cleanor-storage-lab
Open datasets & reproducible benchmarks
Reproducible benchmarks and open datasets: image compression (JPEG/WebP/AVIF/JPEG XL), the HEIC-to-JPG size tax, cloud storage cost per GB, and phone storage capacity.
The language of phone cleanup, defined
Most storage-cleanup confusion is really vocabulary confusion. These plain-language definitions make the categories clear, what counts as a duplicate, why similar photos are harder, where to start, so the right feature or app choice becomes obvious.
3concept guides
21defined terms
Open datareproducible sources
Concept guides
Start with the right mental model
Each guide clears up one cluster of cleanup language before you pick a route.
Device storage glossary
Definitions for device storage, screenshots, downloads, large videos, duplicate photos, similar photos, and the categories that shape cleanup decisions.
Read guideDuplicate photos vs similar photos
A glossary page that explains the difference between duplicates and similar shots so users know why the cleanup logic, confidence level, and deletion order are different.
Read guideAI photo editing glossary
Definitions for AI photo editing, source portraits, identity preservation, age transformation, headshots, realism, and the terms that shape portrait-editing choices.
Read guideKey terms
Straight definitions
The core cleanup vocabulary in one place. Each term links into the guide it belongs to.
Device storage glossary
Open guide →- Device storage
- The overall space available on the phone, including photos, videos, downloads, apps, cached data, and other files that add up over time.
- Large videos
- Heavy video files, recordings, or exports that often create the fastest visible storage recovery when deleted or moved first.
- Screenshots
- Low-risk reference captures that usually have lower emotional value than camera photos and often make a good first cleanup pass.
- Downloads
- Saved files, attachments, exports, and documents that sit outside the camera roll but still quietly consume storage.
- Duplicate photos
- Obvious repeated images or copies that usually create lower-risk cleanup decisions than similar photos.
- Similar photos
- Separate images that look close enough to compete, which means the user still needs to decide which version is worth keeping.
- Cleanup order
- The practical sequence used to reduce stress and recover space faster, usually starting with heavy files and low-risk clutter before moving into harder judgment calls.
- Review-first cleanup
- A cleanup approach that keeps grouping and confirmation visible before deletion, so users understand what they are removing and why.
Duplicate photos vs similar photos
Open guide →- Duplicate photo
- An obvious repeat or copy of the same image, where the cleanup decision is usually lower-risk because one version does not meaningfully differ from the other.
- Similar photo
- A separate image that looks close enough to compete with another shot, even though it is not an exact copy. The user still needs to decide which version is stronger.
- Best shot
- The version in a similar-photo cluster that best preserves what the user actually wants to keep, such as sharpness, expression, framing, or moment.
- Burst-heavy cluster
- A group of photos captured close together where the difference between shots is small enough that the user still needs a review-first workflow.
- Low-risk cleanup
- A cleanup category where the user can move faster with less fear of deleting something irreplaceable. Exact duplicates and screenshots usually feel lower-risk than similar shots.
AI photo editing glossary
Open guide →- AI photo editor
- A tool that changes or generates portrait variations from a source image, often for professional photos, profile pictures, stylized edits, or creative transformation.
- Source portrait
- The original image used to guide the edit. Sharp facial detail, visible eyes, and stable lighting usually make the source portrait more flexible and reliable.
- Identity preservation
- The degree to which the edited image still feels like the same person rather than a synthetic substitute. This is the key trust metric for profile pictures and headshots.
- Realism
- How believable and socially usable the final portrait feels. A realistic result usually keeps expression, facial structure, and presentation grounded enough for real-world contexts.
- AI professional photo
- A work-facing portrait edited or generated to fit resumes, LinkedIn, team pages, founder bios, and other professional profile surfaces.
- Age transformation
- An edit that moves the portrait toward an older or younger direction while trying to keep the same identity readable in the result.
- Style variation
- A controlled change in wardrobe, background, lighting, polish, or creative direction that produces several usable portrait outcomes from one source image.
- Profile-picture workflow
- The sequence of choosing a source portrait, defining the intended use case, generating or editing variants, and selecting the result that feels most believable for the target surface.
Behind the definitions
The data is open source
These definitions are grounded in real measurement, not opinion. The benchmarks and datasets behind them are public on GitHub, so anyone can check the numbers.
All Cleanor repositories
Open code, open data
The image library behind our on-device tools, the storage datasets behind our research, and a zero-auth MCP server for AI builders, all public under one organization.
FAQ
Common questions
Quick answers to the definitions people ask about most.
What is a phone cleanup glossary for?
It defines the words behind storage cleanup so decisions get easier. Once terms like duplicate photos, similar photos, and cleanup order are clear, choosing the right feature or app becomes obvious.
What is the difference between duplicate and similar photos?
Duplicate photos are obvious repeated copies of the same image, which makes them low-risk to remove. Similar photos are separate shots that look close enough to compete, so you still have to decide which version is worth keeping.
Where should I start cleaning up storage?
Cleanup order matters. Starting with heavy files and low-risk clutter, large videos, old screenshots, and clear duplicates, recovers space fastest and with the least stress before you reach harder judgment calls.
Is the data behind these definitions public?
Yes. The storage benchmarks and datasets that inform these concepts are open source in the cleanor-storage-lab repository on GitHub, so the numbers behind the terms are reproducible.