Preserve source cadence and sample picture fps after analysis

docs/animation-model.md

@@ -333,11 +333,15 @@ Every node may map the frame it is evaluated at:
 ```clojure
 :time {:mode :inherit}                        ; the default, and almost always right
 :time {:mode :map :expose 2 :offset -1 :rate 1.0 :loop? false}
+:time {:mode :map :source-fps 30 :sample-fps 12} ; root: lower picture cadence
 ```
 
 Three features that look unrelated are this one mechanism:
 
 - **exposure** is `⌊f/n⌋·n`,
+- **picture fps** quantises source time to a chosen picture grid, then reads the
+  latest source pose at or before that time; source analysis and audio keep their
+  original cadence,
 - **mouth lead** is `f + k`,
 - **a symbol instance's timing** is `(f - at)·rate + in`, with optional looping.
 
@@ -463,8 +467,11 @@ world transform of the node it rides:
 
 Composed with image-pixels-to-local — **both axes divided by `imgH`**, never by
 their own dimension — the photo is registered with the shapes by construction,
-and an unregistered underlay is merely decorative. Which frame it shows, the
-current one or the held plate frame, is a UI choice and not a stored one.
+and an unregistered underlay is merely decorative. The tracing editor chooses
+which source frame to show under a cel. That reference choice is independent of
+the finished picture fps and does not change the dense analysis track. A cel can
+therefore use any useful source frame as its drawing reference, even when that
+frame is not one of the displayed picture poses.
 
 A photo that has to sit *between* two drawn layers is the case that would make it
 a `:bitmap` node with an op of its own. Nothing wants that yet: a reference is
@@ -527,6 +534,7 @@ Proof that it covers what exists, not just what is wanted:
 | painted background cel | node per layer, `[:geom :pts]` **framed**, `[:style :color]` framed |
 | `mouth lead` | `:time {:offset k}` on performance nodes only |
 | `exposure` | `:time {:expose n}` on the clip root, inherited |
+| picture fps | `:time {:source-fps s :sample-fps p}` on the clip root, applied after analysis |
 | hand correction | an `:over` layer, `:offset` or `:replace` |
 
 The brow row is the one worth looking at twice. `docs/design.md` argues at length

docs/architecture.md

@@ -49,6 +49,11 @@ overrides and kept-frame sets are all in clip-frame space, so a clip slides on
 the timeline without a single stored number changing. Exposure and lead are
 transforms *within* clip space:
 
+Detection retains every source frame. A chosen picture fps samples the frozen
+roto in clip time; it changes neither source-frame count nor the audio clock.
+The set of source frames an artist uses as cel tracing references is another
+selection, independent of the picture fps.
+
 ```clojure
 (defn pose-frame [clip cf]
   (-> cf (expose (:exposure clip)) (shift (:lead clip) (count-frames clip))))
@@ -73,7 +78,7 @@ project
 │   ├── scene                node tree
 │   ├── channels             node+property -> keyframe stream, in cf
 │   ├── overrides            (node, property, cf) -> value, applied last
-│   └── cels                 painted vector layers, per exposure slot
+│   └── cels                 painted vector layers, keyed by chosen clip frames
 └── sequence[]               clip placements: {clip-id, at, in, out}
 ```
 
@@ -81,6 +86,10 @@ project
 it a name is most of the work: `state` in `app.js` is a clip with its analysis
 inlined and its palette global.
 
+Cel keys select where drawings begin and how long they hold. The source frames
+shown beneath a cel while tracing are chosen independently, and picture fps
+only controls which analyzed pose the finished roto displays at a given time.
+
 ### Two things called "track"
 
 `docs/design.md` says "dense track" for the landmark stream. A timeline also

docs/port-plan.md

@@ -4,8 +4,10 @@ Self-contained. You should not need any prior conversation to execute this.
 
 **Implementation status (2026-09-27):** steps 0–6 are in the CLJS frontend.
 Step 6 reads extracted footage from the manifest, detects landmarks with local
-MediaPipe assets, and runs the same freeze path as the synthetic take. Step 7 is
-next: eyes, brows and the pixel-derived mouth interior.
+MediaPipe assets at full source cadence, and runs the same freeze path as the
+synthetic take. The scene time map can sample the frozen roto at a lower picture
+fps without changing source analysis, duration or audio. Step 7 is next: eyes,
+brows and the pixel-derived mouth interior.
 
 ## What arthur is
 
@@ -302,6 +304,10 @@ MediaPipe interop behind one namespace; real frames, real audio, real fps from
 the manifest. **Vendor the wasm** rather than fetching from jsdelivr — it is
 currently the only thing in the tool that silently requires a network.
 
+Decode every source frame for analysis. A lower output picture fps is a time map
+over frozen channels, not a reduced detection track. Selecting source frames to
+trace into cels is independent again and remains outside this port's paint scope.
+
 **Done:** real footage plays back as a rotoscoped mouth.
 
 ### 7 — the rest of measure