Mobile apps: how we stopped using them

[Richlv]

Heya all, I’ve been away from Mapillary forums for a while.
One of the reasons was that the mobile applications seemed to change in ways that didn’t match my usage patterns, and what seemed as obvious usability shortcomings were not improved.

This is primarily from experience with the iOS app, but the pattern could be the same on Android.
I still use the app on iPhone6 - it is stuck on some older version before the redesign… and it has kept that application more acceptable for me. But that device will die very soon, and with it would die my Mapillary app usage.

But why did I write “we”?
After the application changes, we had several OpenStreetMap events. In the past, I have always suggested that participants use Mapillary to take images of streets, footways, shops and anything else during the event. Some of the participants had been to previous events, and they tried to use Mapillary again on their own (as I had not recommended it anymore)… but found it to be too convoluted now. One actually asked me whether it was the same app, because they found it to be so different and not quite usable anymore. Across multiple events, I met three people with the same concerns, and they all decided that Mapillary is not worth the hassle for them.

Will I keep on contributing? Oh yes, definitely. While that old, broken iPhone 6 still works (it runs for 10 minutes on it’s own battery now, so not too long; EDIT: correction, 30 seconds), and with action cameras of course. With 13.4M images uploaded, I have some workflows to capture at least something.

Will I recommend Mapillary in mapping events? Stopped doing that last year. I cannot recommend something I find too much of a hassle myself. This also eliminates any “dedicated” photos from Mapillary - things like storefronts, particular alleys, paths or anything else I would consider useful for both mapping and imagery purposes.

[Richlv]

I’ve also looked for alternative ways to capture images with phone - in case somebody here has recommendations, started that as a new topic at Looking for an alternative capture application .

[filipc]

I am trying the Android native phone app, and OpenCamera, combined with Onedrive. Hints are welcome. I have not studied it yet.
I do not understand, I told Mapillary so many times before that their app now sucks, however nice it looks.
Are we too stupid to use pictures only ?
I use the app for panning while walking a bit.

[Richlv]

Hehe, not sure what usability concerns you had with the apps. I had shared mine (including capture and upload functionality) multiple times, but apparently the app has a completely different target demographic than me - and many people I had suggested the app to earlier.

That’s fine - Mapillary was sold to Facebook/Meta a long time ago, and they can do with the whole project as they please (save for copyright and licensing on the images, as agreed with contributors ).

[boris]

Hi @Richlv thank you for the feedback. Could you summarize what usability concerns you have with the current app? It is hard to build something which works well for everyone, but we’d appreciate hearing your perspective in a little bit more detail.

Best,
Boris, Product Manager, Mapillary

[Richlv]

Oh, sure. I believe I’ve shared that feedback in these forums already, but won’t go searching for those threads now - will list some just off the top of my head.
I’m approaching this from a functional point of view. Things like “screenshots look good in presentations” carry zero weight in this.

a) Auto/manual toggle got removed from the capture screen.
b) …manual mode got completely removed.
c) Extremely cumbersome (nearly impossible with a larger number of images) local save/export.

The above destroy app usage for events like OSM mapping parties, or any other high-added-value capture.

d) Nuked all useful info from the capture screen (captured image count, remaining space, even battery level…).
e) App became much less stable with a large number of images captured (slow operations, frequent crashes).

These on the other hand make any constant capture much more difficult.

There are surely more items, but I haven’t used app much past version 5.6.1.230314, thus not recalling all the ways it got nerfed or could be improved.

[boris]

Thanks for this feedback @Richlv . We agree, no one is trying to create “screenshots that look good in presentations” - we’re trying to build a useful tool for the community to use (and we would like both advanced users and beginner mappers to feel welcome).

For the items you mentioned:

a) the manual capture button is available on the capture screen once you enable it in settings (“allow manual capture”
b) manual mode is available in settings as I mentioned, but it does require you to start a capture first so that manual is “extra” photos on top of the ones that are automatically taken every 3 meters. This was done because Mapillary is predominantly useful for a sequence of captures (more so than individual images), but is your use case more about taking individual images that aren’t connected to each other? Could you explain more?
c) local save/export hasn’t changed recently, but we are making it much easier and faster in an upcoming release coming very soon - stay tuned
d) battery level is available on Android, and coming to iOS in an upcoming release. We’ll also be adding distance of capture (as a metric that might be particularly interesting)
e) overall the app should be more stable now (we added crash tracking and this number has been decreasing) - if you’re seeing a crash we haven’t caught, we’d appreciate a forum thread on that topic

Thank you again for your feedback - it is much appreciated, and we are continually working on (hopefully) improving the platform for everyone.

[Richlv]

Thank you for the reply, will try to cover all the items.

a/b) Yeah, that on-top-of-automatic mode is not useless, but true manual mode was what I used a lot on the phone. Such manual, individual images was a major usecase for me - a few examples:

• Turning and taking photos every 45, 90 or such degrees at intersections etc. Automatic mode takes photos at “whatever” times and is not usable at lower lighting conditions.
• Taking photos of changed objects - for example, I there’s a street that I have imaged every month or two for a while, and new construction starts or a shop changes, I’d explicitly take photos of the changes.
• Taking “high-value” photos - taking images of storefronts, opening hours, buildings, cafes, addresses, paths, streams, steps, notable trees, hydrants and dozens of other categories. These would exclusively be framed to be useful for mapping and non-blurry - those alone would be major reasons not to use automatic mode. Other reasons could include lower battery level, lower available storage, rainy weather, crowded places, heavy traffic, cold weather and probably a bunch more could be added to the list over time.
  Additionally, I’d take such high-value photos (HVP ) manually with the phone wile capturing automatic sequences with the action camera in parallel - and for the past year or two such HVPs do not get to Mapillary at all. The automatic sequences would often not include all that detail.
  Yet another major usecase was OSM mapping parties, where participants would walk in a group and take such HVPs in the manual mode. The experience where other OSM mapping party participants thought it might be the wrong app and shunned the new version was notable.

c) Local save/export didn’t get better (allowing to select and save a larger number of images easily), and I tried the new app a few days ago, and recalled that it actually regressed - now hitting that “upload” icon accidentally nukes all images from being accessible locally. This, again, was useful to save HVPs locally - whether all, or select ones (if the phone was used both for automatic and manual capture). Why? Well, they could sometimes take a very long time to get processed, and they could get all the useful stuff blurred. Saving them locally was extremely useful for mapping parties, and also for normal mapping.
In the past was cumbersome, in the latest versions it’s pretty much a pointless effort.

d) It’s great that battery level is coming back on iOS, but apparently other metrics are not - and it tells a lot that it got removed in the first place.

e) Great that stability is being improved. The version I still use (on iPhone 6) had regressed, compared to older versions, and is crashing often both in capture mode and during uploads.
Still greatly prefer that version over the latest ones, which I haven’t bothered to test/try for stability - not sure whether it would have been better for my usecases.

I guess what was detrimental for motivation was the impression that the app is not designed by somebody with extensive usage experience in different countries, weather conditions, transportation modes and whatnot. Which resulted in usability degradation, in some cases really massively.
As evident by the forum post history, I tried diligently reporting all the problems for a long time, but it’s a lot of effort to donate to Facebook.
I got content with the HVPs not getting on Mapillary and other users skipping it altogether, and I don’t quite see the app improving the highlighted problem areas - will keep on dealing with those photos in other ways.
But definitely not discounting Mapillary completely - if the app would become useful one day again, I’d surely test it and possibly restart using it

[boris]

Thank you so much!

a/b) Great, we’ll add this feature to our backlog and should be able to add this functionality later this year (likely first on Android and then iOS)

c) Thank you for the feedback, we are making local/save export much easier and faster in an upcoming release coming soon (already have the internal build on iOS done) - stay tuned

d) Glad to hear you’ll use the battery level. We’ll be highlighting distance of capture as opposed to number of images because that is a bit more natural for Mapillary looking forward. For example, if a user uploads a video file, how many images is that? We can more naturally think of covering “2 kilometers of roads with fresh Mapillary imagery” as compared to adding 10,000 images (or 1,000 images?).

I do hope you’ll stay tuned and continue providing your feedback especially as new versions are released. We do appreciate it, and are actively trying to incorporate your feedback as well as that of others to make the platform better for all.

[Richlv]

a/b) Thank you so much, that sounds promising. It might be worth stressing that it worked the best many versions ago when the capture screen had a single-tap switch automatic/manual, and it just changed the recording button to start a sequence or take a single photo.
When was this especially useful? For example, when walking in an urban setting, I’d have the automatic mode one, then stop and with two taps take a manual photo of a storefront, then back to automatic, then two-tap photo of a building with a mural etc.
This allowed combining automatic sequences with high-value captures back in… 2016 was it?

c) That also sounds promising, although - has it been designed by somebody who uses the app in this way?
Some teams create specs via a fully open wiki-based process, which often allows to design changes, minimising functional regressions or later rework. Granted, it takes a while to build up community goodwill for productive participation, but it is pretty much free labour…

d) While I can see Mapillary thinking about distance (and no objections to it being displayed), this slightly seems like a design not by “eating own dogfood” method.
If capturing (especially in automatic mode), I care about the number of images / diskspace usage many times more than the distance.
And if contributors care about the number/space, but Mapillary cares more about distance, users might get the latter only. And this will again discourage large scale contributions.
I hear your point about video - while it is not my usecase currently (only used videos years ago with a motorcycle-attached camera) due to static images having much better resolution (very important for actual mapping), some users might prefer it. But even then showing used/remaining space seems like waaaaaay more useful than distance. Sure, users could approximate how much in GB a km is… but why force them to do so, what’s gained that way?

I’m very much looking at this from a contributor point of view. I used to capture 55GB of images per day. With a phone. With Mapillary app. What mattered was diskspace used/remaining, battery level.
Distance? While capturing, my level of interest in it was probably 1 out of 10.
And the app actually handled all that. On iPhone 6. Last version that still works on it starts hanging/crashing with 1/4 of that. As you can guess, that did reduce my phone-based contributions already, as some days I could capture only for the first few hours, and on other days I was so pissed by the app not working, I stashed the phone away to save the day

I guess overall 3 things matter a lot:

1. For whom the app is designed. Is it designed only for car-mounted, first time users that will capture 50 images?
2. Who designs it. Have the people writing specs used the app to capture images 12 hours per day, used it in -20, used it at night time, used it to export some or all images for immediate OSM mapping? While that might seem like a small audience, if the app can support these scenarios, it will be recommended more, and it will surely work for newbies. And the old versions were decent/passable in all these aspects.
3. Is it seen as a zero-sum design. That is, if it is thought that displaying the number of images will confuse a newbie way too much, the app is doomed. Extreme minimalism has been very visible for the past 10 years, and it has sometimes resulted in a flight of users (I gave up on one opensource desktop application which I used daily for years, and donated a lot of time supporting other users, writing docs etc).
   Functionality, if well designed, will not scare away newbies. On the other hand, removing functionality is guaranteed to lose some more dedicated users.
   Imagine, if you will, a default photo app on phones. Now imagine developers deciding that having video feature is too confusing, so it’s gone now. Also image editing, that’s soooo not newbie friendly, out it is. Setting focus/exposure point or anything else related to those params? Already gone last week. Flash control pollutes the interface, so off with its head. That’s what the current Mapillary app feels like

Sorry about the long message, hopefully some parts of it are useful enough - and I will definitely try out some upcoming Mapillary versions. If the app improves, will try further versions as well, maybe it gets back to useful state for my usecases, keeping a positive attitude

[boris]

Thank you again for taking the time to share your detailed feedback. We hear you - and while we probably won’t be able to make everything you are looking for happen (because there are tradeoffs to be had), we hear your feedback and hope you’ll see improvements you appeciate in upcoming releases . Thank you for your patience and continued feedback.

[Richlv]

Thanks, always great to hear about improvements planned.
The part about “probably won’t be able to make everything you are looking for happen” does sound a bit ominous (as in “dream about it”), I’m approaching this as a potential bonus - as I’ve already dropped the usage of the Mapillary app, getting it back in a usable shape would be a bonus. If those improvements are not available, it’s just that the current state will continue.

Although it is slightly sad every now and then - for example, somebody in another city offered to send photos of some buildings where tenants and usage have changed. In the past I would have offered them to do this via Mapillary app (and upload photos to Mapillary in the process).
Currently I dread even explaining the necessary steps to them, and even if I did explain the steps, it’s almost guaranteed they would tell me to go and stuff myself - thus the outcome is the same as not suggesting Mapillary

[filipc]

I use the Mapillary app to be on the safe side.
I place my thumbs on the capture button and on the little cross up left.
After the first picture I immediately press the cross with my left thumb.
Then I do a mobile data upload.

Then I take quality pictures of the environment and the object with Open Camera.

[Yaro]

Hi @filipc, we are working on adding the way to capture sequence manually only, without starting automatic capture. Also, at the open testing version has the ability to save images to shared storage (images you can browse in your default directory explorer app). It looks for me like these two can help to improve your experience with the app.

BR, Yaro

[filipc]

I am not easy to satisfy.

[Richlv]

Saw the new major app update, tried it.
Looks like none of the removed/missing features has returned so far.
While fiddling around and trying to save locally the capture, tapped on the button that looked like “share”, which immediately started the upload (confirmation has been removed), and thus locked the sequence “for upload”.

At this pace, it seems more likely that images on Mapillary could arrive from a Panoramax mobile app, if it appears one day

[boris]

Hi @Richlv - thanks for trying the new app update and your feedback!

To locally download images in the iOS app tap the “Select” button in the image gallery and then select the images you’d like to save and tap the save button in the bottom right. Here’s a screenshot:

Let me know if that’s working for you.

[Richlv]

Thank you for the reply, appreciated. I did figure out that it works this way, but mostly from using the functionality before.
Changing text to icons in some way is understandable, but it does make the app notably more confusing - the icons and the functionality is quite far from standard or immediately intuitive.

In this particular case, the worst experience was from the removed upload confirmation + upload locking the sequence.
Older app versions did not lock for upload, and all sequences & images were accessible even if upload had been started before.

[4004]

To me, the key issue with the app has always been the requirement to use it (for a mobile workflow) - you had to open it, wait for gps lock, wait for compass to get a bearing (almost never), wait for the phone to process everything etc. There is much more friction in this process compared to just a snap using the Camera.app.

I’ve asked for simple “upload from camera roll” functionality a few times, since probably before 2020, but I imagine Mapillary get more data/value out of whole sequences, which is why this functionality was never added and probably never will be

[boris]

@4004 - for uploading from Camera roll, we recommend the Desktop Uploader - which is designed for this function - taking images or video you shoot outside the app and uploading it using your fast computer/Wi-Fi connection.