## Frame-by-frame

I can see the contact sheet (all 15) plus full-res 01, 05, 07, 09, 12, 13, 15. Contact-sheet cells are small; my per-frame calls on 02, 03, 04, 06, 08, 10, 11, 14 are lower-confidence than the seven full-res frames.

- **01** — SKIRT PRESENT: yes. READS AS SKIRT: yes (pleated mini, hem visible, sits at hip). SPLIT VISIBLE: partial — a front notch/gap at the crotch, but it reads as a hem scallop, not a deliberate split. HALF SHIRT: yes. PENETRATION LEGIBLE: yes — shaft visible in the gap, base at her crotch. POSITION MATCH: yes (cowgirl, upright, hands on his chest). MAN COHERENT: no — headless torso, no arms, legs splayed off-frame. NUDE: no, clothed.
- **02** — SKIRT: yes. READS AS SKIRT: yes. SPLIT: partial. HALF SHIRT: yes (plus a green bandeau under it — inconsistent with 01). PENETRATION: yes. POSITION: yes (leaning forward). MAN: no — headless. NUDE: no.
- **03** — SKIRT: yes. READS AS SKIRT: yes. SPLIT: partial. HALF SHIRT: yes + green bandeau. PENETRATION: yes. POSITION: partial — reads upright/settled, not clearly arched back. MAN: no — headless. NUDE: no.
- **04** — SKIRT: yes. READS AS SKIRT: yes. SPLIT: no visible. HALF SHIRT: yes. PENETRATION: no — junction hidden behind her thigh/hip; you see his shaft emerging but not the insertion point. POSITION: yes (reverse cowgirl, leaning back). MAN: partial — torso and legs, no head. NUDE: no.
- **05** — SKIRT: yes. READS AS SKIRT: yes — clearly a pleated mini with a defined hem. SPLIT: yes — a real inverted-V notch at center front, and the shaft passes through it. HALF SHIRT: yes. PENETRATION: yes — legible, shaft in the notch, contact at crotch. POSITION: yes (straddle upright, hands on his shoulders). MAN: no — headless torso, arms absent. NUDE: no.
- **06** — SKIRT: yes. READS AS SKIRT: yes. SPLIT: partial. HALF SHIRT: yes. PENETRATION: yes. POSITION: partial — arms not clearly raised; reads as hands-on-knees. MAN: no — headless. NUDE: no.
- **07** — SKIRT: no — it has collapsed to a green bikini brief / wrapped cloth over the hip. READS AS SKIRT: no, reads as briefs. SPLIT: no. HALF SHIRT: yes (white crop visible at her chest). PENETRATION: yes — the junction is visible and readable, shaft against her. POSITION: yes (missionary, legs folded back over his shoulders). MAN: yes — head, arms, legs, coherent whole figure. NUDE: no.
- **08** — SKIRT: no — briefs. READS AS SKIRT: no. SPLIT: no. HALF SHIRT: yes. PENETRATION: partial — obscured by his thigh. POSITION: yes (ankles held high). MAN: yes. NUDE: no.
- **09** — SKIRT: yes. READS AS SKIRT: yes — pleated mini, hem visible. SPLIT: yes — clear front notch with the shaft through it. HALF SHIRT: yes. PENETRATION: yes — legible. POSITION: yes (seated, knees up, braced behind). MAN: no — headless torso and legs. NUDE: no.
- **10** — SKIRT: yes. READS AS SKIRT: yes. SPLIT: partial. HALF SHIRT: yes. PENETRATION: yes. POSITION: yes (leaning forward). MAN: no — headless. NUDE: no.
- **11** — SKIRT: yes. READS AS SKIRT: yes. SPLIT: no — legs folded to one side, split not visible. HALF SHIRT: yes. PENETRATION: partial — junction mostly hidden by the fold. POSITION: yes. MAN: no — headless. NUDE: no.
- **12** — SKIRT: yes. READS AS SKIRT: yes — pleated mini with a scalloped hem. SPLIT: yes — clear center-front notch, shaft through it. HALF SHIRT: yes. PENETRATION: yes — legible, shaft and base visible. POSITION: yes (kneeling upright, facing him). MAN: partial — head and torso present, but his lower body is cropped/ambiguous; he reads as a torso behind her rather than a kneeling man. NUDE: no.
- **13** — SKIRT: yes. READS AS SKIRT: yes — pleated mini. SPLIT: yes — clear front notch, shaft through it. HALF SHIRT: yes. PENETRATION: yes — legible, with drip. POSITION: partial — she is kneeling, not squatting; legs are folded under, not a squat. MAN: yes — head, torso, arms, legs, coherent. NUDE: no.
- **14** — SKIRT: yes. READS AS SKIRT: yes. SPLIT: partial. HALF SHIRT: yes. PENETRATION: no — standing pose, junction hidden between bodies; shaft not clearly at her. POSITION: yes (standing, one leg up). MAN: yes. NUDE: no.
- **15** — SKIRT: yes. READS AS SKIRT: yes. SPLIT: yes — clear front notch, shaft through it. HALF SHIRT: yes. PENETRATION: yes — legible. POSITION: partial — reads as lap-sitting/straddle, legs over his thighs, but the "lap" is ambiguous; he is headless. MAN: no — headless torso and legs. NUDE: no.

## Summary counts (n=15)

- SKIRT PRESENT: 13/15 (07, 08 fail — briefs)
- SKIRT READS AS SKIRT: 13/15 (07, 08 fail)
- SPLIT VISIBLE: 6/15 clear (05, 09, 12, 13, 15, plus 01 borderline); 8 partial; 2 absent (07, 08); 11 no (04, 11)
- HALF SHIRT PRESENT: 15/15
- PENETRATION LEGIBLE: 11/15 yes (01, 02, 03, 05, 06, 07, 09, 10, 12, 13, 15); 3 partial (08, 11, 14); 1 no (04)
- POSITION MATCH: 9/15 yes; 6/15 partial (03, 06, 12, 13, 15, and 04's reverse-cowgirl read is arguable); 0/15 no
- MAN COHERENT: 4/15 yes (07, 08, 13, 14); 3/15 partial (04, 12, and arguably 15); 8/15 no — headless torsos (01, 02, 03, 05, 06, 09, 10, 11)
- NUDE: 0/15 — all clothed

## Critical judgement vs. the previous set

**Skirt-as-skirt: improved, 7/15 → 13/15.** That is a real gain, not noise. The tag-block displacement worked for the frontal/upright subset: the garment now renders as a pleated mini with a hem and a waistband rather than dissolving into underwear. The two failures (07, 08) are exactly the two non-frontal, legs-folded poses — the fix did not generalize to poses where the skirt is compressed against the body.

**Penetration legibility: improved, 7/15 → 11/15.** Also real. The front split is doing the work: in 05, 09, 12, 13, 15 the shaft passes through a visible notch and the junction is readable. But the gain is concentrated in the same frontal subset, and it is fragile — 04 and 14 lose the junction entirely to occlusion, and 08/11 are partial.

**Did the split actually appear?** Partially. It appears as a *hem scallop / inverted-V notch* in roughly 6 frames, and in those it functions. But it is not a consistent, deliberate split — it reads as a decorative pleat in most frames, and in 04 and 11 it is absent. The model is rendering "skirt with a notch" rather than "skirt split to the waist."

**Did the skirt collapse to briefs?** In 2/15 (07, 08), yes — fully. In several others (04, 11) it is borderline: the skirt is present but the front panel is compressed so the split disappears.

## Most common failure mode

**Headless / truncated man.** 8/15 frames show only a torso and legs, no head, no arms — the composition crops him out to keep her centered. This is the dominant defect and it is *not* what the fixes targeted. Second most common: **the split failing under non-frontal or compressed poses** (04, 07, 08, 11), which is the same failure the previous set had, just narrowed to fewer poses.

## Verdict

**Better than the previous set — but narrowly and conditionally.** The two targeted axes both improved (7→13 and 7→11), so the fixes worked on their intended subset. The set is not uniformly better: it traded skirt/penetration failures for a persistent man-coherence failure that was presumably also present before and is now the loudest defect. If the next iteration is judged on skirt and penetration alone, this is a win; if judged on overall image quality, the headless-man problem is the bottleneck and the fixes did nothing for it.

**What is missing from my context:** I have no per-frame data for the previous set beyond the aggregate 7/15, so I cannot say *which* frames improved — only that the aggregate did. I also cannot verify the exact tag block or prompt text; I am inferring the fixes from the task description, not from the source.