IssueTracker

Keep up the good work.

#googleadswords.com

Can we make an internal mirror bug for this feature request? Thanks!

It appears to be slowly being fixed as new imagery gets released, from which they soon become the default when viewed. Also there are not as many broken lines as before, but this is just my guess.

Replying to comment #43: it is NOT your browser or computer that is making you see the broken coverage; it’s the API.

The Google-contributed Street View policy at https://www.google.com/streetview/policy/ appears to be quite out of date and inconsistent with this new change. Instead of waiting a month or more for new imagery to roll out (in the form of mass updates), the interval has been dramatically cut short to roughly weeks so that the data is always accurate in real-time.

Indeed the problem is usually from the precision of the map that is not sufficient. Instead of reducing the threshold, they should adjust the map using the Street View data and the satellite imagery (when applicable). This is really a bad move Google did, it completely damaged Street View as a whole. This change must be reverted because clearly the roads are not getting adjusted at all, 50 meters was fine to cover most of the cases. And we have no way to adjust them since reporting incorrectly drawn roads leads to nothing. We are stuck with an unfunctional service yet we still pay for the API costs.

When Google decided to multiply those prices by 14 in 2018, we would at least expect an increase in the service's quality. In the end it is the complete opposite:

• because Google started letting anyone make their own blue lines this same year without adding any control to Google Maps users and API users (which is a must have when you let anyone uploading data without any quality control), we are now obliged to suffer this lower quality unofficial coverage
• because Google is not counting on other people to create Street View we don't get new countries covered with official Google Street View (we hope this announcement for Rwanda made on Twitter by the Google Africa account is true and not another false hope)
• because Google decided to lower the threshold to validate Street View data we now have a completely broken service
• because Google decided to lower the maximal distance we can jump to something like 50 meters we are now with even more unusable service

Should I keep going? Since 2018 Street View kept losing quality and nowadays its state has never been so bad. It is really concerning that whoever is in charge of the project doesn't see all the trouble their decisions are making. We really miss the golden age of Street View when Google was adding new countries regularly and being proud of it by doing large announcements and events.

At least in 2021, despite of the removal of Vietnam and the new imagery in Switzerland, we could see some improvement with many updates and new areas added in Mexico, USA, Argentina, Labrador (Canada) etc. but this new way of publishing data progressively is not convincing at all as the data is never complete, resulting in broken blue lines and limited usage. Doing less but larger and more complete updates was quite better.

Maybe in 2022 Google will finally realize how bad the Street View project is managed and take action to get back on tracks otherwise Apple will at some point be superior with Look Around.

Quoting from comment #47: "Maybe in 2022 Google will finally realize how bad the Street View project is managed and take action to get back on tracks otherwise Apple will at some point be superior with Look Around."

Yeah, but while I do agree, I would personally still keep the progressive update system as long as they are able to find a cure for these glitches.

We don't know if we'll see any imagery from this month as the data schedule at https://www.google.com/streetview/explore/ remains unchanged. Maybe later this week or next week it'll get updated, but still expect some inconsistencies in new captures.

Google doesn't take this list very seriously. Only Google Japan seems to regularly update it a bit (and it's still not that thorough). Anyways this is not the topic of this discussion.

It looks like the problem where you can only jump at a limited distance has been fixed (for most locations). But more things to come.

Welp, looks like the random breaks have slowed down a bit as new imagery is being released.

Uploads of past imagery are also more than likely to cause breaks in coverage. I stated this in a forum post of mine: https://support.google.com/maps/thread/146520554/street-view-gen-4-imagery-circa-2016?hl=en

The gaps are caused by the fact Google decided to lower the threshold that defines if a panorama is valid or not (depending of it's distance from the road on the map, lowered from around 50m to around 10m). And because Google Maps is not precise enough it broke everything. The best way to fix it is to revert this change but Google apparently doesn't care and will not revert it. The other way is improving the map accuracy but we need a miracle to happen at this point as there is so much to do (and we can't help as we have no tool like Map Maker anymore to adjust existing roads).

As for the list you shouldn't 100% trust it. I've been tracking the changes in this list since 2020 and I can confirm Google manages it poorly. They like to put everything from a region but they won't actually cover or update it. For example British Columbia has all the subdivisions listed but they will usually just drive in the south, sometimes in the center around Prince George, but never in the north or the west close to the Alaskan border. At contrary they sometimes drive in unlisted places like Labrador in late 2021 that was never listed. But it's quite rare.

Also they listed many countries where cars actually drove in 2021 but the pictures are not published for unknown reasons. Countries such as Romania, Colombia, Slovakia or Philippines used to be regularly updated but now the pictures are blocked (you can clearly see them near borders, when the car entered a neighboring country where updates are published: check around Hungary and you'll find leaks from Slovakia, Austria, Romania, Slovenia and Croatia). In 2021 we saw cars driving in 50 countries, but 23 of them have blocked updates for unknown reasons.

In my opinion the only way we can fix the 'broken' coverage bug is by writing a patch for both the Maps and Street View APIs. But this process would probably take a considerable amount of time.

Quoting from comment #59: "Also they listed many countries where cars actually drove in 2021 but the pictures are not published for unknown reasons."

Well, that is due to the fact that when imagery is captured, according to the Google-contributed Street View imagery policy, it takes a tremendous amount of time to stitch and process everything together. However with the new continuous-updates policy for some regions, the process has become a big hurdle as it is extremely buggy nowadays.

If this issue doesn't get tracked then we'll face extreme problems with the quality of the Maps service. It's just the bottom line. :(

Small follow-up about large areas affected by the shift:

• Chita, Russia was fixed 2 weeks ago and looks good now
• more areas in Russia are affected, for example around Oktyabrsky, Republic of Bashkortostan
• Alaska is still broken in the east, affected area is a circle of approx. 125km radius around Tok
• eastern Thailand close to Cambodia is actually a problem with the map, roads are shifted 20~30m from their actual position; Street View data seems correctly placed and removed because of the change in the validation

All the other broken blue lines are caused by the tolerance's threshold as far as we can see. If we find more areas affected by a shift we will let you know.

So if it has to do with a lower threshold and misdrawn roadways, then I totally agree. These changes have indeed impacted the quality of the Street View service.

Is it ever going to get fixed or no? Google has to do something; otherwise it'll be a total mess for Maps and Street View.

Okay... looks like the broken lines are steadily increasing as the first 2022 imagery is being released.

See this related issue, which talks about the nav arrows likely caused by the broken coverage bug: issue 217936671

Thanks for waiting.

We believe this issue is now fixed. If you are still having issues please let us, or better you can file a new separate issue.

This is not fixed at all. Alaska is still with completely broken coverage because of some shift similar to the Azores or Chita. The city of Orsk is also impacted in a similar way. The town of Usinsk in northern Komi Republic received coverage recently and also has this issue. Anyways the core issue is the fact Google decided to lower the tolerance for a blue line to appear on the map and this broke thousands of places. This will only be fixed if this decision is reverted because it doesn't make sense at all to have taken it in the first place. This really ruined everything.

Really? Why close this issue when we still recognize the fact that the bug keeps lingering around? We have to do something! :(

Please hurry up and reopen this issue before it doesn't get tracked! :'(

(Replying to comment #73) Unfortunately, you would most likely have to wait long-term for Google to resolve the issue. The broken coverage bug is a very serious concern we have to take responsibly, and resolving it can be a slow process (somehow).

But, for now, why not take a look at those user-contributed photo spheres in your area?

We're still seeing broken lines everywhere as new imagery is released, so there's clearly no point of closing this issue thinking it's gone. Proof in the image below.

Also, the fact that these lines are broken makes us wonder if there are blurred objects somewhere, as this scenario is common because when something is blurred, all revisions get completely removed or truncated. But that's not the case because this is an API-related bug and despite that the images still function and remain accessible.

Okay, I'm seeing that Google tried removing the 9/10/2021 release data in areas outside the Bay (how dare?!), and the results appear quite disastrous. Now there are broken lines almost everywhere across the country.

The broken coverage bug is still happening. Please reopen this issue! :(

Quoting from comment #80: "It's not that big of a deal if the coverage is still accessible (which it is)."

Yeah, but it can be annoying for some. As the glitch is as severe as software bugs, it can raise concerns for some who are trying to use Maps without any breakage.

(Replying to comment #82) And if that's the case, we REALLY need to reopen this issue. No matter the severity, this bug totally disrupts the Maps experience.

Quoting from comment #59: "In 2021 we saw cars driving in 50 countries, but 23 of them have blocked updates for unknown reasons."

Unknown reasons? I have a better way of explaining that: it's due to privacy concerns and certain Internet laws.

If Google still doesn't do anything about it, we'd be in total pain. Removing the 9/10/2021 release data is NOT feasible enough and this should be reverted.

Any update?

Reminder that this issue is not fixed and won't be until the change about data validation is reverted.

I can only assume that the reason for the removal of these release images was because of a change in the threshold that determines whether a panorama is valid or not. However the change doesn't appear to affect Mountain View, CA as it is still maintaining its own threshold value.

Broken lines are still showing up. Can’t we just reopen this issue?

See related issue 228201902.

Earlier we saw Google completely removing the 9/10/2021 release data from the Mountain View area (which includes Gen 4 data from between March and September 2017, and that doesn't help reduce the number of broken lines. We really need to revert this change.

Are we really not going to take care of this? Many rural areas in the US lost historical coverage and there is no way we can access them today.

Replying to comment #95: Please don't start to spam issues on this topic. Doing that doesn't help with resolving the issue.

(Replying to comment #99) Welp, you got that right. Consider that another part of the March 1 imagery removal.

I still can't access data from the 9/10/2021 update. Blue lines went missing in some German towns, Gen 4 content from 2017 in the US is gone, Gen 3 content from 2019 and 2020 disappeared in the US, Japan lost Gen 1 content after it was restored, and a few areas of Bangladesh lost new content. Google really needs to fix this. :(

The 9/10/2021 data's pano IDs are expiring. Google needs to take action right now because if the data is not recovered/restored, they will forever be lost.

It was recently sought according to issue 232516758 that the threshold value for published content has been increased from 10 to 20. While not as good as before, it did kinda fix the broken coverage bug a bit. Still, I kinda feel bad for the 9/10/2021 data.

Okay, it seems to be getting worse. Whenever I navigate using the keyboard, I see a warning on the bottom that says 'No Street View imagery available here.' This is super awkward. I mean, really? :\

Hmm, I seem to suspect that the 'No Street View imagery available here' warning when navigating may be because the API is trying to access the 9/10/2021 data, which doesn't exist anymore. Speaking of which, I did create issue 241306068 last month to address my case with that data.

It has been realized that another change was made, breaking more coverage mostly in Nigeria.

This change seems related to how far each panorama can be to allow a blue line to exist (and data to be accessible). For some reason on a bunch of Nigerian roads the camera didn't take enough panoramas to have a sufficient density, making data unpublished. This is why many roads got their coverage removed last year.

Attached image shows an example of a road in eastern Nigeria (https://goo.gl/maps/xr2mPPDzs16s28867), which was published this week for the Google for Africa event. With the API the panorama positions were gathered and displayed as red dots, with connections between them displayed as red lines. As you can see the panoramas have inconsistent distances between them and at some point the blue line will stop appearing because the distance becomes too long. This affects many roads covered with the gen 3 Trekker made between 2018 and 2020 in many parts of Nigeria, probably the camera was faulty or misconfigured.

While it can be understandable to remove data that is too far from a road on the map to avoid misplaced data (though this is in 95% of the cases a problem from the map having wrongly traced roads), reducing the threshold for distances between panoramas is not logical. This brings no benefit and breaks things. Can this be modified to allow higher distance between panoramas to bring back those blue lines?

Also reminder that the Street View coverage around Tok in eastern Alaska is still broken when it was perfectly fine before. The data got shifted and needs to be re-shifted to align it on the roads again. This is a similar fix to do than the one your did for the Azores or Russian cities.

Quoting from comment #110: ”For some reason on a bunch of Nigerian roads the camera didn't take enough panoramas to have a sufficient density, making data unpublished. This is why many roads got their coverage removed last year.”

Insufficient density? Hmm… that probably also went for the 9/10/2021 Street View data. But honestly, it took forever for the data to appear on Maps, and I would not be happy to hear if the removal occurred for the same reason.

The 9/10/2021 Street View data still remains inaccessible. Imagery from the update has already been removed from the API database, so I’m not sure if the data will ever return.

(Replying to comment #115) I am NOT talking about the newer data. What I’m trying to focus on is the other past data that didn’t make it live to Street View until September 10, 2021, and nowadays that data is no longer accessible. Please look at the screenshots on issue 241306068 for reference.

It seems like you’re not familiar with the mass removal that occurred on March 1, 2022, and images affected by it were those made part of the 9/10/2021 update. Let me be specific: the imagery was dated between 2008 (in Japan) and 2020 (in North America), and were accessed through the Time Machine. At the time of the update I first found Gen 4-quality imagery from 2017 across various cities in the US. It wasn’t until moments later that I came across other images captured with past-Gen cameras (all of which were entirely copyright 2021), like Gen 3 content in the western US and newly-added coverage in a part of Germany mostly in Gen 2 quality, and the return of blurry Gen 1 imagery in Japan. Unfortunately, that content didn’t last long enough, and that’s something Google hasn’t sorted out. Newer images, especially those released through the continuous updates system, are NOT affected.

To see for yourself, try this link: https://goo.gl/maps/TVJcGuBt9PxQ7m7x8. This one led to an image dated December 2015, and to describe what was originally seen, visible LIDAR sensors and 2021 copyright watermarks indicated that it was released September 10, 2021. If you click on it today, Street View will load but it fails with a ‘No Street View imagery available here’ warning, subsequently returning an image different than what I initially explained. That’s because the imagery’s pano IDs expired months after they were removed, and the removal is likely because of insufficient densities at the time of their release (as explained in comment #110). For your convenience, attached here is a screenshot of the ACTUAL image described.

So what the heck is your concern? :\

(Replying to comment #115) Also check out this map. The driveway outlined was once filled with blue lines and its imagery was dated June 2017 in Gen 4 quality. To describe what was seen, the rooftop of the Street View car was visible, as the car was a Japanese model with a lowered camera (very unusual to see these kinds of Street View cars in North America). With the removal occurring March 1, 2022, its blue lines have been removed, and while only half of the blue lines remains, it does not lead to the actual data.

(Replying to comment #115) Alright, I seem to have misunderstood. You should know there is a feature in Maps that allows you to view historical captures of the same road, which they call the Time Machine. In Street View where it’s available, click on the capture date on the top-right corner, and you’ll see a timeline of captures, starting from the oldest (2007) to the latest (this year). That’s where the data is missing. Is that clear now?

(Replying to comment #119) You may think the data I mentioned from the examples is still around, but literally, it's not. I'm trying to be specific. Try this link: https://goo.gl/maps/3FojuB2m3pt6Qf768. Attached here is a demonstration of the problem.

Quoting from comment #119" "Also, this issue isn't even related to insufficient densities, so if you want to try I'd recommend opening a separate issue instead of spamming the same issue over and over again."

And if I did that, it would simply get closed as a duplicate of issue 241306068 because they're already handling my case, even though I haven't heard back from them. I'll try, but just be honest, they're not going to believe me.

(Replying to comment #119) Have you ever seen/read this: https://virtualstreets.org/index.php/2021/09/12/a-new-google-street-view-update-happened/ ?

There's proof that the data once existed. Several months since the data was removed, some of the links there no longer work. Clearer now?

(Replying to comment #123) No, I'm talking about a specific set of imagery from the update belonging to the 'historical' category. The imagery was released September 10, 2021 and remained until March 1, 2022 From that page here is a list of links that currently don't work because of that:

• https://goo.gl/maps/cpDFNWMUsLUNx3PX7 - Was expecting a Gen 4-quality photo in Ceuta in northern Africa. Original capture date unknown (no, not the current 2022 coverage).
• https://goo.gl/maps/9vzt84opYPQqwurx5 - Was expecting May 2021 coverage in Dhaka, Bangladesh. This addition has been undone; currently there is only 2013 coverage.
• https://goo.gl/maps/WqzuNcTGUXCEipg8A - Was expecting February 2008 Gen 1 coverage in Japan. This addition has been undone.
• https://goo.gl/maps/DAxtqEPirBGrvzVx9 - Was expecting a Gen 4-quality photo in Alaska. Original capture date unknown (no, not the current 2021 coverage).
• https://goo.gl/maps/RCbeA5ikVSmYGMXK7 - Was expecting a photo from July 2015 in San Marino, Italy. This addition has been undone; it now returns a user-contributed photosphere instead.
• https://goo.gl/maps/MxLUuQvY5Sa1oTLJ7 - Was expecting a Gen 3-quality image in Luxembourg. Original capture date unknown.
• https://goo.gl/maps/Mpx9WXHr8LsJQ79U6 - Was expecting an image somewhere in Indonesia. Camera generation and original capture date unknown.

While the removal was not mentioned on that page, it is a very annoying issue for me as I cannot access that data anymore. And if you were to access the data's pano IDs past their expiration dates in the API, you'll get a ZERO_RESULTS error. So, literally, this has NOT been fixed at all.

[Replying to comment #123] Alright, I may have described it incorrectly. Only certain parts of the data from the update were accessed during their availability (September 10, 2021 - March 1, 2022). I'll be providing images I attached on issue 254502855, all which I've took during the data's existence. Other images are unaffected.

[Replying to comment #123] Okay, maybe not densities. I suspect the cameras that took the photos may have been severely misconfigured, and that could be a key reason to why the data got removed. If this was to be corrected, the data could then be instantly restored.

Unfortunately, broken blue lines are on the rise again thanks to huge updates being rolled out in some European countries. However, it doesn't mean that the imagery is inaccessible; you can just drop Pegman anywhere if Street View imagery exists.

Auugh, misdescribed again! Ahem... alright, I'm thinking most of the 9/10/2021 Street View data had that very short lifespan due to poor data precision or some other misconfiguration, and that could be why the data didn't get released until that time. And the reason that update was so important was to provide more historical records for research and quality updates to existing coverage. Should be easy to understand.

Oh, another thing: I believe the 9/10/2021 update is closely related to the 10/8/2020 update, which lifted caps that prevented newer data from being published in some California and Texas cities due to elevated privacy concerns. Numerous stay-at-home orders amidst a raging pandemic allowed Google enough time to publish the newer data in those areas, and since then future captures have been allowed to be published. But unlike the 9/10/2021 update, the 10/8/2020 update data never faced a removal. So, if Google can understand how important the 9/10/2021 update data was, the imagery from that update could be published once again, hoping it will never again get removed.

(Replying to comment #130) O_O

Umm... so you're saying that the release was accidental and when Google noticed they intentionally removed it for privacy? Sad. :'(

But at least the extra Gen 3 and Gen 4 data in North America would still be good enough. Not really an issue.

Quoting from comment #132: "From memory, a lot of the 2017 gen 4 was completely without any blurring which is a massive privacy issue that would cause legal problems for Google, so they would've had to remove all the images from that time period as a precaution."

I don't remember the imagery being released without any blurring at the time of Gen 4's announcement and soon being removed, but the majority of the early Gen 4 images that were released in 2021 actually had all the license plates and faces already blurred. However, the only exception was with the March 2017 imagery around the Googleplex in Mountain View, which shockingly had the license plates and faces unblurred, and upon noticing I had to request manual blurring on those license plates and faces; this process turned out to be successful. Still, though, I kinda feel bad for those images, and if those repetitive blurring requests of mine were what caused the removal, then that'd be final.

Quoting from comment #132: "The 2017 gen 4 and 2020 gen 3 in the US was test imagery, and Google never publishes test imagery so it also makes sense for this to be removed in my eyes."

They shouldn't be considered test images if those cars were travelling throughout various parts of the US. In my opinion, I would only consider those images to be tests if the cars were strictly driving in the Bay Area. Speaking of tests in the Bay Area, a few already-built Gen 4 cars were spotted at various times in 2016, a year before those cars officially started spreading throughout North America. Though the data captured by those cars during that year was not made part of the 9/10/2021 update (as I originally expected), it was sought that the LIDAR sensors used by the Gen 4 cars were also used on some of the existing Gen 3 cars throughout the Bay, and data from those vehicles were in fact published in the update.

Quoting from comment #135: "The Gen 4 camera was officially announced in September 2017 so I believe those images were in fact later testing and not meant for publication."

Well, you should be aware that a few imagery from those test Gen 4 vehicles were actually released at that time. It's mainly found in the cities of Phoenix, AZ (https://goo.gl/maps/pBuwVFbuhcYvg7eG6) and Santa Cruz, CA (https://goo.gl/maps/rD88JKpDDTK5akZK9), but scatters of August 2017 Gen 4 data do also exist in parts of the western US (e.g., https://goo.gl/maps/9HLyvbsJ7bzJ6Kmf6). So, this should make sense to you.

Quoting from comment #135: "Before Gen 3 launched in January 2011 nationwide, cars were spotted driving as far away as Florida and San Diego in late 2010 but these images are also unpublished."

Actually, a fraction of data from these vehicles at that time exists in those areas. This example shows proof: https://goo.gl/maps/2de6HGHMxrWaaZrN7

However, the rest of the Gen 3 data from 2010 does remain absent and wasn't even included in the 9/10/2021 update.

Quoting from comment #132: "The 2017 gen 4 and 2020 gen 3 in the US was test imagery, and Google never publishes test imagery so it also makes sense for this to be removed in my eyes."

It doesn't really matter, though. I would still republish at least the early Gen 4 data in the US to provide more historical records for research and add quality updates, but this would probably need to be discussed with the folks at Google per your explanation.

(Replying to comment #132) It's not just the 2017 or 2020 test imagery. You should also be aware that when an update occurred on October 8, 2020, it included imagery from several test Gen 3 vehicles in areas where the caps existed before the update. See this example: https://goo.gl/maps/cbaSs6NQ7uEyoERq5. It's dated October 2018, and it was captured by a test vehicle whose imagery was never published elsewhere. And if no removal occurred, it means that Google can publish test imagery after a long while.

Dear Google employees,

It is possible to block da...@gmail.com from spamming the issue tracker please? This account keeps spamming non pertinent content all the time, creating way too much noise in our email inbox. Many people follow tickets to get news about potential changes or fixes and are very bothered by this.

Thank you for understanding.

Quoting from comment #141: "The September 2021 update was very poorly done and what you have now is what you'll be stuck with from this update, the removed coverage isn't returning."

Uhh... okay, I see what you mean now. :\

When I read that article mentioned on comment #122, it does say that the update did bring a number of 'technical problems' in which imagery from prior updates was deleted, and in most cases caused a lot of broken coverage throughout the world. Even after Google undid the update back in March 2022 (which was never mentioned in that article), it didn't fully fix it until like several months later, though countries like Vietnam and parts of Switzerland continue to see missing coverage to this point.

I think what happened was that when they lowered the threshold for Street View content at the time of the update, it resulted in a lot of roadways on the Map going far out of their original position, which might have created this 'broken coverage' phenomenon. So, if that's the case, you're probably right: the update itself was very faulty. Maybe next time they should think twice before applying the new changes along with the update and determine if it would cause bugs in the API or not, and hopefully the quality of the Street View service would continue to be in good shape.

Though it was great seeing those images from the update, the removal could have not occurred. I would probably let Google consider restoring a portion of the update in North America, without causing any breaks in coverage (at least the Gen 4 content would be good enough, even if they were just merely tests because some of the vehicles drove around areas where no new coverage or quality updates existed, and as long as there's blurring).

Quoting from comment #141: "The update that you keep rambling about was over 15 months ago now and was one of the most broken updates of all of Street View which ultimately lead to the removal of the old update system."

Hmm... actually, the update may not be responsible for the breaks. It rather has to do with the API itself because when Google implemented the new 'continuous updates' system in some countries immediately after the update, this feature was not fully complete as it could have contained errors in the code, and hence as new data is released through this system, the likelihood of coming across a large number of broken blue lines is seemingly high. So, deleting the data from the 9/10/2021 update is practically infeasible.

https://faq.whatsapp.com/465883178708358/?ref=share

(Replying to comment #145) Why are you posting comments unrelated to this issue? That's considered spam. Also, you provided your full email address, which is a violation of your privacy. You better delete those posts before trouble arises.