Passes the sniff test
I’m usually skeptical of privacy claims like these, but this app impressed me. I used a cellular data monitor to check where it connected, when it transferred data, and how much it transferred.
On first launch, it connected to a CDN and downloaded roughly 15 MB. That appeared to be the preview templates shown on the startup screen. I also observed about 100 KB of traffic involving RevenueCat and Firebase.
During testing, switching away from the app to check the monitoring tools consistently triggered another 20–30 KB transfer to app-analytics-services.com. The pattern resembled Firebase analytics traffic I’ve observed in other apps using event analytics. The privacy policy mentions collecting telemetry data, although the destination and transfer size alone don’t establish exactly what was being sent (just app state, or meta data too?).
What impressed me most was that transfer sizes stayed consistent regardless of the images I used. I tested very large and very small images, along with complex and simple ones, and saw no corresponding increase in traffic. That supports the claim that media stays on the device since a variable payload size could indicate different results from image compression. It doesn’t conclusively rule out tiny, fixed-size thumbnails or other image-derived data, but I saw no evidence of full images (or even useful size thumbnails) being uploaded.
I also left the app running in the background for several days and saw no unexpected network activity.
One thing to watch for: if you use iCloud Photos with “Optimize Storage” enabled, selecting an image may require downloading its original from iCloud. In my testing, that could cause a data spike attributed to the app requesting the image. That download shouldn’t be mistaken for the app uploading your photo.
The app worked in airplane mode with images already stored locally. The only failures I encountered were when selecting photos whose originals were still in iCloud. It also worked normally with cellular data enabled on the phone but disabled specifically for this app.
Overall, the privacy claims appear consistent with what I observed, and the results were impressive. I’d like the privacy policy to explicitly explain which telemetry events are collected, what information they contain, and how long that information is retained. Beyond that, top marks!
I’m excited to see what other on-device AI projects this team develops.
On first launch, it connected to a CDN and downloaded roughly 15 MB. That appeared to be the preview templates shown on the startup screen. I also observed about 100 KB of traffic involving RevenueCat and Firebase.
During testing, switching away from the app to check the monitoring tools consistently triggered another 20–30 KB transfer to app-analytics-services.com. The pattern resembled Firebase analytics traffic I’ve observed in other apps using event analytics. The privacy policy mentions collecting telemetry data, although the destination and transfer size alone don’t establish exactly what was being sent (just app state, or meta data too?).
What impressed me most was that transfer sizes stayed consistent regardless of the images I used. I tested very large and very small images, along with complex and simple ones, and saw no corresponding increase in traffic. That supports the claim that media stays on the device since a variable payload size could indicate different results from image compression. It doesn’t conclusively rule out tiny, fixed-size thumbnails or other image-derived data, but I saw no evidence of full images (or even useful size thumbnails) being uploaded.
I also left the app running in the background for several days and saw no unexpected network activity.
One thing to watch for: if you use iCloud Photos with “Optimize Storage” enabled, selecting an image may require downloading its original from iCloud. In my testing, that could cause a data spike attributed to the app requesting the image. That download shouldn’t be mistaken for the app uploading your photo.
The app worked in airplane mode with images already stored locally. The only failures I encountered were when selecting photos whose originals were still in iCloud. It also worked normally with cellular data enabled on the phone but disabled specifically for this app.
Overall, the privacy claims appear consistent with what I observed, and the results were impressive. I’d like the privacy policy to explicitly explain which telemetry events are collected, what information they contain, and how long that information is retained. Beyond that, top marks!
I’m excited to see what other on-device AI projects this team develops.
Show more
T:Tested :)
好用的软件
十分喜欢这个软件
Zzhaisi
Worthless
I paid a lot for this app and it’s useless because it only allows you to replace perhaps 1/4 of the face. I bought the premium yet it still has a button in the app that says “upgrade to premium”.
R1Random person 11122233
好用的软件
十分喜欢这个软件
Zzhaisi
Response from developer
很高兴你能喜欢我们的产品，谢谢你的使用。
Meh - seems to be limits and poor detection
Arbitrary length limit that seems to prevent face detection after a certain amount of time.
Mmrice99
