# Changelog of Video Frame Extractor - Keyframes & Scenes, 100 Videos/Run (`automa-flow/video-frame-extractor`) Actor

- **URL**: https://apify.com/automa-flow/video-frame-extractor/changelog.md
- **Full Actor documentation**: https://apify.com/automa-flow/video-frame-extractor.md

## Changelog

### 2026-10-19: Scene detection on longer videos

#### Fixes

- `scenes` mode no longer stops videos longer than about 12 minutes at the default `processingTimeoutSeconds`. The scene scan decodes the whole video, so it may now run for 1.5 times the video's length when that is longer than `processingTimeoutSeconds`.

#### Improvements

- When a scene scan still runs out of time, the video row says what to change: 2048 MB memory, a higher `processingTimeoutSeconds`, or `keyframes` mode. It is not charged.
- The retry advice in `RUN_SUMMARY` now states that a video you send again in a new run is processed and charged again in full.

### 2026-10-19: First version (0.1)

The first public release on the Apify Store.

#### Features

- Extract frames from up to 100 direct video URLs, or from a Dataset of a previous Actor with `datasetId`, `sourceField`, `idField` and `datasetOffset`.
- Five modes: `count` (evenly spaced frames), `interval` (one frame every N seconds), `timestamps` (frames at the seconds you list per file), `scenes` (frames at detected scene changes, with a score) and `keyframes` (frames at the stream's keyframes).
- One Dataset row per stored frame with a public image link, and one status row per video. Views: Frames, Videos and Failed.
- JPEG or WebP frames, an optional width cap and an optional free contact sheet per video.
- A restarted run republishes finished rows without charging them again.

#### Pricing and limits

- $0.002 per stored frame, $0.005 per processed video, and $0.005 per started minute scanned in `scenes` mode. Failed videos, failed frames, retries and contact sheets are not charged.
- 1-200 frames per video, 256 MiB per file by default (1 GiB maximum), 2 GiB per run by default.