Newbie question - Can we only view Mapillary images in OSM, but can't add them to OSM?

I am very new to how these tools work.

I was hoping to drop some panoramic photos onto a map, to increase the coverage of a outdoor adventure park. Essentially to allow people to have a virtual look around, before actually going out there, since its a reasonable distance from the city.

But I then found that Mapillary doesn’t allow us to tweak the location of the images. Nor does it let us rotate panoramas so they point in the correct direction.

I looked at OSM, and it had a Mapillary plugin, so I was trying to use that to view and change the images in OSM.

But I am getting the strong impression that I have the wrong idea. And that Mapillary images will never appear directly on OSM when viewing. (Editing, yes of course)

And since Mapillary images are uneditable for location / rotation, Mapillary is useless for me and anyone who doesn’t have very accurate GPS logger and a compass that is accurate.

And of course I very likely have the wrong idea about Mapillary.

Mapillary_tools are very versatile. Tweaking of the image position and direction is easily done prior to processing/uploading. The simplest way is to edit the relevant GPS tags in the image EXIF, check their position (say) as an overlay on Viking/OSM, them process/upload them.

AFAIK there is no facility to change position/direction information once uploaded. It is however possible to delete your upload via the Mapillary GUI and upload the same (modified) etc set.

Keep in mind that the standard OSM viewing GUI is only one way to display the OSM information. If there is an image link within the OSM node then with suitable coding it can be made visible. I don’t personally know of any other website etc that does this automagically but is worth a look.

See Photo linking - OpenStreetMap Wiki

1 Like

My starting position when investigating how to add photos to maps was that I wanted to capture a limited number of single 360 images in a location, at particular points. This was based on previous limited experience with 360 photos and creating simple walk throughs. Or Google Street View with photospheres. (It appears that photos uploaded to Google Maps as part of a location now ignore the GPS data and don’t create photo spheres. Ridiculous.)

Part of my learning has been that these street view tools are expecting long sequences of images so they can merge them together into “walkable” street view maps.

This may well be an option for this little project, but I’m trying to keep it simple so that volunteers can have an easier time if they just want to add a few more images later.

And it seems like none of the street view tools (Mapillary, Panoramix) allow uploading a number of images, and having an option to have them as separate images, not part of a sequence. This would allow them to appear as image hot spots on a map.

It is not at all obvious what external tools to try to tweak the position and direction data. You mention Viking, which I’ll have a look at. As one of about 10 I’ve looked at.

Given that Mapillary seems to have no desire to enable image position / direction editing etc, it would be REALLY helpful to have a section of the help that gives viable 3rd party options for altering image / gpx metadata. I expect that people who have been working in this space for a bit know of some options.

But those of us coming in new would benefit greatly by some pointers in that direction.

I managed to find JOSM, that lets us import images with GPS data, then (using some plugins) we can move them on the map to a more correct position. And change their direction. Then write that data back into the jpgs. Which yes, can then be imported back into Mapillary as a sequence.

This actually would work, except I can’t make all the images separate, not as part of a sequence. Well, actually, a sequence would be fine, if I could hide the lines on the map between each image. Is that possible?

Cheers

I would suspect that Google may have reigned in what may have the chance of violating GDP. Just an opinion!

Once again the mapillary_tools (CLI) creates photo sequences, so theoretically one could create specific route walk throughs by entering the right spacing/distance/angle limits. I have in the past uploaded single sequence images because they exceeded the default limits, although in my case it wasn’t actually intended.

These are the process parameters that affect sequence splitting and NOT regarding close images as duplicates. This is a rather old help dump so please check the current version

–cutoff_distance CUTOFF_DISTANCE
Cut a sequence from where the distance between
adjacent images exceeds CUTOFF_DISTANCE. [default:
600.0]
–cutoff_time CUTOFF_TIME
Cut a sequence from where the capture time difference
between adjacent images exceeds CUTOFF_TIME. [default:
60.0]
–duplicate_distance DUPLICATE_DISTANCE
The maximum distance that can be considered “too
close” between two images. If both images also point
in the same direction (see --duplicate_angle), the
later image will be marked as duplicate and will not
be upload. [default: 0.1]
–duplicate_angle DUPLICATE_ANGLE
The maximum camera angle difference between two images
to be considered as heading in the same direction. If
both images are also close to each other (see
–duplicate_distance), the later image will be marked
as duplicate and will not be upload. [default: 5.0]

I don’t do any any image repositioning, instead always using GPS data. I have however used Viking to validate my local EXIF processing (for camera shutter delay) by overlaying the image on the OSM map at road intersections. I know Google Earth also allows georef’d image overlays. I don’t know the license/reuse limitation of that, but I suspect the image could be overlaid on the overhead imagery and the view/co-ords copied/pasted as needed. I get the impression Viking will do similar with background georef’d overhead imagery. There have however been posts in this forum about manual repositioning so I hope someone else will chime in.

I get the impression that although what you are suggesting might be good and useful, it isn’t really what Mapillary was created for. (Might be good if @boris would chime in here) Help in that respect then will come from other users doing uncommon things rather than from formal help. Also keep in mind that such data as altitude and image tilt angles are either unreliable or not recorded by Mapillary. Indeed “street level +1-2m and horizontal POV” is always assumed.

I don’t know/use JSOM at all, but if you can create/modify the images then mapillary_tools could be used separately to upload them, for example using (say) 1 (metre) rather than the default 600 for cutoff_distance

Hope this helps

Dear @JokerNZ,

One way to place and rotate in the correct viewing direction is using the JOSM OSM editor with two plugins, photoadjust and photo_geotagging - see GPS location correction - #2 by koninklijke and SOLVED : Drag&drop to adjust geo-location? JOSM works best for me, after trial-and-error, GPXEV and GeoSetter are good alternatives - turns out you’re not the first to ask.

Best, and don’t hesitate to ask is anything is ‘as clear as mud’

Thank you @bob3bob3, @koninklijke for your input.

As I looked at Mapillary and other similar tools, I realised that they weren’t what I thought they were.

I understand now that Mapillary is very much about sequences of “connected” images. As in, Mapillary is able to connect them together into walkable sequences.

And part of the point is to have many images along a route, so that Mapillary tools and other tools (like OSM) that use Mapillary can extract a lot of data.

And that its deliberately got limited ability to change GPS etc in Mapillary because the data that goes in needs to be assumed to be good data.

Hence, people like me wanting to change it after the fact is probably not a good idea.

Its quite a difference use case from what I was expecting, and in many ways, what I need.

Thanks for the tips about the CLI uploader. I am not against getting down into a CLI.

What I’m trying to do is have a reasonable solution for a group of volunteers (who really aren’t technical) to be able to either update some of their own information, or get someone a bit more technical to update their information.

Basically, I’m trying to give them something that is practical for their use, that doesn’t put a burden on their website, and doesn’t leave them totally stranded if I can’t help them with this in the future.

Having said all that, now that I have a better idea of what Mapillary is, I could imagine creating a whole set of sequences to upload into Mapillary, which would give them a nice walkthrough…

At least, it seems that way.

1 Like