Authentic Ground Truth Photography for AI Training

Project Title: Real-World Scene Decomposition Photography (Ground Truth Dataset)
Project Overview
We are developing a state-of-the-art AI model focused on understanding lighting, perspective, and shadows in real-world environments. To achieve this, we require authentic, unedited, real-life photographic data—absolutely no digital manipulation or AI-generated content.
Our mission: Capture 50–200 complete "Series" per submission, with each series deconstructing a single real-world scenario into its fundamental visual components.
The Task: Create 50–200 Complete "Series"
Each series must contain 5–7 photos that systematically break down one authentic scene:
Table
Shot
Description
Example
Shot 1
Full Scene – Authentic "in-action" scenario
Someone cooking in a kitchen, gardening in a backyard, working at a desk
Shot 2
Background Only – Identical environment, person and movable objects removed
Empty kitchen, garden without person/tools, desk without person/items
Shot 3
Isolated Person – Same individual from Shot 1, photographed separately against a different background
Subject against a plain wall or outdoor setting
Shots 4+
Isolated Objects – Minimum 3 unique objects from the scene, shot individually on a neutral surface (white/gray/black backdrop)
Cooking utensils, plants, books, tools, etc.
Total per series: 6–10 individual images
Total per submission: 300–1,000 individual images (50–100 series × 6–10 images)
Technical Requirements (NON-NEGOTIABLE)
Table
Specification
Requirement
Device
DSLR, Mirrorless Camera, or High-End Smartphone (iPhone 12 Pro/Pro Max or newer, Samsung Galaxy S21 Ultra or equivalent)
Format
JPG only, sRGB color space
Resolution
4MP – 50MP
File Size
Maximum 45MB per image
Editing
ZERO digital edits – No filters, no color correction, no cropping, no retouching, no generative AI
Lighting
Natural, authentic lighting conditions only
Deliverables Per Submission
✅ 50-200 Complete Series (approximately 300–1,000 total images)
✅ Organized Google Drive folder with "View" permissions enabled for our team

Filename

Series ID

Shot type (Full Scene/Background/Person/Object)

Object name (if applicable)

Camera/device used

Timestamp

Brief scene description
Trial Submission Requirements
⚠️ MANDATORY: All applicants must complete a trial submission of minimum 50 series before receiving paid assignments.
Table
Phase
Requirement
Compensation
Trial Phase
50+ complete series (300+ images)
Unpaid quality assessment
Production Phase
50–100 series per submission
Paid per submission
The trial submission demonstrates your ability to maintain consistency, follow technical specifications, and manage high-volume workflows. Only approved trials proceed to paid work.
Compensation Structure
Per Submission Rate:
(Each submission = 50–200 complete series / 300–1,000 images)
Volume Opportunity:
We require 60–95 submissions total for this project phase. Each submission contains 50–200 series. Successful contributors will receive numbered submission assignments (#001, #002, etc.) after trial approval.
Estimated Total Project Volume: 3,000–8,500 series (18,000–85,000 individual images)
How to Apply
⚠️ CRITICAL: To be considered, your proposal MUST begin with the words:
"Authentic Data"
Your application must include:
1.
Equipment specification (Camera model or Phone model with specific model year)
2.
Portfolio sample (5–10 examples of unedited, natural-light photography showing consistency)
3.
Confirmation that you can deliver 50+ series for the trial and 50–200 series per paid submission
4.
Workflow description – How you will organize, shoot, and deliver 300–1,000 images per batch
5.
Availability – Timeline for completing trial submission and subsequent paid submissions
6.
Proposed rate per submission (50–200 series)
Selection Process
Table
Step
Action
Timeline
1. Initial Review
Verify "Authentic Data" header, equipment specs, and capacity confirmation
48 hours
2. Trial Assignment
Approved applicants receive trial guidelines and series numbering (#T001–#T050+)
Immediate
3. Trial Submission
Submit minimum 50 series via organized Google Drive with complete metadata
7–14 days
4. Quality Assessment
We evaluate consistency, technical compliance, and metadata accuracy
5–7 days
5. Production Assignment
Successful candidates receive numbered submission assignments (#001, #002, etc.)
Immediate
6. Production Cycle
Submit 50–200 series per deliverable, paid per approved submission
Ongoing
Critical Requirements for High-Volume Production

✅ Systematic workflow – You need a reliable method to track 50–200 distinct scenes, props, and backgrounds

✅ Consistent lighting – All shots within a series must match; across 50+ series, maintain documentation of conditions

✅ Organized file management – Naming conventions must be bulletproof (e.g., Series001_Shot01_FullScene.jpg )

✅ Metadata discipline – Spreadsheet must be 100% accurate for 300–1,000+ rows per submission

✅ Storage capacity – 300–1,000 images × 45MB max = up to 45GB per submission

✅ Backup protocol – Protect against data loss during large-scale production
Absolute Prohibitions
❌ Any digital editing – Automatic rejection if filters, color correction, cropping, or retouching detected
❌ AI-generated content – This project trains AI, it does not use AI
❌ Incomplete series – Every series must have all 5–7 shots; no partial submissions accepted
❌ Missing metadata – Spreadsheet errors will result in rejection
This is a high-volume, precision-focused project for serious professional photographers or organized production teams. If you have the equipment, workflow discipline, and capacity to deliver 50–100 series per submission, we want to hear from you.
Submit your proposal beginning with "Authentic Data" to begin the trial process.
We review applications within 48 hours. Trial submissions accepted on a rolling basis.