There are times when – for reasons of privacy or even a person’s physical safety – you want to make certain parts of a frame in a video unrecognizable so not to give away someone’s identity or the place where you shot the video. While it’s fairly easy to achieve something like that for a photograph, it’s a lot more challenging for video because of two reasons: 1) You might have a person moving around within a shot or a moving camera which constantly alters the location of the subject within the frame. 2) If the person talks, he or she might also be identifiable just by his/her voice. So are there any apps that help you to anonymize persons or objects in videos when working on a smartphone?
KineMaster – the best so far
Up until recently the best app for anonymizing persons and/or certain parts of a video in general was KineMaster which I already praised in my last blog about the best video editing apps on Android (it’s also available for iPhone/iPad). While it’s possible to use just any video editor that allows for a resizable image layer (let’s say just a plain black square or rectangle) on top of the main track to cover a face, KineMaster is the only one with a dedicated blur/mosaic tool for this use case. Many other video editing apps have a blur effect in their repertoire, but the problem is that this effect always affects the whole image and can’t be applied to only a part of the frame. KineMaster on the other hand allows its Gaussian Blur effect to be adjusted in size and position within the frame. To access this feature, scroll to the part of the timeline where you want to apply the effect but don’t select any of the clips! Now tap on the “Layer” button, choose “Effect”, then “Basic Effects”, then either “Gaussian Blur” or “Mosaic”. An effect layer gets added to the timeline which you can resize and position within the preview window. Even better: KineMaster also lets you keyframe this layer which is incredibly important if the subject/object you want to anonymize is moving around the frame or if the camera is moving (thereby constantly altering the subject’s/object’s position within the frame). Keyframing means you can set “waypoints” for the effect’s area to automatically change its position/size over time. You can access the keyframing feature by tapping on the key icon in the left sidebar. Keyframes have to be set manually so it’s a bit of work, particularly if your subject/object is moving a lot. If you just have a static shot with the person not moving around a lot, you don’t have to bother with keyframing though. And as if the adjustable blur/mosaic effect and support for keyframing wasn’t good enough, KineMaster also gives you a tool to add an extra layer of privacy: you can alter voices. To access this feature, select a clip in the timeline and then scroll down the menu on the right to find “Voice Changer”, there’s a whole bunch of different effects. To be honest, most of them are rather cartoonish – I’m not sure you want your interviewee to sound like a chipmunk. But there are also a couple of voice changer effects that I think can be used in a professional context.
What happened to Censr?
As I indicated in the paragraph above, a moving subject (or a moving camera) makes anonymizing content within a video a lot harder. You can manually keyframe the blurred area to follow along in KineMaster but it would be much easier if that could be done via automatic tracking. Last summer, a closed beta version of an app called “Censr” was released on iOS, the app was able to automatically track and blur faces. It all looked quite promising (I saw some examples on Twitter) but the developer Sam Loeschen told me that “unfortunately, development on censr has for the most part stopped”.
PutMask – a new app with a killer feature!
But you know what? There actually is a smartphone app out there that can automatically track and pixelate faces in a video: it’s called PutMask and currently only available for Android (there are plans for an iOS version). The app (released in July 2020) offers three ways of pixelating faces in videos: automatically by face-tracking, manually by following the subject with your finger on the touch-screen and manually by keyframing. The keyframing option is the most cumbersome one but might be necessary when the other two ways won’t work well. The “swipe follow” option is the middle-ground, not as time-consuming as keyframing but manual action is still required. The most convenient approach is of course automatic face-tracking (you can even track multiple faces at the same time!) – and I have to say that in my tests, it worked surprisingly well!
Does it always work? No, there are definitely situations in which the feature struggles. If you are walking around and your face gets covered by something else (for instance because you are passing another person or an object like a tree) even for only a short moment, the tracking often loses you. It even lost me when I was walking around indoors and the lens flare from the light bulb at the ceiling created a visual “barrier” which I passed at some point. And although I would say that the app is generally well-designed, some of the workflow steps and the nomenclature can be a bit confusing. Here’s an example: After choosing a video from your gallery, you can tap on “Detect Faces” to start a scanning process. The app will tell you how many faces it has found and will display a numbered square around the face. If you now tap on “Start Tracking”, the app tells you “At least select One filter”. But I couldn’t find a button or something indicating a “filter”. After some confusion I discovered that you need to tap once on the square that is placed over the face in the image, maybe by “filter” they actually mean you need to select at least one face? Now you can initiate the tracking. After the process is finished you can preview the tracking that the app has done (and also dig deeper into the options to alter the amount of pixelation etc.) but for checking the actual pixelated video you have to export your project first. While the navigation could/should be improved for certain actions to make it more clear and intuitive, I was quite happy with the results in general. The biggest catch currently is that the maximum export resolution is 720p. I wouldn’t say it’s a deal breaker per se as 720p is still an acceptable resolution for certain use cases, particularly if you publish something to the web/social media. But 1080p would definitely be better. I talked to the developer and he/she said that they will try to add 1080p export in one of the next updates. I suppose the face-tracking feature is a computing hog and sifting through 1080p footage might (still) be somewhat challenging to not-so-powerful Android devices. But I’m sure current flagship processors like the Snapdragon 855/865 or Samsung’s recent Exynos SoC’s should be able to handle tracking in FHD. An additional feature that would be great to have in an app that has a dedicated focus on privacy and anonymization, is the ability to alter/distort the voice of a person, like you can do in KineMaster.
There’s one last thing I should address: The app is free to download with all its core functionality but you only get SD resolution and a watermark on export. For 720p watermark-free export, you need to make an in-app purchase. The IAP procedure is without a doubt the weirdest I have ever encountered: The app tells you to purchase any one of a selection of different “characters” to receive the additional benefits. Initially, these “characters” are just names in boxes, “Simple Man”, “Happy Man”, “Metal-Head” etc. If you tap on a box, an animated character pops up. But only when scrolling down it becomes clear that these “characters” represent different amounts of payment with which you support the developer. And if that wasn’t strange enough by itself, the amount you can donate goes up to a staggering 349.99 USD (Character Dr. Plague) – no kidding! At first, I had actually selected Dr. Plague because I thought it was the coolest looking character of the bunch. Only when trying to go through with the IAP did I become aware of the fact that I was about to drop 350 bucks on the app! Seriously, this is nuts! I told the developer that I don’t think this is a good idea. Anyway, the amount of money you donate doesn’t affect your additional benefits, so you can just opt for the first character, the “Simple Man”, which costs you 4.69€. I’m not sure why they would want to make things so confusing for users willing to pay but other than that, PutMask is a great new app with a lot of potential, I will definitely keep an eye on it!
As always, if you have questions or comments, drop them below or hit me up on the Twitter @smartfilming. If you like this article, also consider subscribing to my Telegram channel (t.me/smartfilming) to get notified about new blog posts and receive the monthly Ten Telegram Takeaways newsletter about important things that happened in the world of mobile video.
For an overview of all my blog posts click here.