fix(render): size a group's own tile and a picture shape's own tile in points too

Review of #1693 found two more places that drew a tiled picture fill in the
shape's own (child-space) units inside a scaled group.

A group nested in a scaled group stored its own tiled fill unconverted, so the
members that take it (grpFill) got oversized tiles. The owning group now
converts the fill with per_point_xy when it stores it, so inheriting members,
directly or through a chain of grpFill groups, get the converted tile.

A picture shape (p:pic) whose blip fill is a tile passed its own fill straight
to paint::picture. It now goes through paint::tiled_picture_per_point, which
tiled_per_point also uses.

Tests: a group's tile passed to its members and a picture shape's own tile, each
in outer groups scaled 1, 0.25, 1600 and 2x0.5; both fail without the change.

Docs: the Picture and texture fill checklist row is re-assessed (remains P) with
its status line and revision history, and the ROADMAP progress log row mentions
both cases.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

ROADMAP.md

@@ -148,7 +148,7 @@ app's JSON control channel (`deckcraft --control PORT`).
 
 | Date | What landed |
 |---|---|
-| 2026-11-02 | Tiled picture fills in groups: a tile (`a:blipFill` with `a:tile`) keeps its size, offset and anchor in slide points whatever the group's child-coordinate scale, uniform or stretched unevenly, nested or flipped, and tiles with `rotWithShape="0"` keep their size in points at any zoom, as PowerPoint draws them (measured on its own PNG export of a synthetic probe deck: every grouped case within ~3/255 of the ungrouped control's own difference from PowerPoint, which is ~13/255 from resampling). Before, a group whose child space was ~1800 times smaller than its box showed one flat texel of the texture (a Venn diagram with paper-textured circles and a hatched intersection) |
+| 2026-11-02 | Tiled picture fills in groups: a tile (`a:blipFill` with `a:tile`) keeps its size, offset and anchor in slide points whatever the group's child-coordinate scale, uniform or stretched unevenly, nested or flipped (also a group's own tile fill passed to its members, and a picture shape's own tile), and tiles with `rotWithShape="0"` keep their size in points at any zoom, as PowerPoint draws them (measured on its own PNG export of a synthetic probe deck: every grouped case within ~3/255 of the ungrouped control's own difference from PowerPoint, which is ~13/255 from resampling). Before, a group whose child space was ~1800 times smaller than its box showed one flat texel of the texture (a Venn diagram with paper-textured circles and a hatched intersection) |
 | 2026-10-31 | No Fill shape interiors allow selection and editing of underlying objects, even when a quick style remains; hit-testing uses the renderer’s fill resolution. Outlines, rotation, text targets and actual fills keep their hit behaviour (#1621) |
 | 2026-10-30 | Print-preview tests use the URL emitted by egui and its exact file path; previews are explicitly removed before the macOS process exits |
 | 2026-10-30 | `slide.actions` lists shape click and hover actions and text-run hyperlinks (including table cells and grouped shapes) for audit; listing never runs an action. The UI pane is a follow-up |

crates/render/src/paint.rs

@@ -274,13 +274,21 @@ pub fn picture(ctx: &mut RenderContext, pres: &Presentation, pf: &PictureFill, b
 /// returned as it is.
 pub fn tiled_per_point(fill: &Fill, per_point: (f64, f64)) -> std::borrow::Cow<'_, Fill> {
     let Fill::Picture(pf) = fill else { return std::borrow::Cow::Borrowed(fill) };
-    let PictureMode::Tile { tx, ty, sx, sy, flip, align } = &pf.mode else { return std::borrow::Cow::Borrowed(fill) };
+    match tiled_picture_per_point(pf, per_point) {
+        std::borrow::Cow::Borrowed(_) => std::borrow::Cow::Borrowed(fill),
+        std::borrow::Cow::Owned(p) => std::borrow::Cow::Owned(Fill::Picture(p)),
+    }
+}
+
+/// [`tiled_per_point`] for a picture fill itself (a picture shape's own fill is not a [`Fill`]).
+pub fn tiled_picture_per_point(pf: &PictureFill, per_point: (f64, f64)) -> std::borrow::Cow<'_, PictureFill> {
+    let PictureMode::Tile { tx, ty, sx, sy, flip, align } = &pf.mode else { return std::borrow::Cow::Borrowed(pf) };
     let (kx, ky) = per_point;
     if !(kx.is_finite() && ky.is_finite() && kx > 1e-12 && ky > 1e-12) || ((kx - 1.0).abs() < 1e-9 && (ky - 1.0).abs() < 1e-9) {
-        return std::borrow::Cow::Borrowed(fill);
+        return std::borrow::Cow::Borrowed(pf);
     }
     let mode = PictureMode::Tile { tx: tx * kx, ty: ty * ky, sx: sx * kx, sy: sy * ky, flip: flip.clone(), align: align.clone() };
-    std::borrow::Cow::Owned(Fill::Picture(PictureFill { mode, ..pf.clone() }))
+    std::borrow::Cow::Owned(PictureFill { mode, ..pf.clone() })
 }
 
 /// Where the first tile of a tiled picture fill goes in `bounds`: its corner, edge middle or

crates/render/src/renderer.rs

@@ -436,7 +436,8 @@ impl Renderer {
                 // A group's own fill (a group filled with its parent's passes that on).
                 let own = match &s.fill {
                     Some(Fill::Group) | None => self.group_fill.clone(),
-                    Some(fl) => Some((fl.clone(), m, Rect::new(0.0, 0.0, x.w, x.h))),
+                    // A tile is given in points; this group's own space may be scaled by its parents.
+                    Some(fl) => Some((paint::tiled_per_point(fl, self.per_point_xy(m, depth)).into_owned(), m, Rect::new(0.0, 0.0, x.w, x.h))),
                 };
                 let outer = std::mem::replace(&mut self.group_fill, own);
                 // A group's outer shadow is cast by everything in it, as PowerPoint draws it: the
@@ -763,7 +764,8 @@ impl Renderer {
                 let bounds = Rect::new(0.0, 0.0, w, h);
                 ctx.push_clip_layer(outline);
                 let t = f.opts.gif_times.iter().find(|g| g.0 == s.id).map(|g| g.1);
-                paint::picture(ctx, f.pres, &paint::resolved(rctx, pf, None), bounds, m, t);
+                let pf = paint::tiled_picture_per_point(pf, pt_xy);
+                paint::picture(ctx, f.pres, &paint::resolved(rctx, &pf, None), bounds, m, t);
                 ctx.pop_layer();
             }
             // An object DeckCraft doesn't model (an embedded OLE object, an unknown graphic): its

crates/render/src/tests_wild.rs

@@ -506,6 +506,68 @@ fn a_tile_keeps_its_size_in_a_group_with_scaled_child_coordinates() {
     }
 }
 
+/// A slide with `shape` (built for a box of `w` x `h` child units) in a group at (100, 100) that is
+/// 240 x 60 pt on the slide and scaled `kx` x `ky` against its child space, drawn at 1 px per point.
+fn in_a_scaled_group(kx: f64, ky: f64, p: Presentation, shape: impl FnOnce(f64, f64) -> Shape) -> Image {
+    let (w, h) = (240.0 / kx, 60.0 / ky);
+    let group = Shape {
+        id: ShapeId(21),
+        xfrm: Some(Xfrm::new(100.0, 100.0, 240.0, 60.0)),
+        kind: ShapeKind::Group { children: vec![shape(w, h)], child: Xfrm::new(0.0, 0.0, w, h) },
+        ..Default::default()
+    };
+    let mut p = p;
+    std::sync::Arc::make_mut(&mut p.slides[0]).shapes = vec![group];
+    render_slide(&p, 0, &RenderOpts { scale: 1.0, ..Default::default() })
+}
+
+#[test]
+fn a_tile_a_group_passes_to_its_members_keeps_its_size_inside_a_scaled_group() {
+    use deckcraft_model::style::{PictureFill, PictureMode};
+    // The group that owns the tiled fill sits inside another group whose child space is scaled: the
+    // members that take it (`grpFill`) get tiles of 60 x 30 pt, as with no outer group.
+    let blue = Rgba::rgb(0, 0, 255);
+    for (kx, ky) in [(1.0, 1.0), (0.25, 0.25), (1600.0, 1600.0), (2.0, 0.5)] {
+        let mut p = Presentation::default();
+        let media = p.add_media("t.png", "image/png", half_red_half_blue()).expect("media");
+        let mode = PictureMode::Tile { tx: 0.0, ty: 0.0, sx: 10.0, sy: 10.0, flip: "none".into(), align: "tl".into() };
+        let img = in_a_scaled_group(kx, ky, p, |w, h| {
+            let member =
+                Shape { id: ShapeId(23), xfrm: Some(Xfrm::new(0.0, 0.0, w, h)), fill: Some(Fill::Group), line: no_line(), ..Default::default() };
+            Shape {
+                id: ShapeId(22),
+                xfrm: Some(Xfrm::new(0.0, 0.0, w, h)),
+                fill: Some(Fill::Picture(PictureFill { media, mode, ..Default::default() })),
+                kind: ShapeKind::Group { children: vec![member], child: Xfrm::new(0.0, 0.0, w, h) },
+                ..Default::default()
+            }
+        });
+        let (a, b, c) = (img.pixel(115, 115), img.pixel(145, 115), img.pixel(175, 115));
+        assert!(close(a, RED, 30) && close(b, blue, 30) && close(c, RED, 30), "outer group {kx} x {ky}: {a:?} {b:?} {c:?}");
+    }
+}
+
+#[test]
+fn a_picture_shapes_own_tile_keeps_its_size_inside_a_scaled_group() {
+    use deckcraft_model::style::{PictureFill, PictureMode};
+    // A picture (`p:pic`) whose blip fill is a tile rather than a stretch.
+    let blue = Rgba::rgb(0, 0, 255);
+    for (kx, ky) in [(1.0, 1.0), (0.25, 0.25), (1600.0, 1600.0), (2.0, 0.5)] {
+        let mut p = Presentation::default();
+        let media = p.add_media("t.png", "image/png", half_red_half_blue()).expect("media");
+        let mode = PictureMode::Tile { tx: 0.0, ty: 0.0, sx: 10.0, sy: 10.0, flip: "none".into(), align: "tl".into() };
+        let img = in_a_scaled_group(kx, ky, p, |w, h| Shape {
+            id: ShapeId(22),
+            xfrm: Some(Xfrm::new(0.0, 0.0, w, h)),
+            kind: ShapeKind::Picture { fill: PictureFill { media, mode, ..Default::default() } },
+            line: no_line(),
+            ..Default::default()
+        });
+        let (a, b, c) = (img.pixel(115, 115), img.pixel(145, 115), img.pixel(175, 115));
+        assert!(close(a, RED, 30) && close(b, blue, 30) && close(c, RED, 30), "group {kx} x {ky}: {a:?} {b:?} {c:?}");
+    }
+}
+
 #[test]
 fn an_upright_tile_in_a_turned_shape_keeps_its_size_in_points_at_any_zoom() {
     use deckcraft_model::style::{PictureFill, PictureMode};

docs/parity-checklist.md

@@ -2,7 +2,9 @@
 
 > **Generated:** 2026-10-14 (bullets/numbering remain D after empty-line marker review; summary by `cargo xtask parity`; rows re-scored by the 2026-10-14 community PR wave, duplicate Thesaurus, AutoCorrect, Artistic effects and Corrections rows folded; Preview in editor re-scored 2026-10-13; earlier rows re-scored 2026-10-10 against PowerPoint for Mac 16.113.4; video/animated GIF row re-scored 2026-10-11, #171) · **Target:** Microsoft PowerPoint for Mac 16.113 (Microsoft 365)
 
-> **Last updated:** 2026-11-01 · Locked-shape and no-outline regressions added for #1621; fill row re-assessed: remains D; No Fill hit-testing now follows resolved appearance. Scores unchanged.
+> **Last updated:** 2026-11-02 · Tiled picture fills keep their size in scaled groups and upright tiles (#1693); Picture and texture fill row re-assessed: remains P, since only how tiles are sized changed. Scores unchanged.
+>
+> **Previously (2026-11-01):** Locked-shape and no-outline regressions added for #1621; fill row re-assessed: remains D; No Fill hit-testing now follows resolved appearance. Scores unchanged.
 
 The summary between `<!-- SUMMARY -->
 **Weighted breadth parity: 84%** over 191 features.
@@ -240,4 +242,5 @@ Open P0 items: Contextual tabs (Shape Format, Picture Format, Table Design/Layou
 
 | Date | Change | Summary |
 |---|---|---|
+| 2026-11-02 | trivial | Tiled picture fills keep their size in scaled groups and upright tiles (#1693); Picture and texture fill row remains P and summary scores are unchanged |
 | 2026-10-31 | trivial | No Fill interior hit-testing follows resolved fill (#1621); fill row remains D and summary scores are unchanged |