WNS B1 + B2 v3L: the delivery set (Ty's pick), version 7

VCC, Oct 29 2026. Order: CoS e-35d3cc19 (Ty: "I like the v3L"); version 7 = CoS e-186841d6 (B2 v3L-r: F-4 + F-6, picture only), resume e-bce59c28.

Draft for review. Nothing is published; ads stay off until the Editor passes and Ty approves. Approval is tied to the md5s below (YT-A03, SV-A04).

New in version 7 (CoS e-186841d6, B2 v3L-r; picture only): F-4 (DR-03, result first): B1's 3 real team-cut Shorts ("Example") now sit under Ty's chin for 0-1.17 s, while he asks "What do you need from me?". The label "For business owners with videos" stays, so frame 1 = his face + the result + who it's for. F-6 (SV-E08): "You don't give us" pops as one phrase at 5.30 s (no ~7-frame "You don't"). B2 has new md5s: 9:16 c22c2dd1, -fb 8b7cfb13, 4:5 0da94824 (were 66ed37c7 / b1afc68a / 85108803). Proof it is picture only: each file carries the delivered file's audio stream, copied: packets 44d930e4 / 352f7d5d / 44d930e4 = delivered; decoded PCM on 0-1.35 s and 5.30-5.53 s identical; audio start 0.000, 992 / 912 / 992 packets. Frame diff vs the delivered files: only frames 0-35 (the hook) and 159-172 (5.30-5.77 s, the caption) moved; every other frame is ≥ 42.3 dB (two encodes of the same picture). 9:16 framing: the crop already sits at his take's top edge, so he is pushed in (FULLBODY x 1.30-1.32) until his hair clears the question (22.9 px); the fan runs from under his chin to the frame bottom (front card 320 px wide) and "Example" sits at the fan's top-left. The 4:5 uses FULLBODY-1.03 with the fan as built. B1 is unchanged from version 6. Byte-exact B2 files for the CoS: artifact art_01m63y2z9r0ak1111yfthasgpc.

Version 6 (CoS e-b5d56f21, honesty fix): B1's hook label now reads "1 video → up to 3 Shorts" (was "1 video → 3 Shorts"). The offer is "up to three Shorts, depending on usable footage", so the label keeps the qualifier (SV-E08, WUI-A04). One line at 60 px (was 68 px), inside x 180-900. Only the label frames moved; the audio packets are identical to version 5. B1 has new md5s (table below); B2 was unchanged from version 5. Proof it is picture only: the audio packet md5s are identical across Ty's R3 files (b5beaa6d / 400fd82c / a3194cc6), the 04:00Z files (2247ff5a / 89498e49 / dec07c7f) and the new files: 9:16 b39ac632, -fb 82a49095, 4:5 b39ac632. The frame diff against the 04:00Z files shows only frames 0-58 moved (the label on the fan, 0-1.97 s); every other frame is ≥ 42.5 dB (9:16) / ≥ 42.6 dB (4:5), which is two encodes of the same picture.

Version 5 (two fixes I found after the 03:25Z hand-off; picture only, audio packets identical to the files Ty watched): (1) B1: the "1 video → 3 Shorts" label re-popped for 2 frames at 0.58 s, where the second hook beat starts. It now pops once, from frame 1 (only frames 18-20 moved; every other frame matches the watched file to ≥ 42.4 dB). (2) B2: his lids drop on frames 237-243 (7.90-8.10 s, silent), right before the 8.13 cut. His eyes-open frame 236 holds over them (the Editor's B3-5 method; only those frames moved). The label sits inside the 4:5 safe box (text x 180-900, y 40-1310) on every frame. The v2 | v3 | v3L comparison stays on version 3.

Carried over from v3, unchanged: his words, takes and order are v2's (SV-E02). Kit text is not shown under the team lines (free-tool drafts; e7fb125). Caption style A, the patch16 fold, the Pixabay 609167 bed, no AI anywhere (AI info OFF, altered/synthetic = No).

Open for the Editor gate: B1-2 (card rows → him on camera), the two hook labels, and "No channel access" / "3+ minutes". R3 is locked (Ty: "sound is good in B1").

Closed: the lifter release (Shorts 2/3). Ty confirmed it on Oct 29 2026, paid ads included ("lifter said we can use her. permission approved", via the CoS). It is in the rights ledger for B1-B4 (SV-E07). B1, B2 and B3 show her, in all 3 files (9:16, -fb, 4x5). B4 does not show her.

Players: review copies (Helix 25 MiB cap): picture re-encoded once at 5.5 Mb/s (PSNR 49.3-50.5 dB vs the delivery file, measured), sound = the delivery audio bit for bit (packet md5s match). The exact files are in the VCC repo, jobs/wns/2026-10-27-round2/deliverables/v3L/.

B1 "Three Shorts our team cut from one video"

Label: 1 video → up to 3 Shorts (on the fan of his 3 real Shorts; CoS e-b5d56f21)

9:16 master (= the -fb picture)

4:5 Meta feed

Hook 0-3 s

One frame per second

Files

Delivery file	Use	md5 (approval is tied to this)
wns-b1ty-real-example-v3L.mp4	9:16 master (YouTube Short, AAC 48k)	`71d121f6550c9c12f21949168c0cf633`
wns-b1ty-real-example-v3L-fb.mp4	9:16 Meta Reels/Stories (AAC 44.1k; picture bit-identical to the master)	`502fdf55ad47c02849e0b603dba890fd`
wns-b1ty-real-example-v3L-4x5.mp4	4:5 Meta feed (1080x1350, re-laid out)	`1b62f58b52a1efd2666e4c73e627744d`
wns-b1ty-real-example-v3L.captions.srt	captions (transcript record; burned in, not uploaded)	`8020c846ea8ea6f8fed6c84198ebeaac`
wns-b1ty-real-example-v3L.dialogue-only.wav	dialogue-only master (YT-B09)	`174ed0ad05ab9b17f5695936d862ae7a`

edit map (EDL) · captions SRT

Gates 9:16, every frame (806): text outside the safe box 0, pictures outside SV-B05 0, type on a face 0 (820 faces), caption vs head min 17.0 px / 0 overlaps, smallest text 36 px, blinks flagged 0 of 553 -> PASS
Gates 4:5, every frame (806): text outside the safe box 0, pictures outside SV-B05 0, type on a face 0 (862 faces), caption vs head min 10.2 px / 0 overlaps, smallest text 36 px, blinks flagged 0 of 553 -> PASS
Probe + loudness (3 MP4s): H.264 High, yuv420p, BT.709 tv, SAR 1:1, 30 fps CFR, fast-start, no black frames; -15.0 LUFS / -1.8 dBTP
Words: whisper on the final = his read, word for word: True
Shared QC: handbook_qc (speech gate) + mix_check: exit codes 0/0/0/0/0; FAIL lines 0

B2 "What do you need from me?"

Label: For business owners with videos (CoS wording; site: "for business owners who already record videos")