Local Browser Tool
See what metadata your Midjourney downloads carry, remove it locally, and verify the result — all in your browser, with no file uploads.
Checking plan
Local Browser Tool
Checking plan
Drop photos hereTap to choose photos
Drop up to 3 images per batch (JPEG, PNG, or static WebP, 15 MB each, 30 MB total). The tool inspects supported metadata fields, cleans them losslessly, and re-scans the output. 10 images per day, shared across the homepage image tool, EXIF Remover, and this page.Drop up to 3 images per batch (JPEG, PNG, or static WebP, 15 MB each, 30 MB total). The tool inspects supported metadata fields, cleans them losslessly, and re-scans the output. 10 images per day, shared across the homepage image tool, EXIF Remover, and this page.
Supported metadata is removed automatically on this device. You can also paste images from the clipboard. · Up to 3 images · 15 MB each · 30 MB total
10 of 10 free images remaining today · Upgrade to Pro
Midjourney downloads may contain different metadata depending on the download path and format. No single field is guaranteed, and absence of metadata does not mean an image is not from Midjourney or is not AI-generated.
Images may contain EXIF tags such as Software or ImageDescription, XMP properties, and IPTC fields when present. These can include descriptive text, but this tool does not interpret any Description field as a prompt, Job ID, UUID, or proof of Midjourney origin.
An image file can carry a possible C2PA content-credentials marker when present. This tool detects the structural marker but does not cryptographically verify it. The IPTC DigitalSourceType value trainedAlgorithmicMedia is a generic self-declaration that can be added or removed by anyone and cannot identify Midjourney as the source.
The metadata embedded in a downloaded image depends on how you obtained it. Different paths can produce different file formats and metadata profiles.
Images downloaded from the Midjourney website may arrive as PNG or JPEG with embedded EXIF, XMP, IPTC, or C2PA data when present. The specific fields vary by image and may change as the platform evolves.
Images saved from the Midjourney Discord bot may carry different metadata than web downloads when present. Discord's own processing can also strip or alter certain fields before the file reaches your device.
The tool strips supported metadata containers — EXIF, XMP, IPTC, and C2PA markers — without modifying compressed image data. ICC color profiles and decode-essential structures are preserved.
Compressed pixel data remains untouched. EXIF Orientation is preserved by default so your image displays correctly; you can choose to remove it. ICC color profiles and other decode-essential structures are kept intact. No re-encoding occurs.
C2PA markers carry content-provenance information. When a possible C2PA marker is detected, cleaning pauses and no output is produced for that image until you provide explicit consent to remove it.
Metadata is not the only thing that can indicate where an image came from. A source filename itself can disclose labels or identifiers when present.
A filename may contain labels, identifiers, or other text that indicates where the image came from. These are visible to anyone who receives the file, regardless of whether internal metadata has been removed.
This tool defaults to generic filenames: clean-image-<session>-01.<ext>, where the session token is a 16-character hex string generated once in your browser, never stored or reported. Files are numbered in accepted-file order. You can switch to source-derived naming (<original-name>-clean.<ext>) before downloading.
Removing file metadata is a specific, limited operation. It is important to understand what remains unaffected.
Cleaning does not affect visible watermarks, invisible watermarks, SynthID, steganographic signals, or features that visual classifiers may detect. These exist within the image data itself and are outside the scope of metadata removal.
Midjourney maintains its own records of generated images. Removing metadata from a downloaded file does not delete the image from Midjourney's servers, your generation history, or any other external system.
Many jurisdictions and platforms require disclosure when content is AI-generated. Removing metadata does not change these obligations.
Removing metadata from an AI-generated image does not waive any legal, regulatory, or platform-specific requirement to disclose that the image was created with AI. You are responsible for meeting applicable disclosure obligations.
The IPTC DigitalSourceType value trainedAlgorithmicMedia is a voluntary, generic self-declaration. It can be added or removed by anyone and does not identify Midjourney or any specific platform. Its presence or absence proves nothing about an image's origin.
The entire process runs locally in your browser. No files are uploaded to any server.
Drop or select up to 3 Midjourney images (JPEG, PNG, or static WebP, 15 MB each). The tool reads and displays supported metadata fields found in each file.
Remove supported metadata losslessly with one click. Compressed image data stays untouched, and ICC color profiles are preserved. Orientation is kept by default. If a possible C2PA marker is detected, cleaning pauses until you provide explicit consent; no output is produced without it.
A re-scan of the cleaned file confirms which fields were removed. Download your cleaned images with generic filenames by default, or switch to source-derived naming before downloading.
Common questions about removing metadata from Midjourney images.
Other metadata tools on this site that may be useful.
