All your photos. Compressed with AI.

PixelDrive uses AI to compress photos down to as little as 10% of their original size without compromising on quality or resolution. Users can link their existing Dropbox storage account (with Google Drive coming soon), and compress all photos stored on those platforms.

Would you recommend this product?
6 Reviews5.0/5
10% ? Next Pied Piper 😅
Impressive numbers, 10% with no quality loss! 👀
So I tried this. I upscaled my photo using Upscalepics.com up to 8 times, got to about 1 MB in size, and then shrank it using PixelDrive and got it down to 511 KB only. Can't see any loss in quality. So the both work amazingly well together

The design needs some serious improvement both visually (too cluttered with useless stuff) and for the interaction (verrrrrry tedious on my iPhone). I see potential, maybe this would be best as an API rather than they providing the dashboard for you.


The few images I tested appeared absolutely identical between the compressed and original versions, even upon very close inspection.


The user interface is SO bloated. Why is there even a “Contacts” area on a photo-storage service?!

Upvote (1)Share
@jeffreybennett This keeps feeling more and more like Pied Piper XD.
Thanks @chrismessina for the hunt! PixelDrive uses AI to compress photos down to as little as 10% of their original size with virtually no change in quality or resolution. To be clear, it is definitely a lossy algorithm, but the cool thing about AI is that it learns what information it can remove from each individual pixel without compromising the overall structural integrity of the image. We first compress to our own format (called MLVX, for Machine Learning Visual Extension), but then apply JPEG on top of it to maintain compatibility with everything else that is out there. The advantage of our own format is that it is able to reproduce a super high-fidelity version of the original image when decoded, because it compresses without any reduction in resolution or quantization of colors. We'd appreciate any feedback you have!
@fdoumet what about other emerging formats like webp and webm? Or HEIC?
@chrismessina Great question! We mainly use MS-SSIM (which measures the perceived quality of digital pictures) to determine the quality of the compressed image. For the same bitrate (# of bits per pixel) of the output image, we beat all existing formats