Freelance Photographer - Training Dataset Capture

Freelance Photographer - Training Dataset Capture Type: Fixed-price, per-batch Location: Remote/on-location (global) Client: Glass Imaging; overseen by Head of Image Quality Engineering About the project Glass Imaging builds AI-powered image processing. We need photographers to capture large volumes of images used to train our neural networks. These are dataset images, not portfolio work; aesthetic value doesn't matter. What matters is technical compliance, scene variety, and consistency. Required equipment - Camera: Sony A1 II or Sony A7RVI (required, no substitutes) - Lens: Sony GM 24mm or GM 70-200mm f2.8 or GM 100mm Macro 2.8 (required, no substitutes) - Tripod required for low-light scenes Capture settings (mandatory) - ISO 100 - RAW File Type: Uncompressed - Resolution: 50MP or 67MP (native of the camera) - NR Shooting mode: ON, Number of Shots: 32, 30fps burst mode (max drive speed) - Deep depth of field: minimal to no bokeh or foreground/background blur (stop down as needed; entire scene should be in focus); Generally between f8-f13 (depends on subject distance, lens focal length. Can use a depth-of-field calculator if needed to figure out the correct depth of field to get the scene in focus, or check on live view). Also see example images. - No in-camera creative styles, filters, or crops Files that deviate from these settings will be rejected. Scene requirements Each batch must cover a variety of scene types, including: - Textures (fabric, stone, wood, metal, print, etc.) - Foliage and natural detail - Architecture and man-made structures (especially repeating structures like office skyscraper windows, railings etc) - People — public/incidental only: street-style scenes where people are not the primary subject. No posed portraits or close-ups of identifiable individuals that would need a model release to be signed. - Indoor scenes (homes, offices, retail, varied artificial lighting) - Outdoor scenes (daylight, overcast, shade, golden hour) - Low light / night scenes (ISO 100 on tripod still, avoid motion blur) - Macro images of textures, text, products, fabrics, natural foliage etc (all in focus, no shallow depth of field) - A range of different focal lengths and scene distances across the lenses specified. We will provide a set of sample images, please follow the rough types of content we show but add diversity. A per-batch scene checklist and quota will be provided at kickoff. Deliverables - Target 1000 captures (bursts of 32 each) total - Batches of 100 compliant RAW captures (each capture = one 32-shot NR sequence) - Organized in batches for upload - Delivered via Google Drive or WeTransfer (we will provide a login / upload folder) - Original unedited files only — no post-processing Payment - Fixed price per accepted batch of 100 captures meeting spec - Batches reviewed by our Image Quality team; non-compliant captures must be reshot or are excluded from the count - Ongoing work available for photographers who deliver consistently Rights & usage All delivered images become the property of Glass Imaging with full rights transferred for machine learning training and related use. Images will not be published commercially as photographs. We will provide a photo release form to sign. To apply, please include - Confirmation that you own or can reliably rent a Sony A1 II / A7RVI with one or more of the specified lenses - Your location and the scene types easily accessible to you - Rough weekly capture capacity - A few sample images; technical quality matters more than artistry