diff --git a/docs/re/f6-unit7-one-moving-quad.md b/docs/re/f6-unit7-one-moving-quad.md
new file mode 100644
index 00000000..24f3221e
--- /dev/null
+++ b/docs/re/f6-unit7-one-moving-quad.md
@@ -0,0 +1,61 @@
+# F6 unit 7 — the title has exactly **ONE moving quad**: the lights are inside the texture
+
+**Status: ✅ measured ⟨capture⟩ ×1.** 2026-09-02.
+
+**Question:** how many distinct moving quads does the title draw, and how big?
+**Look at:** every quad, no size and no sample-count filter.
+**Not covered:** screenshots as images; what the lights look like.
+
+---
+
+The port raised a real caveat: my earlier scans filtered on height (>1.2 NDC) and
+on sample count (≥20), so a population of many **small** lights would have been
+invisible to both by construction. Dropping both filters:
+
+```
+title era, ALL quads          distinct shapes: 59
+                                      MOVING:   2
+                                      static:  57
+moving shapes:
+   1.38 x 3.15 NDC  (883 x 1134 px)   n=1140   dx 3.38
+   1.39 x 3.15 NDC  (890 x 1134 px)   n= 614   dx 3.37
+SMALL moving shapes (<0.5 NDC both axes):  0
+```
+
+> **Two moving keys, and they are the two rounding-split halves of one element —
+> `pteff03`. There is exactly one moving quad on the title, and not a single small
+> one.**
+
+## 🔴 That answers the human's actual uncertainty
+
+They said they could not tell whether the game renders **one light per line** or
+**one light covering several close lines**. It is the second:
+
+> **One quad, 883 × 1134 px, sweeping across. The multiple lights the human sees
+> are painted INTO its texture, not drawn as separate moving objects.**
+
+A port trying to reproduce "several lights" as several elements would be modelling
+the wrong thing — the game moves one large sprite whose image already contains the
+traces.
+
+📌 The port's caveat was right to raise and the check excludes the population it
+warned about. Third time in this exchange that pattern has held: **asking was
+worth it even when the specific worry turned out not to apply.**
+
+## ⚠️ What I have NOT done, and it was the literal ask
+
+The human asked for **screenshots** of the title. I have not taken any. I answered
+the counting question from the draw stream instead, because it is a stronger
+instrument for *how many and how big* — it reads the guest's own vertex buffer
+rather than pixels, and it cannot miss a faint light the way an image can.
+
+**But "what do the lights look like" is a question about an image, and the draw
+stream cannot answer it.** If that is what is wanted, it is a separate deliverable
+and it is still open.
+
+## Reach
+
+⟨capture⟩ ×1, one boot, the title era of one window, every quad in the stream.
+Static elements are not counted as lights here — 57 of them exist and some may be
+glows that pulse in place rather than travel; **this unit counts motion, not
+brightness.**