screenshat vs ML Data Factory

screenshat captures full-page web screenshots at the sizes you actually ship. Paste one URL, choose from 18 presets, and get social cards, mobile viewports, and high-resolution exports back in one job. If you've ever tried to grab a clean OG card on a phone, you know the drill: screenshot, crop, resize, repeat for every size, and the result still only covers the visible viewport. screenshat renders the page at every size you picked and hands them all back together. WHAT YOU CAN DO - Capture a page across 18 presets in one run: Open Graph and social cards, iPhone, Pixel, and iPad viewports, plus 2K through 16K output in portrait and landscape. - Watch each screenshot appear as soon as its preset finishes instead of waiting for the whole job. - Select any combination of finished shots to save to Photos or share as PNGs, or download the complete job as a ZIP. - Retry only the presets that failed instead of starting everything over. - Send a link from Safari with the Share or Action extension. The URL is waiting in the capture form next time you open the app. - Give slow pages time to settle: wait for full load, network idle, or add a fixed delay before the shot. - Reopen captures from on-device history to check progress or export the files again. THE HONEST PART - Public pages only. Sites behind a sign-in won't render. - Captures run on the screenshat server, not on your phone. Your URL goes to the server; the PNGs come back. - Full-page means full page: the rendered document top to bottom at the preset's exact pixel size, not a viewport crop. - No account. The in-app history index stays on your device; capture jobs and PNGs remain on the server until deletion or the roughly 30-day retention window expires. ON IPAD Sidebar navigation, a wider capture form, and a list-and-detail history view. Made by Luke Steuber. Questions: luke@lukesteuber.com

ML Data Factory is a professional-grade mobile utility engineered to bridge the gap between raw video footage and the high-quality datasets required for machine learning. Designed for data engineers, sports analysts, and AI researchers, this app allows you to curate, manufacture, and package machine learning datasets entirely from your iPhone or iPad. Stop wasting time with bulky desktop software. Whether you are in the field or at your desk, ML Data Factory provides a streamlined, high-throughput pipeline to transform raw video into structured, AI-ready training data. Key Features: Precision Data Pipeline: Seamlessly ingest raw video from iCloud, the local Files app, or your Photo Library. Support for bulk-uploading allows you to queue multiple videos for efficient, uninterrupted processing. Direct-Link Ingestion: Paste direct video URLs to stream and download high-quality footage directly into your processing queue. Frame-Accurate Scrubbing: Utilize high-precision seeking controls to pinpoint the exact millisecond of an action. Whether you are analyzing a sports swing or a manufacturing process, you have full control with frame-by-frame navigation. ML-Ready Visual Guides: Built-in 224x224 crop guides ensure your visual subjects are perfectly framed for popular machine learning models (e.g., ResNet, MobileNet), ensuring data consistency before you ever snap a frame. Dynamic Dataset Labeling: Create, manage, and toggle custom categories on the fly. Organize your dataset into structured folders that match your specific training labels. Customizable Capture Length: Adjust your capture window dynamically—choose between 30, 60, 90, 120, or 150 frames to match the specific needs of your ML model's temporal requirements. One-Tap Manifest Export: With a single tap, the app compiles your categorized clips into a structured dataset_manifest.json and bundles everything into a single .zip archive. Easily AirDrop or transfer your entire, ready-to-train dataset to your workstation for immediate use in Create ML, PyTorch, or TensorFlow. Technical Specifications: High-Speed I/O: Built on the AVFoundation framework for efficient, asynchronous video processing and extraction. Optimized Performance: Fully compliant with modern iOS 18+ standards for maximum stability and performance on the latest hardware. Data Integrity: All dataset operations are performed using atomic write protocols to ensure no data is corrupted or lost during the extraction and zipping process. User-Centric Design: High-contrast, dark-themed industrial interface designed for high-focus professional environments.