Video Frame Extractor — YouTube, TikTok & MP4 to Images & OCR
Pricing
from $0.99 / 1,000 frame extracteds
Video Frame Extractor — YouTube, TikTok & MP4 to Images & OCR
Extract high-res still frames, thumbnails & scene-change keyframes from MP4, YouTube, TikTok, Instagram Reels, Facebook, X or M3U8 links: evenly spaced, every N seconds, scene cuts, or exact timestamps. Outputs JPG/PNG/WebP URLs & Base64 for GPT-4o Vision, Claude, Gemini & n8n.
Pricing
from $0.99 / 1,000 frame extracteds
Rating
0.0
(0)
Developer
Axon Studio
Maintained by CommunityActor stats
0
Bookmarked
2
Total users
1
Monthly active users
2 days ago
Last modified
Categories
Features
Share
Extract high-resolution still frames, video thumbnails, scene-change keyframes, and composite Storyboard Contact Sheet grids from any direct video file (.mp4, .mov, .webm, .mkv, .m3u8) or social video URL (YouTube, Shorts, TikTok, Instagram Reels, Facebook, X/Twitter, Vimeo, Bilibili) into permanent public image URLs (.jpg, .png, .webp) and optional inline Base64 Data URIs.
Built on native FFmpeg + ffprobe + yt-dlp for multimodal Vision AI pipelines (OpenAI GPT-5, Anthropic Claude Vision, Google Gemini), n8n / Make automations, video OCR, and ad-intelligence storyboarding. 100% self-contained with zero external API keys required — priced from $0.99 / 1,000 frames with an instant free preview in Apify Console.
| Need | Actor |
|---|---|
| Video Frame, Thumbnail & Scene Extractor (视频抽帧/九宫格分镜提取器) | axon-studio/video-frame-extractor (Current) |
| Douyin Videos, 1080p MP4, Comments & Shop (抖音无水印视频与带货) | axon-studio/douyin-scraper |
| Bilibili Videos, Danmaku & Comments Suite (哔哩哔哩 B站全能母舰) | axon-studio/bilibili-scraper |
| WeChat Official Account Articles & Markdown (微信公众号文章深搜) | axon-studio/wechat-official-account-scraper |
| Taobao & Tmall Detail, Search, Shop & Reviews (淘宝详情/搜索/店铺) | axon-studio/taobao-detail-scraper |
| Taobao & Tmall Real-Time Sales & Stock Search (淘宝天猫销量库存) | axon-studio/taobao-tmall-scraper |
| Weidian Products, SKUs, Stock & Agent Links (微店商品与尺码库存) | axon-studio/weidian-scraper |
| Goofish / Xianyu Second-Hand Deals (闲鱼二手商品与卖家) | axon-studio/goofish-xianyu-scraper |
| Sina Weibo Search, Posts & Hot Trends (新浪微博全能抓取) | axon-studio/weibo-scraper |
Copy to your AI assistant (Claude, ChatGPT, Cursor & Windsurf)
Help me use axon-studio/video-frame-extractor on Apify to extract high-resolution keyframes, thumbnails, scene-change snapshots, and storyboard contact sheets from MP4, YouTube, TikTok, X, or Instagram Reel videos for Vision AI and OCR analysis. Read https://apify.com/axon-studio/video-frame-extractor.md for complete documentation, input schema, output fields, and pricing. Configure the optimal input JSON for my use case.
What Can This Video Frame Extractor Do? (Features)
| Capability | Built-In Advantage |
|---|---|
4 Precision Sampling Modes (mode) | Extract Evenly Spaced frames across the video (evenly_spaced), Every N Seconds (every_n_seconds), automatic Scene Change Detection (scene_change with pixel-histogram sceneScore), or Exact Timestamps (timestamps). |
Storyboard Contact Sheet Grid (includeContactSheet) | Automatically stitches all extracted frames into a single composite Storyboard Contact Sheet grid image (configurable columns 1–10) — cutting Vision LLM token costs by up to 80%. |
Custom Time Range Clipping (startTime / endTime) | Restrict frame sampling to any specific segment of a video (e.g., from 00:05 to 00:45) without processing the entire file. |
| Direct Files & 1,800+ Video Platforms | Accepts direct .mp4, .mov, .webm, .mkv, .m3u8 URLs (including Google Drive & Dropbox) as well as YouTube, Shorts, TikTok, X/Twitter, Instagram Reels, Facebook, Vimeo, and Bilibili links. |
Smart De-Blur Filter for OCR (skipBlurryFrames) | Applies FFmpeg's multi-frame thumbnail=n=5 sharpness analysis to pick the crispest frame in each window for slide decks and on-screen text OCR. |
| Permanent Image URLs + Optional Base64 | Saves every frame and contact sheet to the Apify Key-Value Store (imageUrl) with live previews in the Output table, plus optional inline data:image/...;base64,... (base64DataUri). |
How to Extract Video Frames in 3 Easy Steps
- Enter
videoUrls: Paste one or more.mp4/.m3u8file URLs or TikTok, X/Twitter, YouTube, Vimeo, or Instagram Reel links (or run immediately with the prefilled Sintel movie trailer). - Choose
mode& Options: Selectevenly_spaced,every_n_seconds,scene_change, ortimestamps, set optionalstartTime/endTime, and keepincludeContactSheet: trueenabled to generate a composite storyboard grid. - Click "Start": View the visual Frame Preview Gallery directly in the Output tab, or feed the
imageUrl/base64DataUriarray into OpenAI, Claude Vision, Gemini, or n8n.
Video Frame Extractor Use Cases
| Persona / Team | Multimodal AI & Automation Use Case |
|---|---|
| AI Agent & Vision LLM Builders | Convert .mp4, TikTok, X, and YouTube videos into composite Contact Sheet grids or discrete imageUrl / base64DataUri arrays for OpenAI Vision, Anthropic Claude, and Google Gemini. |
| Ad Creative & Marketing Agencies | Run scene_change mode on viral TikTok, Reels, and Shorts ads to dissect 3-second visual hooks, pacing, scene cut frequency, and CTAs. |
| Webinar, Lecture & Slide Deck OCR | Enable scene_change + skipBlurryFrames: true to automatically capture a sharp 1080p screenshot every time a presenter changes slides. |
| Compliance & Content Moderation | Sample frames every N seconds (every_n_seconds) within specific time ranges across user-generated video uploads for automated brand-safety inspection. |
Input
All 15 parameters are optional and prefilled ("required": []). Below is the complete 1-to-1 parameter reference table and JSON input examples:
| Parameter | Type | Default | Description |
|---|---|---|---|
videoUrls | array[string] | ["https://media.w3.org/2010/05/sintel/trailer.mp4"] | Direct video file URLs (.mp4, .mov, .webm, .mkv, .m3u8) or social video links (YouTube, Shorts, TikTok, Instagram Reels, Facebook, X/Twitter, Vimeo, Bilibili). |
mode | string | "evenly_spaced" | Sampling strategy: "evenly_spaced", "every_n_seconds", "scene_change", or "timestamps". |
count | integer | 10 | In evenly_spaced mode: exact number of frames per video (1–200). In every_n_seconds and scene_change modes: maximum frame cap per video. |
intervalSeconds | integer | 5 | Seconds between consecutive captures when mode is "every_n_seconds" (1–3600). |
sceneSensitivity | string | "medium" | Sensitivity for "scene_change" mode: "low" (0.45), "medium" (0.30), or "high" (0.15). |
timestamps | array[string] | [] | Target timestamps when mode is "timestamps" (e.g., ["00:05", "00:20", "00:40"] or ["5", "20", "40"]). |
startTime | string | "" | Optional start timestamp (in seconds "5" or "00:05" / "00:00:05.000"). Frame extraction begins from this point. |
endTime | string | "" | Optional end timestamp (in seconds "45" or "00:45" / "00:00:45.000"). Frame extraction stops at this point. |
imageFormat | string | "jpg" | Output image encoding format: "jpg", "png", or "webp". |
maxWidth | integer | 1280 | Maximum image width in pixels (160–3840), preserving aspect ratio without upscaling. |
includeContactSheet | boolean | true | Stitches all extracted frames from each video into a composite Storyboard Contact Sheet grid image (extractionMode: "contact_sheet"). |
contactSheetColumns | integer | 5 | Number of columns per row in the generated Storyboard Contact Sheet grid image (1–10). |
skipBlurryFrames | boolean | false | Enables FFmpeg's thumbnail=n=5 sharpness filter to select the least motion-blurred frame. |
includeBase64 | boolean | false | Embeds the complete data:image/...;base64,... string in base64DataUri. |
proxyConfiguration | object | {"useApifyProxy": false} | Optional proxy settings for geo-restricted social videos. |
1. Mode evenly_spaced + Storyboard Contact Sheet Grid (JSON)
{"videoUrls": ["https://media.w3.org/2010/05/sintel/trailer.mp4"],"mode": "evenly_spaced","count": 10,"includeContactSheet": true,"contactSheetColumns": 5,"imageFormat": "jpg","maxWidth": 1280}
2. Mode scene_change + Custom Time Range Clipping (JSON)
(Tip: Combine "startTime" and "endTime" with "scene_change" or "every_n_seconds" to sample keyframes from a specific video segment)
{"videoUrls": ["https://x.com/SpaceX/status/1845442658390335784"],"mode": "scene_change","sceneSensitivity": "medium","startTime": "00:05","endTime": "00:35","count": 10,"includeContactSheet": true,"skipBlurryFrames": true}
Output
Each extracted frame (plus the composite Contact Sheet grid when includeContactSheet: true) is pushed as a structured JSON item to the default Dataset and rendered as a visual thumbnail in the Output table:
{"imageUrl": "https://api.apify.com/v2/key-value-stores/abcdef123456/records/video_01_frame_0001.jpg","frameIndex": 1,"timestampSeconds": 2.609,"timestampFormatted": "00:00:02.609","extractionMode": "evenly_spaced","sceneScore": 0.2,"imageFormat": "jpg","width": 1280,"height": 546,"fileSizeBytes": 64210,"imageKey": "video_01_frame_0001.jpg","videoIndex": 1,"videoUrl": "https://media.w3.org/2010/05/sintel/trailer.mp4","videoTitle": "trailer","videoDuration": 52.209,"videoWidth": 1280,"videoHeight": 546,"videoCodec": "h264","videoFps": 24.0,"base64DataUri": "(Disabled — Enable 'Include Base64' in Input to embed full Base64 string)","extractedAt": "2026-11-17T15:30:00Z"}
Output Field Definitions
| Field | Type | Description |
|---|---|---|
imageUrl | string | Permanent public HTTPS URL pointing directly to the extracted frame or contact sheet in the Apify Key-Value Store. |
frameIndex | integer | 0 for the composite Storyboard Contact Sheet grid; 1..N for individual sequential frames. |
timestampSeconds / timestampFormatted | number / string | Exact position in seconds (2.609) and formatted timestamp (00:00:02.609). |
extractionMode / sceneScore | string / number | Sampling mode used (evenly_spaced, every_n_seconds, scene_change, timestamps, or contact_sheet) and visual change score (0.0–1.0). |
imageFormat / width / height / fileSizeBytes | string / integer | Image format (jpg/png/webp), scaled pixel dimensions, and file size in bytes. |
imageKey / videoIndex / videoUrl / videoTitle | string / integer | KV Store key, 1-based video index, source URL, and resolved video title. |
videoDuration / videoWidth / videoHeight / videoCodec / videoFps | number / integer / string | Native video stream duration, resolution, codec (h264/vp9/av1), and frame rate. |
base64DataUri / extractedAt | string | Inline data:image/...;base64,... URI (when includeBase64: true) and UTC extraction timestamp. |
Data Coverage (Always Populated vs. Conditional)
| Coverage Tier | Fields | Condition |
|---|---|---|
Always Populated (100%) | imageUrl, frameIndex, timestampSeconds, timestampFormatted, extractionMode, sceneScore, imageFormat, width, height, fileSizeBytes, imageKey, videoIndex, videoUrl, videoTitle, videoDuration, videoWidth, videoHeight, videoCodec, videoFps, base64DataUri, extractedAt | Present on every extracted record (0 null columns). |
Multimodal AI & API Integration (Python, JavaScript, cURL & n8n)
Python (apify-client)
from apify_client import ApifyClientclient = ApifyClient("YOUR_APIFY_API_TOKEN")run_input = {"videoUrls": ["https://media.w3.org/2010/05/sintel/trailer.mp4"],"mode": "evenly_spaced","count": 10,"includeContactSheet": True,"contactSheetColumns": 5,"imageFormat": "jpg","maxWidth": 1280,}run = client.actor("axon-studio/video-frame-extractor").call(run_input=run_input)for frame in client.dataset(run["defaultDatasetId"]).iterate_items():print(f"[{frame['timestampFormatted']}] #{frame['frameIndex']}: {frame['imageUrl']}")
JavaScript / Node.js (apify-client)
import { ApifyClient } from 'apify-client';const client = new ApifyClient({ token: 'YOUR_APIFY_API_TOKEN' });const run = await client.actor('axon-studio/video-frame-extractor').call({videoUrls: ['https://media.w3.org/2010/05/sintel/trailer.mp4',],mode: 'scene_change',sceneSensitivity: 'medium',count: 10,includeContactSheet: true,});const { items } = await client.dataset(run.defaultDatasetId).listItems();console.log(items);
cURL (Synchronous Execution)
curl -X POST "https://api.apify.com/v2/acts/axon-studio~video-frame-extractor/run-sync-get-dataset-items?token=YOUR_APIFY_API_TOKEN" \-H "Content-Type: application/json" \-d '{"videoUrls": ["https://media.w3.org/2010/05/sintel/trailer.mp4"],"mode": "evenly_spaced","count": 10,"includeContactSheet": true}'
How Much Does It Cost to Extract Video Frames? (Pricing)
This Actor uses transparent Pay-Per-Event (PPE) billing with no monthly subscription and no bandwidth surcharges:
| Billing Event Name | Trigger Condition | Price |
|---|---|---|
frame-extracted (Primary) | 1 per extracted frame or contact sheet saved to Dataset & KV Store | From $0.99 / 1,000 frames (Gold tier; $1.50 / 1,000 base rate) |
video-processed | 1 per video stream probed and processed | $5.00 / 1,000 videos ($0.005 per video) |
- Free Apify Plan Preview Policy: Free-tier users can extract up to 1 video and 10 frames per run (plus 1 composite Contact Sheet grid) with no credit card required.
FAQ & Legal Compliance
What video formats and platforms are supported?
Direct .mp4, .mov, .webm, .mkv, .avi, and .m3u8 HLS links (including Google Drive and Dropbox) as well as public links from TikTok, X/Twitter, YouTube, Instagram Reels, Facebook, Vimeo, and Bilibili.
Can I pass the extracted frames directly to OpenAI, Claude Vision, or Gemini?
Yes. Pass either the composite Contact Sheet imageUrl (or individual frame URLs) directly into Vision LLM image blocks, or set includeBase64: true to receive inline data:image/jpeg;base64,... strings.
Is it legal to extract frames from public videos?
This utility processes publicly accessible video URLs provided by the user for analysis, storyboarding, and AI workflows. Ensure your usage complies with applicable copyright laws and platform terms of service.