stay in step as the note at the top of this rule
	   says. A transform, not re-cased markup, so putting it back is
	   one declaration and no database write. */
	text-transform: uppercase;
}

/* THE SIZE LADDER IS GONE — the clamp above replaces all four of its steps
   and the three phone blocks further down.

   It also means this title no longer computes identically to .sfcg__title
   in category-grid.css, which still steps 30/48/60/72/112 off sohub's own
   table. The two were deliberately matched on 2026-11-10 ("same font style
   and size font like section 2"), so this is a real departure — the hero is
   the only one of the pair that has to fit inside a 100svh panel, and that
   is what forced them apart. Section 2's own type is untouched. */

.sfop__line{
	display: block;
	will-change: transform, opacity;
}

/* ----------------------------------------------------------------------------
   CLEAR THE SOCIAL RAIL ON PHONES.

   .sfsr is a position:fixed column of four buttons pinned to the left edge on
   every page, z-index 90, --sfsr-size 46px wide (40px below 782px).

   It first collided when the title sat at 40svh on phones: measured at
   375x812 the rail occupied [0, 323, 40, 166] and the title
   [12, 325, 351, 97], so it covered the first letter of all three lines and
   RIDE. read as IDE.

   THAT PARTICULAR OVERLAP IS GONE — the title is bottom-aligned now, and at
   375x812 the rail ends at 489 while the title starts at 699, 210px clear.
   The indent stays anyway, because of this, at the foot of social-rail.css:

       @media (max-height: 420px){ .sfsr{ top:auto; bottom:12px; } }

   Under 420px tall the rail stops being vertically centred and anchors to
   the BOTTOM left — which is exactly where this title now lives. Measured at
   812x375: rail [0, 163, 46, 190] and the title's text running from y=193,
   straight through it. Horizontally clear only because the text starts at
   x=56.

   So bottom-aligning did not remove the need for this, it moved which
   viewports need it — from tall phones to short ones.

   WHICH IS WHY THE INDENT IS NOW SCOPED BY HEIGHT, not applied to every
   phone. Applied at all widths it pushed the title a visible 46px off the
   left gutter on ordinary portrait phones that no longer need it at all —
   the operator saw it immediately and asked why the title sat so far in.

   The clearance is only required when the title and the rail can occupy the
   same band:

     - tall portrait (600px+): rail centred, bottom-aligned title well below
       it. 375x812 measured 210px of clearance. No indent.
     - short / landscape (<=560px tall): the rail is either bottom-anchored
       by its own max-height:420px rule or centred low enough to meet the
       title. Indent.
     - 768px and wider: the 16px inset plus the 40px gutter already put the
       text at x=56, past the rail's 46px. Nothing needed, which is why this
       query stops at 767.

   560px is the threshold rather than 420 because the rail is centred until
   420 and a bottom-aligned title reaches up to meet it before then; the
   margin covers the taller headline this ramp now produces.

   Indenting only the title, not the whole panel: nothing else in here sits
   at the left edge. The rail belongs to another section and is not touched.
   ---------------------------------------------------------------------------- */


/* INDENT ONLY NOW. The three font-size steps that were here are in the
   clamp above — and the shortest of them never applied anyway: this block
   and the plain (max-width: 767px) one that followed it carried the same
   specificity, so the later one won and a 560px-tall phone was handed the
   LARGER size. A size that depends on height belongs in one expression,
   not in two blocks racing on source order. */
@media (max-width: 767px){
	.sfop__title{
		padding-left: 23px;
	}
}

/* The third line carries the brand colour, exactly as sfd__word--accent does
   one band down. accent-1, never a hex: the theme ships #ec1e79 and the live
   theme wins. */
.sfop__line--accent{
	/* ─────────────────────────────────────────────────────────────
	   THE THIRD LINE CARRIES SECTION 2's SHEEN, NOT A FLAT MAGENTA
	   (operator, 2026-11-13: "for text RESEARCH i want you make it
	   effect text like you make it in section 2 for text
	   PERFORMANCE"). Ink -> magenta -> ink, clipped to the glyphs and
	   looping for ever, so it reads as a light passing through the
	   word rather than as a second colour.

	   IT IS `.sfcg__word` VALUE FOR VALUE — the stops, the 220%
	   size, the 5s ease loop and the keyframe positions are copied
	   from category-grid.css. SO THIS IS A FOURTH PLACE CARRYING
	   THOSE LITERALS rather than one shared token, for exactly the
	   reason the type note above gives: a custom property substitutes
	   at computed-value time on the element where it is DECLARED, so
	   a :root token declared in one stylesheet and read from another
	   resolves by load order and fails silently. The others are
	   category-grid.css (.sfcg__word), video-panel.css (the marquee
	   row and the statement accents) and bike-ribbon.css
	   (.sfbr__fat--snd). CHANGE THEM TOGETHER OR NOT AT ALL.

	   `width: fit-content` IS LOAD-BEARING, and it is the one thing
	   section 2 did not need. .sfop__line is display:block, so the
	   background's positioning area is the whole title block — the
	   word sits at the left of it and the sweep is spread across the
	   panel's full measure instead of across the word. fit-content
	   shrink-wraps the box to the text and changes nothing else; an
	   inline-block would also shrink-wrap and would re-align the line
	   on the strut's baseline, which can move the line box.

	   The negative-margin/padding pair gives the last glyph the room
	   the background box would otherwise clip, as .sfcg__word's does.

	   BOTH background-clip LONGHANDS. With only one of them Blink
	   paints a solid magenta band with invisible type inside it —
	   the failure reviews.css and category-grid.css both record.
	   ───────────────────────────────────────────────────────────── */
	width: fit-content;
	margin-right: -.12em;
	padding-right: .12em;

	color: transparent;
	background-image: linear-gradient(90deg,
		#fff                                          0%,
		var(--wp--preset--color--accent-1, #ec1e79)  48%,
		#fff                                        100%);
	background-size: 220% 220%;
	-webkit-background-clip: text;
	background-clip: text;
	animation: sfopSheen 5s ease infinite;
}

@keyframes sfopSheen{
	0%   { background-position: 10% 0 }
	50%  { background-position: 91% 100% }
	100% { background-position: 10% 0 }
}

/* ---------------------------------------------------------------------------
   4 · the button
   ----------------------------------------------------------------------------
   The reference puts a glass "latest news" card here, linking to a PDF. The
   operator asked for a single SHOP NOW instead (2026-11-10), so this is the
   brand magenta rather than the card's bg-black/10 + 18px backdrop blur.

   accent-1, never a hex: the theme ships #ec1e79 and the live theme wins.
   --------------------------------------------------------------------------- */

.sfop__cta{
	position: relative;
	z-index: 1;
	flex: none;

	display: inline-flex;
	align-items: center;
	justify-content: center;
	gap: 14px;

	padding: 19px 28px;
	border-radius: 4px;
	background: var(--wp--preset--color--accent-1, #ec1e79);
	color: #fff;
	text-decoration: none;

	font-family: 'Chakra Petch', system-ui, sans-serif;
	font-weight: 700;
	font-size: 13px;
	letter-spacing: .16em;
	text-transform: uppercase;
	line-height: 1;
	white-space: nowrap;

	will-change: transform, opacity;
	transition:
		transform .4s cubic-bezier(.19,1,.22,1),
		box-shadow .4s cubic-bezier(.19,1,.22,1),
		filter .3s ease;
}

/* On mobile it takes the width the reference's card takes — the full
   measure, pinned to the bottom. */
.sfop__cta{ width: 100%; }

.sfop__cta:hover,
.sfop__cta:focus-visible{
	transform: translateY(-2px);
	filter: brightness(1.08);
	box-shadow: 0 10px 30px rgba(236,30,121,.32);
}

.sfop__cta:focus-visible{
	outline: 2px solid #fff;
	outline-offset: 3px;
}

.sfop__cta-arrow{
	display: block;
	width: 14px;
	height: 14px;
	flex: none;
	transition: transform .4s cubic-bezier(.19,1,.22,1);
}
.sfop__cta:hover .sfop__cta-arrow,
.sfop__cta:focus-visible .sfop__cta-arrow{
	transform: translateX(4px);
}

/* ---------------------------------------------------------------------------
   5 · desktop
   --------------------------------------------------------------------------- */

@media (min-width: 768px){
	.sfop{
		/* The gutter went 8 -> 16 -> 12 across 2026-11-10. 16 was chosen so
		   more of the carbon weave showed and the panel read as a plate laid
		   on the surface rather than a hairline seam; the operator then asked
		   for less padding, and 12 keeps that reading while taking a quarter
		   off. It is still double the phones' 6.

		   THE RADIUS NO LONGER STEPS UP HERE. It was 20px, the theme's
		   --radius-2xl, on the reasoning that the corner should not scale
		   with the inset - growing both together starts to look like a
		   phone screenshot. That still holds; what changed is the target.

		   IT IS 12px AT EVERY WIDTH NOW, MATCHING THE NAVBAR (2026-11-10,
		   operator: "please make it standadize .. navbar and hero video
		   use 12px of the border radius"). The bar sits 24px above this
		   panel at the top of the page, close enough that 10 against 20
		   read as a mistake rather than as two decisions. The other half
		   is --sf-navbar-bar-radius in assets/css/navbar.css - CHANGE
		   BOTH OR NEITHER.

		   Not a shared token on purpose: this file only loads on the
		   front page and navbar.css loads everywhere, so a :root variable
		   declared in one and read by the other would depend on load
		   order. Two literals and this note. */
		--sfop-inset: 12px;
		--sfop-gutter: 40px;
		--sfop-foot: 48px;
	}

	.sfop__panel{
		flex-direction: row;
		align-items: flex-end;       /* bottom-aligned, not centred */
		justify-content: space-between;
		gap: 40px;
		padding: 0 var(--sfop-gutter) var(--sfop-foot);
	}

	.sfop__title{
		flex: 0 1 auto;
		min-width: 0;                /* or a long line pushes the button out */
	}

	.sfop__cta{ width: auto; }
}

/* There used to be a 768–1000px rule here that dropped the button under the
   title, because at the reference's own type ramp "Research." and the button
   met at those widths. The title came down ~30% on 2026-11-10 and they no
   longer do: measured at 768 the line is ~311px and the button 168px in a
   672px measure, and at 900 it is 356 in 780. The rule is gone rather than
   left in as dead weight — if the title is ever scaled back up, this is the
   thing that has to come back with it. */

/* ---------------------------------------------------------------------------
   6 · the intro's resting state
   ----------------------------------------------------------------------------
   .is-armed is added by an inline script in the markup, during parsing and
   therefore before first paint, so there is no flash of the finished hero
   before GSAP takes over. It is NEVER added when JavaScript is off, and the
   same inline script drops it on a timer if home-opening.js never arrives —
   an invisible hero is the one failure this section cannot have.

   The reference gates its whole intro on document.fonts.ready with the hero
   at opacity 0 behind it. Measured in a throttled tab, theirs sat invisible
   at opacity 0 three seconds after load. We do not copy that: without JS,
   this hero renders.
   --------------------------------------------------------------------------- */

/* THE MEDIA IS NOT IN THIS LIST, and that is the whole of the 2026-11-10
   "auto display the video" change on this side. The video is at full size and
   full opacity from the first frame; only the title and the button wait.
   Putting .sfop__media back here without also restoring its tweens in
   home-opening.js hides the video with nothing left to reveal it. */
.sfop.is-armed .sfop__line,
.sfop.is-armed .sfop__cta{
	opacity: 0;
}

@media (prefers-reduced-motion: reduce){
	/* The intro does not run at all; make sure nothing is left hidden if the
	   preference changes after the class was added. */
	.sfop.is-armed .sfop__media,
	.sfop.is-armed .sfop__line,
	.sfop.is-armed .sfop__cta{
		opacity: 1;
	}
	.sfop__cta{ transition: none; }
	.sfop__line--accent{ animation: none; }
}
