Loading...
Loading...
Printable Reference
Folder naming, export checklists, delivery packages, and the redact-vs-preserve decision tree.

What This Pack Covers
Export Checklist
Reliable steps for getting images out of MJ with metadata intact
Folder System
Project, style, and client folder structures that scale
QA & Delivery
Pre-delivery checks and client package templates
Decision Tree
When to preserve, strip, or selectively redact metadata
Midjourney Export Checklist
Follow these steps every time you export images from the Midjourney web app to ensure nothing is lost in transit.
Single Image Download
Open the image in the MJ web app at full resolution
Click the download button (not right-click save)
Save to your project folder using the naming convention
Verify metadata: open file properties and confirm the Description field contains your prompt + parameters
Batch Download
Select images using checkboxes in the MJ web grid
Click the bulk download button to get a ZIP file
Extract the ZIP into your project folder
Batch rename files using your project naming convention (see Naming Cheat Sheet)
Spot-check 2-3 files to confirm metadata is present (single and batch downloads contain identical metadata)
MJ Metadata Is Identical for Single and Batch Downloads
Every downloaded image includes the same embedded fields: Description (prompt + parameters + Job ID as text), Digital Image GUID (Job ID UUID), Author (your MJ username), Creation Time, IPTC Digital Source Type, and XMP Toolkit. There is no difference between single and batch export.
Post-Download Organisation
Sort into folders
Move files into project/client/style folders (see Page 2)
Review & cull
Move rejected images to _archive/ rather than deleting
Back up originals
Keep unmodified downloads separate from edited finals
License: All checklists and templates on this page are released under the MIT License. You are free to use, copy, modify, and distribute them in personal and commercial projects without attribution.
Disclaimer: These workflows are provided "as is" without warranty of any kind. They are educational examples for common Midjourney workflow operations. Numonic Labs Ltd and its contributors are not liable for any damages arising from their use.
1
Page 2 of 5
A consistent folder structure means you can find any image months later without relying on search. These patterns work on any OS and sync cleanly with cloud storage.
Project-Based Structure
Best for campaigns, product launches, and one-off creative briefs.
MJ/ ├── 2026-03/ │ ├── hero-redesign/ │ │ ├── raw/ # Unmodified MJ downloads │ │ ├── finals/ # Edited, approved versions │ │ └── _archive/ # Rejected variations │ └── product-shots/ │ ├── raw/ │ └── finals/ ├── 2026-02/ │ └── brand-refresh/ └── _templates/ # Reusable prompt templates
Style Reference Collections
For curating and reusing --sref codes across projects. Name folders by the visual style, not the sref code (codes are in the metadata).
MJ/
└── styles/
├── watercolour-botanical/
│ ├── reference-samples/
│ └── project-outputs/
├── 90s-retrofuture/
└── clean-product-render/Client Work Structure
Keeps client deliverables separate from internal exploration. Includes a dedicated delivery/ folder for final packages.
MJ/
└── clients/
└── acme-corp/
├── spring-campaign/
│ ├── raw/
│ ├── finals/
│ ├── delivery/ # What the client receives
│ └── notes.txt # Brief, feedback, revisions
└── logo-concepts/Image Naming Convention
MJ auto-generates names from prompts, which are often long and cryptic. Rename using:
project_mj_YYYYMMDD_seq.png
Example: hero-redesign_mj_20260309_001.png
MJ Saved Searches That Map to Folders
Create saved searches in the MJ web app to mirror your folder structure for quick access:
MJ Web App Folders Work Great for Casual Use
The MJ web app has folders, folder groups, and bulk actions that are excellent for organising while you work. The local folder system above is for the post-export lifecycle: once images leave midjourney.com and live on your hard drive, NAS, or cloud storage.
2
Page 3 of 5
Use these checklists before sending AI images to clients, publishing to social media, or archiving for long-term storage.
Pre-Delivery QA Checklist
Resolution & Format
Image resolution meets deliverable spec (e.g. 1024x1024, 1920x1080)
Aspect ratio matches brief requirements (--ar parameter)
File format is correct (PNG for lossless, JPEG for web delivery)
Upscaled if needed (MJ upscale or external tool)
Content Review
No visible artifacts, distorted hands/text, or hallucinated watermarks
Colours render correctly on both screen and print (if applicable)
No unintended brand logos, trademarked elements, or identifiable faces
Metadata & Privacy
Metadata reviewed: Description field contains prompt text (check if this should be preserved or stripped)
Prompt privacy check: ensure no confidential client names, proprietary terms, or sensitive keywords in the Description
IPTC Digital Source Type set to trainedAlgorithmicMedia (MJ does this automatically)
Client Delivery Package
What to Include
Final images (renamed, organised by use case)
Usage notes: dimensions, file formats, intended placement
AI disclosure statement (see decision tree on Page 4)
License summary: MJ Terms of Service grant commercial rights to paid subscribers
Sample Disclosure Statement
"These images were generated using Midjourney, an AI image generation tool, and may have been edited or composited. They are licensed for commercial use under Midjourney's Terms of Service (paid plan). IPTC metadata identifies the digital source type as AI-generated."
Delivery Format Options
Standard delivery, preserves folder structure
Ongoing collaboration, easy updates
Professional delivery with metadata preview
Internal Archive Checklist
Raw MJ downloads preserved separately from edited finals
Metadata kept intact on archived originals (never strip archive copies)
Project notes file included: brief summary, key prompts used, style references
Folder follows naming convention from Page 2
Backed up to at least one additional location (cloud, NAS, or DAM)
Automate with a DAM
A Digital Asset Manager like Numonic can automate QA checks, metadata preservation, and delivery package creation. Drop files in your sync folder and let the system handle naming, tagging, and archive copies.
3
Page 4 of 5
Not every image should keep its metadata. Use this decision tree to determine what to preserve, strip, or selectively redact before sharing.
Is this image for a client?
YES
Does the client know it's AI-generated?
YES
Preserve + Disclose
Keep all metadata. Include AI disclosure statement in delivery package. Transparent and audit-friendly.
NO
Strip + Discuss
Remove prompt metadata before delivery. Initiate AI usage policy conversation with client. Don't hide it — address it.
NO
Where is this being published?
Social Media
Selective Redact
Strip prompt text (privacy). Keep IPTC source type as trainedAlgorithmicMedia (compliance). Platforms may read this for labelling.
Internal / Portfolio
Preserve Everything
Keep all metadata intact. Your future self will thank you when you need to reproduce or iterate on a style.
Privacy Preset Recommendations
Full Preserve
Keep everything: prompt, parameters, Job ID, GUID, author, timestamps, IPTC source type. Best for internal archives and portfolios.
Use: Archive, portfolio, internal
Compliance Only
Strip prompt/parameters from Description. Keep IPTC Digital Source Type (trainedAlgorithmicMedia) and Creation Time. Meets disclosure requirements without exposing creative process.
Use: Social media, public content
Full Strip
Remove all AI-specific metadata. Only basic file information remains. Use only when contractually required or when metadata would cause issues.
Use: Specific client requirements only
EU AI Act Disclosure Deadline: August 2026
From August 2026, the EU AI Act requires that AI-generated content published in the EU be clearly labelled as such. The IPTC Digital Source Type field (which Midjourney sets automatically to trainedAlgorithmicMedia) is the emerging technical standard for machine-readable compliance. Even if you strip prompt text, preserve this field for any content that may reach EU audiences.
4
Page 5 of 5
Tear this page out. Everything you need to understand MJ metadata and parameters at a glance.
MJ Metadata Fields Reference
| Field | Contains |
|---|---|
| Description | Full prompt text + parameters + Job ID (as a single text string)This is where everything lives. No separate structured fields for --sref, seed, etc. |
| Digital Image GUID | Job ID as a UUIDUnique identifier for the generation job |
| Author | Your Midjourney usernameThe account that created the image |
| Creation Time | Timestamp of generationWhen the image was created on MJ servers |
| IPTC Digital Source Type | trainedAlgorithmicMediaMachine-readable AI disclosure (C2PA / EU AI Act ready) |
| XMP Toolkit | XMP metadata toolkit identifierStandard XMP packaging information |
Key Insight: Everything Is in Description
MJ packs the entire prompt, all parameters (--ar, --sref, --stylize, etc.), and the Job ID into a single Description text field. There are no separate structured fields for individual parameters. To extract specific values, you need to parse the Description string.
Common -- Parameters
| Parameter | What It Does |
|---|---|
| --ar | Aspect ratio (e.g. --ar 16:9, --ar 3:2) |
| --stylize | How strongly MJ applies its aesthetic (0-1000, default 100) |
| --chaos | Variation in results (0-100, default 0) |
| --weird | Experimental/unusual outputs (0-3000, default 0) |
| --sref | Style reference URL or code for consistent styling |
| --v | Model version (e.g. --v 6.1) |
| --no | Negative prompt — elements to exclude |
| --tile | Generate seamlessly tiling patterns |
| --q | Quality/compute time (0.25, 0.5, 1) |
Key URLs
Numonic MJ Organisation Guide
numonic.ai/guides/midjourney-organisation
Numonic Naming Cheat Sheet
numonic.ai/guides/organise-ai-generated-images/naming-cheat-sheet
Midjourney Documentation
docs.midjourney.com
IPTC Photo Metadata Standard
iptc.org/standards/photo-metadata
EU AI Act Overview
artificialintelligenceact.eu
Automate your entire MJ workflow
Numonic preserves MJ metadata automatically, applies your naming conventions, and handles the redact-vs-preserve decision with configurable privacy presets. Drop files in your sync folder—everything else is handled.
numonic.ai → Try free
numonic.ai
AI-Native Asset Management
Free to share. Attribution appreciated.