How I Cleaned Up the Thousands of Photos and Videos I Had Scattered Across the Internet

After years of treating cloud storage like an infinite digital attic, I found myself paying for five different subscriptions and still not knowing where half my family photos lived. The breaking point came when I tried to find a specific birthday video from 2018 and spent three hours digging through Google Drive, iCloud, Dropbox. And two forgotten external drives. It wasn't just annoying-it was expensive, insecure, and utterly unsustainable. Here's how I untangled the mess and cut my monthly storage bill by 70%. And why the process taught me more about digital hygiene than any best-practices guide ever could.

The hard truth: most people are paying for redundant storage they don't need. And their most precious memories are scattered across services with no single source of truth. In my case, I discovered 1. 2 TB of identical Photos duplicated across platforms, along with 400 GB of corrupted or irrelevant files that had been backed up and never reviewed. What follows is a step-by-step blueprint for anyone drowning in digital clutter-built on actual command-line tools, metadata forensics. And a painful but rewarding audit of every single file.

Assessing the Damage: Why You Have More Copies Than You Think

Before I could clean, I needed to map every copy of every file. This wasn't just about counting duplicates; it was about understanding the topology of my digital life. I started by listing every cloud service I'd ever used: iCloud (since 2011), Google Photos (2015), Google Drive (2016), Dropbox (2013), OneDrive (2018). And an old Amazon Photos account from a Prime subscription. Each had been set to auto-upload from my phone at different times, overlapping with itself and with local backups on two external SSDs and an old laptop's hard drive.

The real insight came when I ran a script to tag every file with its origin timestamp, not just the filesystem date. ExifTool (official ExifTool documentation) became my best friend. By extracting the original capture date from JPEG and HEIC metadata, I could order photos chronologically regardless of where they'd been uploaded. This revealed a startling pattern: between 2015 and 2020, I had uploaded the same set of vacation photos to three different services within a span of weeks, each time thinking I was finally being organized. In reality, I was multiplying the problem.

Auditing Subscription Costs: The Financial Wake-Up Call

I pulled up every subscription bill from the last 12 months iCloud 2TB ($9. 99/mo), Google One 2TB ($9. 99/mo), Dropbox Plus 2TB ($11. 99/mo), OneDrive 1TB bundled with Office 365 ($99. 99/yr), plus a 5TB external HDD I'd bought "just in case" ($130). The total came to roughly $540 per year, or $45 per month. And for whatTurns out, less than 800 GB of unique, non-duplicated photos and videos. The remaining 3. 2 TB across all services was pure redundancy, corrupted files. And junk (screenshots, memes, accidental burst shots that no one would ever view).

The math was undeniable: I could move everything into a single object storage bucket and a cheap NAS at home, dropping the monthly to around $15. But first, I had to know exactly what I was keeping. That required a deliberate, multi-pass deduplication strategy-not a one-click "duplicate cleaner" app. Which often misses near-duplicates and metadata variations.

Deduplicating at Scale: Tools That Uncovered Hidden Pairs

Simple file-hash duplicate checkers (like DupeGuru, DupeGuru official site) are great for exact copies. But most of my duplicates weren't byte-for-byte identical. Google Photos had recompressed my 12MP JPEGs into WebP. While iCloud had kept the original. Dropbox had duplicated some files during sync conflicts, appending " (1)" to filenames. I needed a three-tier approach:

• Tier 1 - Exact byte-level hash (SHA-256): Ran on every local and downloaded cloud file. Caught 40% of duplicates-mainly sync artifacts and re-uploads.
• Tier 2 - Perceptual hashing (pHash): For photos resized or recompressed. Tools like vImage` in Swift or Python's imagehash` library flagged near-duplicates with the same subject but different dimensions.
• Tier 3 - Date + size heuristics: Videos longer than 5 minutes with identical capture timestamps and file sizes (within 1%) were marked as strong duplicates, even if codecs differed.

I wrote a small Python pipeline that combined these passes and output a JSON report. The report showed that 62% of my 14,000+ items were likely unnecessary copies. That's 8,680 files I could safely delete-if I was careful.

Metadata Purification: The Overlooked Cleanup Step

Before consolidating, I realized that many files had incorrect EXIF data? Some older photos had no date at all (defaulting to 1970-01-01). And others had been stripped of location data by sync services. This would break any chronological organization system later. I used ExifTool in batch mode to write the original capture date back into JPEG files that lacked it, pulling the timestamp from the filename pattern I'd used on my phone (IMG_YYYYMMDD_HHMMSS).

For videos, MP4 metadata is trickier. I relied on FFmpeg to extract creation time from the QuickTime metadata atom if available. And fall back to the file's birth timestamp when that was missing. I also stripped GPS coordinates from photos I didn't want geotagged (like indoor shots) and kept them only for travel images. This took two evenings but was essential for a usable final archive.

Consolidation Strategy: One Source of Truth (and a Backup)

After deduplication and metadata cleanup, I moved everything to a single cloud storage bucket (Backblaze B2) with lifecycle rules to automatically archive older photos to cold storage after 90 days. At the same time, I set up a Synology NAS at home running a synchronized copy via rsync. This gave me three copies (local, B2 hot, B2 cold) but only one active master. The rule: any new photo or video goes only into one specific folder on the NAS. And then the NAS syncs up to B2. No syncing from phone straight to cloud-that's how the duplication started.

I also wrote a small script that watches that folder and runs ExifTool + pHash daily to catch any accidental re-uploads before they proliferate. It's not glamorous. But it's the only way to prevent the scatter from creeping back in.

The Cost Reduction: From $540

After cleaning up, my monthly subscription costs dropped from $45 to $12. Here's the breakdown:

• Canceled iCloud 2TB → use only the free 5GB (for device sync only)
• Canceled Google One 2TB → use free 15GB for email/docs
• Canceled Dropbox Plus 2TB → no subscription needed
• Kept Office 365 Family (already paid for, includes 1TB OneDrive but now unused for photos)
• Added Backblaze B2: ~$6/mo for 300GB of unique data (after dedup)
• Added NAS electricity and hard drive depreciation: ~$6/mo

Total saved: ~$400/year. More importantly, I now have a single, searchable, backup-tested archive. When I need a photo from 2016, I can find it in under 30 seconds.

Lessons Learned: Why Digital Hoarding Is a feature, Not a Bug

The most valuable lesson wasn't technical-it was psychological. Cloud storage providers intentionally make it easy to upload and hard to prune. Auto-upload from a phone is a convenience that turns into a trap. I now set a manual upload schedule: photos stay on my phone for a month, then I move them to the archive during a dedicated weekly review. That review also forces me to delete the 90% of screenshots and blurs that no one needs.

Another insight: metadata is the easiest thing to lose. WebP conversions, platform-specific compression, and accidental edits can strip EXIF data silently. Always keep an original, unmodified copy of your camera's output. In my case, I re-downloaded the original JPEGs from Google Photos' "Takeout" feature (using the "original quality" option) and compared their hashes against what I had locally. The mismatches turned out to be Google's recompression-I deleted the local compressed versions, keeping only the originals.

Future-Proofing: Automating the Cleanup Routine

I now run a monthly maintenance script that:

• Scans the master archive for new duplicates (cross-referencing hashes)
• Reports files older than 90 days that haven't been marked as reviewed
• Generates a list of broken or non-playable video files using FFmpeg's error-checking
• Validates that the NAS rsync job completed without mismatches

This routine isn't technically hard. But it requires discipline. I wrote the script in Bash and Python and scheduled it via cron. The output goes to a Slack channel only I can see-a small daily reminder that digital hygiene is an ongoing practice, not a one-time project.

Frequently Asked Questions

1. What's the best deduplication tool for photos? For exact copies, use DupeGuru or a custom SHA-256 script. For near-duplicates (recompressed or resized), use perceptual hashing libraries like imagededup (Python) or imagehash. Never trust a tool that doesn't let you preview files before deleting.
2. How do I safely export photos from Google Photos? Use Google Takeout (takeout, and googlecom). Choose "original quality" to avoid recompression, since expect months of files if you have many-be patient and expect duplicates.
3. Should I keep multiple cloud subscriptions for redundancy? No. That's expensive and confusing. Use one storage provider for the active copy. And a second service (or local backup) for disaster recovery. The 3-2-1 rule works with two physical locations, not two cloud bills.
4. What's the best way to organize photos without losing metadata? Keep the original folder structure from your camera (e g., DCIM/100MEDIA), but also use a tool like DigiKam or PhotoStructure that reads dates from EXIF and files them into YYYY/MM for easy browsing. Avoid renaming files unless you preserve the original timestamp in the name.
5. How do I know if a video file is corrupted? Use FFmpeg with -v error -i file, and mp4 -f null -It will print errors for corrupt frames or incomplete headers. Batch check with a script that logs failures.

Conclusion: You Don't Need More Storage, You Need Better Storage

Cleaning up my scattered photos wasn't just about saving $400 a year-it forced me to confront how I consume and preserve digital artifacts. The tools exist, the methodologies are proven,, and and the cost savings are realBut the hardest part is starting: you have to admit that your current system is broken, even if it feels organized. Take one afternoon to map your subscriptions, run one hash scan. And commit to a single source of truth. Future you will thank you when searching for a memory takes seconds, not hours.

Ready to take control? Start with a spreadsheet of all your storage services and billing. Then download ExifTool and a duplicate finder. The next step is just opening the first folder,

What do you think

Was the pain of cleaning up thousands of files worth the $400 annual savings,? Or would you rather just pay for convenience and ignore the clutter?

Is it possible for a non-technical person to follow a similar metadata-based cleanup,? Or does this approach remain the domain of engineers with terminal access?

Given that cloud providers actively encourage hoarding through auto-upload, do you think they have an ethical responsibility to help users prune duplicates and reduce costs?