O.k. I finally sent an e-mail to someone with a Phd. in satellite imagery
Which university offer a Ph.D. in Satellite Imagery?
Isn't that just a specialization within the field of Physics or Engineering? Besides most of the imagery in the maps application are aerial not satellite.
Mapping programs get their position data from mathematical and geo spacial algorithms not from photographic imagery. Satellite images are said to be 10m or 1m or whatever which is related to the relationship to one pixel. When you view a hybrid image, the image data is scaled to match the geo spacial zoom level not the other way around. That is why you don't have smooth zooming but instead have fixed levels because the images are pre scaled. The geo spacial data is more precise than the image data because the the original images are rectangular which are distorted and stretched along their edges so they can be tiled. They only approximately match the geo spacial data.
Which university offer a Ph.D. in Satellite Imagery?
Isn't that just a specialization within the field of Physics or Engineering? Besides most of the imagery in the maps application are aerial not satellite.
Mapping programs get their position data from mathematical and geo spacial algorithms not from photographic imagery. Satellite images are said to be 10m or 1m or whatever which is related to the relationship to one pixel. When you view a hybrid image, the image data is scaled to match the geo spacial zoom level not the other way around. That is why you don't have smooth zooming but instead have fixed levels because the images are pre scaled. The geo spacial data is more precise than the image data because the the original images are rectangular which are distorted and stretched along their edges so they can be tiled. They only approximately match the geo spacial data.
You are absolutely correct. I tried to keep my post brief.
He's a gentleman with a degree in Engineering who specializes in Image Processing and Interpretation. He wrote his thesis on estimation and algorithms for satellite image deconvolution.
It sounds like both you and he would get along well.
Also, if you read that post I mentioned that he responded with exactly what you stated (shortened). The older, Google, maps are indeed aerial when I had actually believed they were satellite based. I was wrong.
Also notice where I stated that I REALLY shortened his response to me because I didn't really think it was anything most people here would want to read. Keep in mind that I have gotten rude comments about providing too much information. Who knew right? Apparently most here would rather read "Apple is doomed!".
This glut of errors is exactly why they need to prioritize and fix the high visibility errors first.
if ("Columbia, SC" == strSearch) then
location.lat = 34°00?2?N
location.lon = 81°02?39?W
end if
If you've worked on software, you know sometimes imperfect fixes NOW are a heck of a lot better than waiting on perfection. That doesn't take months.
If you can't hard code a server-side fix to negate the power of a print ad in 2012, you've got a serious problem. As soon as you knew there was a print ad with a specific point, you include something like the above and everyone who tries the point from the ad goes, "Huh? Apple gets that right." Insane.
Look, I don't have a lot of Apple stock, but I've got over $20k of it (and that's after "profit taking" several times). I like Apple. They've done well. This is crap.
Hard to let that one go. Columbia, SC, established 1786. "new and fancy" stuff, indeed. You must be British (or Egyptian or Iranian, etc) to call that "relatively new".
Three points:
1) You have no idea what their fix queue looked like before ship. It's a really lousy bet that they sat around for months twiddling their thumbs presuming everything was great, as the article implies; rather, they were very likely fixing things right through ship with an effectively infinite queue. We might quibble about their priorities when looking at ridiculous results in particular cases, but you can't just assume they weren't doing anything.
2) Your fix is way too specific to be useful. If they had to fix everything that way they'd have millions of fixes. They had (perhaps still have) some pretty bad search algorithms that didn't narrow effectively, but the way to fix those is not to apply thousands of one-offs but instead to improve the search criteria. It's obvious in the Columbia, SC case that they were ignoring the SC part (the same is true of some of the screwball New York City search results) and even without the user entering that discriminator nearness-to-location would be a very good guess. So, do you do your one-line fix that works for the one very specific fix, and iterate thousands of times, or do you take maybe a few days to think about your search algorithm and and fix it almost everywhere all at once? Maybe (probably) you end up doing a mix of both.
I know they've been making algorithmic adjustments because I did a search the day after iOS 6 came out and it got useless results, then did the same search four days later and it was spot on. I'm guessing, mind you, that they didn't suddenly hand-optimize the search for an REI in northeastern Massachusetts. It's not like a missing Statue of Liberty. So, they got better *very* fast.
This is the old optimized-code-versus-improved-algorithm tradeoff. You can do pretty well in the very short term by hyper optimizing specific cases, but better algorithms are going to crush that approach in the longer term, and probably end up being less work too. Should they be spending the time on zillions of one offs as you suggest, or core improvements? I'd go with core improvements almost every time.
3) The "new and fancy stuff" I was talking about was not city locations, it was the 3D views where they are doing some pretty interesting projections by combining multiple kinds of data. You get melted bridges because they're projecting a flat image on the topo data and the image doesn't know anything about roads or bridges or landmarks, it's just pixels. That looks ridiculous, but seen in that light it's very easy to see why it happened and why it's not easy to fix: The bridge data doesn't exist in the topo map dataset, nor is there any notion of "this is a bridge" in the satellite image. There might be some of that information in the street map data, but most likely they're going to have to go in by hand and correct every bridge et al.
It has taken Google years of these kinds of algorithmic improvements and individual manipulations to get to the point they are now. No way was Apple going to be able to match that out the door. The difference in quality versus Google is a good source of jokes, no question, but Apple really is updating this stuff daily just like Google and other map providers do. Improved algorithms and many manual corrections are already obviously being made. That's what they ought to be doing, right?
Lambasting them for not holding this product in limbo for years while they perfect it is just ridiculous. They needed turn-by-turn two years ago, they couldn't just keep using Google data and accept major limitations like that.
If they had to fix everything that way they'd have millions of fixes. They had (perhaps still have) some pretty bad search algorithms that didn't narrow effectively, but the way to fix those is not to apply thousands of one-offs but instead to improve the search criteria.
Actually it is a little of both. One of the big advantages that Google has is their search business. They know where every, thing, person, place, dog, cat and car is. So yes it does take hundreds of millions of data points to make a good map search engine. The algorithms come into play after the data points are indexed. Apple is getting their data from numerous sources so I think the job of indexing, purge and merge is more difficult because the sources are disparate. The Columbia mistake is not an algorithm error, it is a typo in a database somewhere in the table 'countries' in the field 'country_name' that spells Colombia incorrectly, it is not spelled like the university. They simply have to find it and fix that typo.
Only most times they aren't known to the software house. Software released with known bugs is a sign of deception.
Only many times they are. You have software to ship to a deadline and a list of bugs you haven't eliminated. You have to decide if any are show stoppers. You've never developed serious software if you think this never happens. Not only with software. You're designing a car and there's a list of flaws you can't eliminate in time for the launch. Guess what happens. No one has ever engineered a perfect product; the better engineering teams get closer to eliminating all known flaws and minimising any that surface later.
The real question here is why Apple pulled the plug on Google without putting this out for extensive real-world testing. That seems to me idiotic when they had that option, especially if there were serious flaws known to the development team.
IOS maps based on Google maps didn't and still doesn't.
The End.
I think this is the primary reason most iOS users will probably find Apple Maps more useful for them than the previous maps app. It doesn't have anything to do with Google maps being more/less accurate, or listings searches being more/less complete, or Flyover being more/less useful than Streetview.. It's that users won't have to purchase a standalone nav app, with the free Apple version good enough for most.
This is exactly like FCP X. Apple had to make a move. Historically, Apple's v1.0 products are NEVER good. It's not supposed to be good. It is a strategic move for Apple Inc. Of course, legacy systems such as Windows XP are much more reliable and developed. That is the advantage of using old, legacy products! Sometimes they work great.
Apple will fix this Maps app like they always do. Right now the people piling on are the sheeple. This is the same old Apple. Making a strong move with a v1.0 product. Expecting it to be polished is lunacy. This isn't Microsoft with a 5 year Beta period. If you have mission critical applications, by all means, don't use Apple. That's not what it is for.
Comments
Originally Posted by Vadania
How high does your list go? Is this something that you collaborated on with the other moderators? Can you post the whole list? ...or e-mail it to me?
Just about 50 now.
Quote:
Originally Posted by Vaelian
blah, blah, blah
iOS 6 Apple maps has voice guidance.
IOS maps based on Google maps didn't and still doesn't.
The End.
Quote:
Originally Posted by Vadania
O.k. I finally sent an e-mail to someone with a Phd. in satellite imagery
Which university offer a Ph.D. in Satellite Imagery?
Isn't that just a specialization within the field of Physics or Engineering? Besides most of the imagery in the maps application are aerial not satellite.
Mapping programs get their position data from mathematical and geo spacial algorithms not from photographic imagery. Satellite images are said to be 10m or 1m or whatever which is related to the relationship to one pixel. When you view a hybrid image, the image data is scaled to match the geo spacial zoom level not the other way around. That is why you don't have smooth zooming but instead have fixed levels because the images are pre scaled. The geo spacial data is more precise than the image data because the the original images are rectangular which are distorted and stretched along their edges so they can be tiled. They only approximately match the geo spacial data.
You are absolutely correct. I tried to keep my post brief.
He's a gentleman with a degree in Engineering who specializes in Image Processing and Interpretation. He wrote his thesis on estimation and algorithms for satellite image deconvolution.
It sounds like both you and he would get along well.
Also, if you read that post I mentioned that he responded with exactly what you stated (shortened). The older, Google, maps are indeed aerial when I had actually believed they were satellite based. I was wrong.
Also notice where I stated that I REALLY shortened his response to me because I didn't really think it was anything most people here would want to read. Keep in mind that I have gotten rude comments about providing too much information. Who knew right? Apparently most here would rather read "Apple is doomed!".
Quote:
Originally Posted by rufwork
This glut of errors is exactly why they need to prioritize and fix the high visibility errors first.
if ("Columbia, SC" == strSearch) then
location.lat = 34°00?2?N
location.lon = 81°02?39?W
end if
If you've worked on software, you know sometimes imperfect fixes NOW are a heck of a lot better than waiting on perfection. That doesn't take months.
If you can't hard code a server-side fix to negate the power of a print ad in 2012, you've got a serious problem. As soon as you knew there was a print ad with a specific point, you include something like the above and everyone who tries the point from the ad goes, "Huh? Apple gets that right." Insane.
Look, I don't have a lot of Apple stock, but I've got over $20k of it (and that's after "profit taking" several times). I like Apple. They've done well. This is crap.
Hard to let that one go. Columbia, SC, established 1786. "new and fancy" stuff, indeed. You must be British (or Egyptian or Iranian, etc) to call that "relatively new".
Three points:
1) You have no idea what their fix queue looked like before ship. It's a really lousy bet that they sat around for months twiddling their thumbs presuming everything was great, as the article implies; rather, they were very likely fixing things right through ship with an effectively infinite queue. We might quibble about their priorities when looking at ridiculous results in particular cases, but you can't just assume they weren't doing anything.
2) Your fix is way too specific to be useful. If they had to fix everything that way they'd have millions of fixes. They had (perhaps still have) some pretty bad search algorithms that didn't narrow effectively, but the way to fix those is not to apply thousands of one-offs but instead to improve the search criteria. It's obvious in the Columbia, SC case that they were ignoring the SC part (the same is true of some of the screwball New York City search results) and even without the user entering that discriminator nearness-to-location would be a very good guess. So, do you do your one-line fix that works for the one very specific fix, and iterate thousands of times, or do you take maybe a few days to think about your search algorithm and and fix it almost everywhere all at once? Maybe (probably) you end up doing a mix of both.
I know they've been making algorithmic adjustments because I did a search the day after iOS 6 came out and it got useless results, then did the same search four days later and it was spot on. I'm guessing, mind you, that they didn't suddenly hand-optimize the search for an REI in northeastern Massachusetts. It's not like a missing Statue of Liberty. So, they got better *very* fast.
This is the old optimized-code-versus-improved-algorithm tradeoff. You can do pretty well in the very short term by hyper optimizing specific cases, but better algorithms are going to crush that approach in the longer term, and probably end up being less work too. Should they be spending the time on zillions of one offs as you suggest, or core improvements? I'd go with core improvements almost every time.
3) The "new and fancy stuff" I was talking about was not city locations, it was the 3D views where they are doing some pretty interesting projections by combining multiple kinds of data. You get melted bridges because they're projecting a flat image on the topo data and the image doesn't know anything about roads or bridges or landmarks, it's just pixels. That looks ridiculous, but seen in that light it's very easy to see why it happened and why it's not easy to fix: The bridge data doesn't exist in the topo map dataset, nor is there any notion of "this is a bridge" in the satellite image. There might be some of that information in the street map data, but most likely they're going to have to go in by hand and correct every bridge et al.
It has taken Google years of these kinds of algorithmic improvements and individual manipulations to get to the point they are now. No way was Apple going to be able to match that out the door. The difference in quality versus Google is a good source of jokes, no question, but Apple really is updating this stuff daily just like Google and other map providers do. Improved algorithms and many manual corrections are already obviously being made. That's what they ought to be doing, right?
Lambasting them for not holding this product in limbo for years while they perfect it is just ridiculous. They needed turn-by-turn two years ago, they couldn't just keep using Google data and accept major limitations like that.
Quote:
Originally Posted by jimfrost
If they had to fix everything that way they'd have millions of fixes. They had (perhaps still have) some pretty bad search algorithms that didn't narrow effectively, but the way to fix those is not to apply thousands of one-offs but instead to improve the search criteria.
Actually it is a little of both. One of the big advantages that Google has is their search business. They know where every, thing, person, place, dog, cat and car is. So yes it does take hundreds of millions of data points to make a good map search engine. The algorithms come into play after the data points are indexed. Apple is getting their data from numerous sources so I think the job of indexing, purge and merge is more difficult because the sources are disparate. The Columbia mistake is not an algorithm error, it is a typo in a database somewhere in the table 'countries' in the field 'country_name' that spells Colombia incorrectly, it is not spelled like the university. They simply have to find it and fix that typo.
Quote:
Originally Posted by Vaelian
Only most times they aren't known to the software house. Software released with known bugs is a sign of deception.
Only many times they are. You have software to ship to a deadline and a list of bugs you haven't eliminated. You have to decide if any are show stoppers. You've never developed serious software if you think this never happens. Not only with software. You're designing a car and there's a list of flaws you can't eliminate in time for the launch. Guess what happens. No one has ever engineered a perfect product; the better engineering teams get closer to eliminating all known flaws and minimising any that surface later.
The real question here is why Apple pulled the plug on Google without putting this out for extensive real-world testing. That seems to me idiotic when they had that option, especially if there were serious flaws known to the development team.
Quote:
Originally Posted by Lerxt
How pathetic to watch a bunch of people try to justify the crap quality of Apple Maps when compared to Google Maps.
So if one area is mapped poorly, all areas are mapped poorly?
Quote:
Originally Posted by hill60
iOS 6 Apple maps has voice guidance.
IOS maps based on Google maps didn't and still doesn't.
The End.
I think this is the primary reason most iOS users will probably find Apple Maps more useful for them than the previous maps app. It doesn't have anything to do with Google maps being more/less accurate, or listings searches being more/less complete, or Flyover being more/less useful than Streetview.. It's that users won't have to purchase a standalone nav app, with the free Apple version good enough for most.
What iOS Maps Troubles?
Oh yeah, I remember.
Some "remote islands" () (-DED) and obscure locations are having some issues, aren't they?
Quote:
Originally Posted by Mac_128
First I've heard of it. There's a problem with Maps? Who are these developers?
;-)
Really there is a problem with the maps? Just knew it now too! :P developers? ahhhh maybe Orange! lol
This is exactly like FCP X. Apple had to make a move. Historically, Apple's v1.0 products are NEVER good. It's not supposed to be good. It is a strategic move for Apple Inc. Of course, legacy systems such as Windows XP are much more reliable and developed. That is the advantage of using old, legacy products! Sometimes they work great.
Apple will fix this Maps app like they always do. Right now the people piling on are the sheeple. This is the same old Apple. Making a strong move with a v1.0 product. Expecting it to be polished is lunacy. This isn't Microsoft with a 5 year Beta period. If you have mission critical applications, by all means, don't use Apple. That's not what it is for.
So Apple make unreliable software and fixed it when people are complaining?
Originally Posted by bwik
Historically, Apple's v1.0 products are NEVER good. It's not supposed to be good.
Yeah… that's not…
Quote:
Originally Posted by Tallest Skil
Yeah… that's not…
Apple was never good
Apple bugs! It doesn't surprised me
http://maps.google.com/?ll=42.177653,-81.301575&spn=1.876624,4.205017&t=m&z=9
also shows the same in icloud's find my phone - but iOS 6 Maps app does show the correct label.
Quote:
Originally Posted by lilgto64
Hmmm - according to Google Maps - Lake Erie is now a Quarry rather than a lake.
http://maps.google.com/?ll=42.177653,-81.301575&spn=1.876624,4.205017&t=m&z=9
also shows the same in icloud's find my phone - but iOS 6 Maps app does show the correct label.
Click on "what's here" and it's labeled as Lake Erie in Google Maps. Have no idea what "quarry" is referring to tho.
quarry? well, it is a destruction in our mother earth..
Business is business.
I'm sure most of the main media knew too well before hand.
The media also seem to have got well and truly pre warned to never mention GPS being in 7" tablets no matter what.