Jump to content

Matching Google Street View Images - Best Practices


Matt McDonald
 Share

Recommended Posts

Increasingly, I am being asked to place building rendering in Google Street View images to determine visibility of the project. I am given very little information on existing or proposed grade, I will get a spot elevation for finish floor but that is about it.

 

I then try and determine elevations from the Street View image based on the elevation indications in Google Earth. From there I try and match my field of view as best I can by modeling some of the curbs seen in the satellite images.

 

I imagine we can all agree this is not a highly accurate process but I don't have much say in the matter.

 

I'm curious to hear what tips and tricks others might share re: matching Street View images.

 

One of the things I want to look into is determining the lat & long of the Google car when the image was taken and see if the Civil might be able to provide me with lat & long of the proposed project. That would be a start.

 

I also wonder if I can't model a camera rig in Max which simulates the camera ball on top of the Google car.

 

So...a couple of thoughts there. If you have anything to share, I'd love to hear it.

 

Matt

Link to comment
Share on other sites

I usually tell people that the in the time it takes for me to put the building the the street view (which I'll be billing them for) they can fly me first class to the location, put me in a 5-star hotel, expense my meal at the best restaurant in town, and let me take my own way more accurate and high res photographs of the location.

 

Using Google maps images and 3D models is pretty close to jamming a square-ish peg in a round hole. You'll get it in there, but it will take a lot of jamming and cussing to do so.

 

If you have to use them, I use this tool to grab the high res version of the street view (http://www.purebasic.fr/english/viewtopic.php?f=27&t=50248&sid=239fbe4d34f9c5cfdd4f2dd42c57ea9a) The legality of the tool is highly suspect, so use at your own risk.

 

For me, the best practice is to not do it at all. It's a lot of effort for very little in return. Most of the time it's far easier if you grab a Google Earth map, draw quick outlines of the surrounding buildings, extrude them to rough heights, and then place your new building in there.

Link to comment
Share on other sites

I sometimes use Google streetview to get basic data on an extended site. I just screencap/paste into Photoshop and correct the perspective. I agree that it would be better to just go shoot it myself, but often don't have the time.

 

Scott--thanks for the link. The funny thing is, the intersection the guy uses for an example is just down the street from where I've been working, about 1/4 block away.

Link to comment
Share on other sites

Besides all the legality involved, you need to consider that those pictures are taken with a very wide angle lens, then they are stitched and adjusted to fit their virtual 3D environment, after all this process the exactitude of those image is very limited, the barrel distortion alone can trow off several feet away of the real dimension.

You should use just to eye bole some schematics and draft presentations, but try to make exact visibility studies it is far from reality.

Link to comment
Share on other sites

- I understand that attempting to photomatch google street view images isn't a good idea. It's an argument I have had and lost.

- It would be nice to go and shoot proper photos for this process but there isn't the time or budget for it.

- These renderings are not anything beautiful. They are for a retail operation which is evaluating potential sites. The retailer wants to be sure the building is visible from the road.

Link to comment
Share on other sites

I good way of scaling photos is to use car parking spaces which are all usually a very standard size (2.5x5m in the UK) and from that you could establish a scaled plan in order to model the buildings relatively accurately, using maybe a few front building textures taken from street view. Once you have modelled the immediately surroundings and the new building you could then move your camera to match the existing building locations in the street view. This is probably as accurate as you can be. But definitely you need to be stating a disclaimer in your contract that the views are approximate only and not 'verified views'.

Link to comment
Share on other sites

Which version of Max do you have? In the newest version there is a match perspective tool, that will help you align your model to the perspective of the image.

 

Good luck with using perspective match with google street :)

 

BTW it would be nice if we would be able to recreate whole scenes by using this technology: http://www.patriciogonzalezvivo.com/2014/pointcloudcity/

Edited by mali22
Link to comment
Share on other sites

Generally you can get a rough but usually good enough idea of context by navigating around a site in street view. Generalize it as much as possible, keeping priority only on the biggest features of context such as neighboring buildings, roads, sidewalks etc.

 

Those big features are what the audience will be looking for. Math and science will get you close at best, you must use your eye and Photoshop skill to fill in the inevitable holes.

 

Photocomp Street View.jpg

 

Public.jpg

 

If there are concerns with accuracy, it will require a site visit to get the photographs in person and specifically for this purpose so that you can understand the context completely.

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
 Share

×
×
  • Create New...