73% of AI video cases don't say how they were made — and the reason isn't the creators
When we designed the tag schema we included one basic field: how was this video generated — text-to-video, image-to-video, reference-to-video, or first-and-last-frame?
It is the precondition for learning anything from a prompt. To study how a case was written, the first thing you need to know is whether it had a reference image or a specified first/last frame, because those three prompt styles are written completely differently. After filling in all 1,274 records, more than seventy percent of that field is empty.
Where the data comes from
The generation-mode field was determined from three sources: explicit statements on the original publication page, structural features of the prompt text itself (for example, whether it declares a first frame), and official labels on source sites. When none of the three yields an answer, we record "mode unverified" rather than guessing.
| Generation mode | Cases | Share |
|---|---|---|
| Mode unverified | 930 | 73.0% |
| Text-to-video | 164 | 12.9% |
| Reference-to-video | 128 | 10.0% |
| First/last-frame-to-video | 49 | 3.8% |
| Image-to-video (unsplit) | 3 | 0.2% |
Our first hypothesis was that creators are simply lazy about writing it down. Slicing by genre appeared to support that — the spread between highest and lowest is nearly 50 percentage points:
| Genre | Unverified rate | Sample |
|---|---|---|
| Action & physics | 94.3% | 70 |
| Transition editing | 86.0% | 43 |
| Short-drama dialogue | 79.0% | 224 |
| … | ||
| Sci-fi | 53.6% | 56 |
| Martial arts / wuxia | 45.1% | 71 |
It looks like different communities have different habits. But the real cause is something else.
Finding 1: The missing rate is a mirror of the source channel
We re-cut the data by provenance — where each case came from:
| Provenance | Sample | Unverified rate |
|---|---|---|
| official | 6 | 0.0% |
| source_page_public_prompt | 64 | 68.8% |
| creator-verbatim (copied from the original post) | 645 | 52.4% |
| beatapi_indexed_prompt_with_webm | 312 | 100.0% |
| indexed_from_awesome_minimax_h3_prompts | 74 | 100.0% |
| apimodels_original_or_indexed | 63 | 100.0% |
| official_or_official_index_prompt | 37 | 100.0% |
| Republished by static.minimax-h3.io | 20 | 100.0% |
The pattern is perfectly clean: every official primary channel has a 0% unverified rate; every third-party aggregation mirror has a 100% unverified rate.
That is a far stronger result than the genre cut. It says "mode unverified" is not a property of creators — it is information stripped out during propagation. The original post often says it. The aggregation site scrapes it, that line disappears, and every downstream indexer is permanently in the dark.
Finding 2: The only two complete sources aren't the community's main channels
Aligning hosting providers against unverified rate makes it starker:
| Video host | Sample | Unverified rate |
|---|---|---|
| pub-…r2.dev (official republishing channel) | 20 | 0.0% |
| h3-field-notes-production.up.railway.app | 648 | 52.2% |
| video.twimg.com (native X video) | 155 | 92.9% |
| media.beatapi.io | 312 | 100.0% |
| raw.githubusercontent.com | 60 | 100.0% |
| external-cdn.morphic.com | 44 | 100.0% |
| static.minimax-h3.io | 20 | 100.0% |
Meanwhile 87.4% of the whole corpus originates on X (Twitter) — which happens to be the venue least likely to carry structured metadata.
Finding 3: Among the 344 verified cases, short drama under-indexes
Restricting to the 344 cases with a verified mode, the genre mix differs from the whole:
| Genre | Verified cases |
|---|---|
| Product ad | 57 |
| Character performance | 51 |
| Short-drama dialogue | 47 |
| Martial arts | 39 |
| City & architecture | 34 |
Short-drama dialogue is the largest genre overall (224 cases, 17.6%) but ranks only third in the subset that can say how it was made (47 cases, 13.7%). The more a genre depends on specific reference material, the more likely its generation mode is lost in propagation.
What you can do with this
If you learn prompt craft from public cases: verify the generation mode first. If you cannot, don't copy the prompt. The same line — "the camera slowly pushes in" — is written very differently for text-to-video versus reference-to-video, and copying across layers wastes hours. Our advice: only study cases with a verified mode. In this library there are 344 of them.
If you build datasets or indexes: this is worth noting down. Scraping from aggregation sites systematically drops structured fields, and the loss is not random — the more thorough the aggregation, the cleaner the fields get. To preserve this class of information you have to capture it further upstream.
If you publish your own work: state the generation mode in the post body, even just "reference-to-video." It costs nothing and materially raises the chance your case is classified correctly by downstream indexers.
Method and limitations
- "Mode unverified" does not mean "unknown." It means none of our three sources yielded the answer. The video has a definite generation mode; it just was not recorded.
- The inference itself has error. Detecting prompt-text features (such as a literal
first frame) can misfire; those cases carry aweak classification basisflag — 26 of them. - On correlation: this piece reports a strong association between source channel and field completeness. It does not claim single causation — channels may also differ in which genres they prefer to collect.
- All figures use the 1,274 index records, without de-duplication by original publication (see the separate piece on dedup scope).
Reproduce it
Filter by generation mode, or group by source site: https://cases.aishifu.shop/