OpenStreetMap
DigitalGlobe, now Maxar [1], is transitioning the OpenStreetMap imagery endpoints (from “DigitalGlobe Standard” to “Maxar Standard”; and from “DigitalGlobe Premium” to “Maxar Premium”). OSM users should see content parity between the old and the new layers by June 30, 2019. Starting July 1, 2019, there will be only two Maxar layers available from within the OSM editors – labeled with Beta.
Our new imagery services use http/2 protocol. Please note that there are some known speed issues with the JOSM editor. A JOSM plugin for http/2 support is expected before the end of July (ticket #17861). In the meantime, increasing the concurrent connections (e.g. to 20) will help. iD editor already supports http/2.
The good news: imagery will be more recent and more frequently updated with our latest Maxar mosaic content. We hope that our improved production processes (bundle block adjustment, atmospheric compensation, continuous updates, etc.) will benefit OSM contributors.
As always, thanks to everyone within OSM for all your support and feedback. Shout out to the maintainers and contributors for iD and JOSM who have been fantastic to work with. Look for Maxar at State of the Map and State of the Map US; we will be attending and presenting more information on this new image service. And please feel free to send us feedback. Happy editing!
討論
由Omnific於2019年6月28日23:04發表的評論
Thank you for all your contributions! DigitalGlobe access has been an incredible asset to the OSM community.
由@kevin_bullock於2019年6月28日23:17發表的評論
Appreciate the feedback Omnific!
由AkuAnakTimur於2019年6月29日04:28發表的評論
Thank you Maxar/DigitalGlobe for sharing the imageries! Are there any channel to submit feedbacks e.g. a small patch of cloudy area in my area of editing interest?
由ndrw6於2019年7月1日08:22發表的評論
Thank you, I’ve been using your imagery for a while now, good stuff. Some places have been updated for the first time since 15 years. It immediately rendered my GPX traces obsolete.
Could you shine some light on image quality (resolution/clarity) and alignment? I would like to understand what the limiting factors are (e.g. technology, licensing) and if there is a short or long-term improvement expected. For example, imagery at google.com or even bing.com(!) is clearer, sometimes a lot clearer, than what I see in jOSM.
由Zverik於2019年7月2日13:01發表的評論
Thanks for the update, Kevin! In some cities currently DG Premium has more recent imagery than any of the Maxar layers. I hope the latter will be updated soon :)
由@kevin_bullock於2019年7月2日16:07發表的評論
AkuAnaktimor - sure, please send a coordinate!
ndrw6 - the resolution of our satellites range from 30cm-50cm per pixel. This is amongst the best resolution available via satellite. The reason why there are higher resolution images on Google or Bing is because they are also using aerial surveys in places like the US and Europe. Due to the lower flying aircraft, they can resolve a better resolution.
Zverik - we should have parity or better soon, or hopefully now. Feel free to send a specific location if you see this issue persist.
Thanks all for the feedback!
由4004於2019年7月2日21:49發表的評論
Thanks for the very welcome updates Kevin! One day I’ll hopefully start actually purchasing small areas of your imagery, but will enjoy it in osm for now
由freebeer於2019年7月3日11:52發表的評論
NDR-w6, additional higher resolution images can be seen in the us from mapbox in some locations, which have trickled up to one of the dg layers in the past in pixellated form. also, esri as a collective of at least hundreds of sources ranges from dg-matching imagery to much higher resolution (i think i saw native z23 details in a small part of us state hawaii). the esri clarity layer is about a decade old but matches what bing used to offer from aerial overflights.
then of course there are localised regional offerings that can be much better with pixel resolution to a handful of cm.
as i do not know your area of mapping interest, i give generic guidelines here based on my experience. as always, it is a compromise whether the ``best’’ (i think these quotes will get swallowed) imagery layer should be the newest, or the clearest.
thanks kevin, i almost always cycle through all tms imageries available for my worldwide mis-mapping and the wide choice you offer is much appreciated.
由angys於2019年7月4日09:55發表的評論
Like what AkuAnaktimor said, some of the imagery are cloudy. Hope MAXAR can update for it and adjust the imagery into right position as now most of the imagery coverage in Malaysia has a big offset compared to older DG Premium imagery. Cloudy area in Malaysia need to replace with a crispy imagery Bahau, Negeri Sembilan 2.81017,102.39802 Seremban, NS 2.72855, 101.95497
由erickdeoliveiraleal於2019年7月5日15:28發表的評論
where I can change concurrent connections?
由HeyItsAdam於2019年7月7日20:33發表的評論
Both of my region’s premium / standard imagery are the same. My assumption is that the premium imagery is only in select locations?
由iriman於2019年7月18日11:55發表的評論
Thank you for your great work. I’m from Iran and now in compare with previous DG imagery many more areas have a more recent and clear imagery!
BTW there is a misalignment about 8m that shows itself on top of the bridge at 34.09814158102221, 49.72359001100064.
由Baloo Uriza於2019年7月30日17:30發表的評論
Can we get some better alignment and resolution on the latest Maxar imagery? Not only is it not lined up, but the way that it’s out of alignment varies quite wildly over distances as short as 100m, and at resolutions that make it impossible to make out lanes.
由@kevin_bullock於2019年7月31日19:08發表的評論
@HeyItsAdam - yes, in some cases we have limited # of images to select from, so it may happen that the premium and standard are the same.
@iriman - that is great news, thank you. We will look into the misalignment issue, for me it’s most clear at the turning circle to the NE of the bridge.
@Paul Johnson - I’ve never seen a 100m misalignment. A few meters to be expected, 10m is considered an extreme case. Can you send a coordinate or an example?
由Baloo Uriza於2019年7月31日20:04發表的評論
@Kevin_Bullock: Really just sliding around Tulsa or along the length of US 412, US 169 or US 75 in Oklahoma, you can see it wobble substantially. I think I was looking up northwest of Cashion, OK when I saw it go totally off the rails.
由@kevin_bullock於2019年7月31日20:23發表的評論
@Paul Johnson - ok I misread your orginal comment, I mistook it as a 100m offset. But you are saying the offset changes within a 100 m stretch of freeway. I am seeing a 3-4m offset along the US 169 here: geo:36.50731,-95.70377?z=19 - but I am not seeing the wobble. If you can send a coordinate I can look into it. (wobble in imagery is usually caused by a bad elevation model, or an old elevation model)
由Prince Kassad於2019年8月1日08:48發表的評論
I am rather disappointed about the disappearance of the DigitalGlobe layers as of today. Some regions had really high-quality imagery that have now disappeared forever, they’re on none of the Maxar layers and on none of the other imagery providers.
See e. g. osm.org/#map=17/50.15857/8.99834&layers=N and generally the whole region there that had crisp imagery on DigitalGlobe Standard - Maxar is more current but much lower-resolution and with considerable offset.
I really hope the old imagery can somehow come back, because I was relying on it for my work.
由CENTRALHUB於2019年8月2日17:57發表的評論
The image distortion in Maxar is extremely horrible and it makes the mapping work become very painful. You can see obvious curve along a straight road now. Houses are facing different direction though they are facing the same direction in the real world.
Besides, in some locations, Maxar is out of dated and it’s data is older than DG Premium imagery.
In summary, Maxar maybe suitable for small scale mapping as supplemental data source. It’s hard for me to trust it for large-scale mapping
由@kevin_bullock於2019年8月5日15:51發表的評論
@Prince Kassad - the only change I can speak to are the old DigitalGlobe layers now becoming Maxar layers. Most users prefer more recent imagery vs higher resolution. It’s a balance we need to keep. With respect to the offset, please use the imagery offset tools in your editor.
@CENTRALHUB - please provide a location for the issue you mentioned. I don’t understand how houses can now be facing a different direction. In most cases, the new Maxar imagery will be newer data and refreshed more often.
由mirror176於2019年8月17日15:59發表的評論
My understanding was that DigitalGlobe standard was lower resolution while Premium being newer, higher resolution including some nonsatellite sources, and better image postprocessing. Was that correct and will that continue or will the two Maxar choices have different meanings? Will there be a way to get an approximate date of the imagery? In Phoenix, AZ I have always had good results of clear imagery without any major skewing of alignment. DigitalGlobe Premium looked to be aerial clarity though often without any steep angles and at a decent time of day for shadows to help see objects heights without often hiding anything. Though new is often great to have, higher resolution, seasonal variations, and differences in time of day and even slight angles from imagery have been necessary to make some of my edits possible. I can start to watch more for strange distortions from Maxar images in areas I work but if even the premium will be significantly lower resolution than Mapbox (=basically the old DigitalGlobe Premium with slightly different final processing) and Bing while having worse offset out of the box compared to DigitalGlobe and Mapbox aerials that i have known offsets for will make it not be my default go to for editing. Whatever the goals and outcome, thanks for being another usable source for us.
由@kevin_bullock於2019年8月19日16:16發表的評論
Hi Mirror176 - the Maxar layers will be satellite only (for the time being). The older DigitalGlobe layers indeed did share some aerial content with the Mapbox layer. This aerial imagery, typically available over the United States and some parts of Europe, is higher resolution, but infrequently updated. You are correct that the aerial imagery is higher resolution than the satellite imagery. That’s the advantage of flying in the atmosphere vs. flying in outer space. There are also some known offsets in the Maxar satellite imagery, I recommend using the adjust imagery tools in JOSM or ID. We are working to improve the accuracy and reduce these offsets. Seeing a 2m offset should be considered normal for satellite imagery. thanks for your feedback!!
由adiatmad於2019年9月12日02:39發表的評論
Thank you for all your incredible support for OSM. I wonder about Maxar vintage like DG vintage layer, will it be available in JOSM? Thanks in advance
由Jason Simonsen於2019年9月18日11:23發表的評論
Firstly, thanks for the imagery.
I am noticing around Brisbane, Australia around a 5 to 10m offset to other image sets.