Short Drama Dataset Review

9:16 source screening · subtitle removal · GB200 production · 50-hour target

Updated August 31, 2026 · static review bundle
19.6 / 50 hours Confirmed accepted runtime
313completed deliveries
1,205accepted ReelShort items
1,003new raw Shorts identified
0production failures

Source Inventory

Only public portrait Shorts without persistent source branding are promoted.

Production and source expansion active

DramaBox

Original production source. Free/public inventory resolved and watermark-filtered.

302 complete
1,082raw
302accepted
6.2hruntime

GoodShort

High branding rate; conservative automatic acceptance retained a small clean subset.

11 complete
324raw
11accepted
2mruntime

ReelShort

Diskless source streaming into NFS, followed by cluster-native OCR and mask preparation.

1,205 accepted · cluster prep active
2,315raw
1,205accepted
13.4hruntime

ShortMax

Strong sample yield with public 1080x1920 Shorts and minimal persistent branding.

Full screening candidate
187raw
28/30sample accepted
37.5smean

NetShort

Strong sample yield with public portrait Shorts and low watermark rejection.

Full screening candidate
138raw
28/30sample accepted
43.7smean

DramaWave

Large inventory, but persistent corner branding materially lowers usable yield.

18/29 sample rejected
613raw
11/29sample accepted
84.2smean

MiniShort

Additional official short-drama inventory queued for full public, portrait, and watermark screening.

Full screening active
65raw
pendingaccepted
9:16format

Delivered Samples

Short excerpts for rapid inspection of final caption-removal delivery encoding.

H.264 CRF16 · audio retained
uB-5XR26to0

DramaBox production sample

DramaBox15sdelivered

Pipeline

Outro filtering, OCR mask generation, ProPainter temporal prior, DiffuEraser 2-step repair, native masked composite, and CRF16 delivery.

Format720x1280 · 30 fps · H.264 + AAC
StatusComplete with receipt
ReviewSelected excerpt only; full-corpus manual QA not run.
1QanjafOKeY

GoodShort production sample

GoodShort15sdelivered

Selection rule

Accepted only after public-access, portrait-format, persistent-logo, and branded-outro screening.

Format720x1280 · 30 fps · H.264 + AAC
StatusComplete with receipt
Source yield11 of 324 Shorts auto-approved.

Operations

Restartable state and durable storage locations.

Compute policyMaximum four concurrent 4xGB200 Jobs; 8-hour deadline per Job.
Local storageAuthenticated downloads stream directly to NFS; cluster performs normalization, OCR, masks, and generation.
Completed roots20260829-dramabox-scale-b001 · 20260830-dramabox-scale-b002 · 20260830-goodshort-auto
Active root/mnt/nfs/vlm-tzhou/h3-subtitle-repair/20260830-reelshort-auto
CleanupQP0 intermediates and completed Kubernetes Jobs are deleted; sources, masks, priors, deliveries, logs, and receipts remain.