New Los Angeles Imagery

Updated Google Earth imagery makes 3D models look better

Today I noticed an imagery update to the Los Angeles area where some of my Google Earth models are located. In my opinion it is an ideal imagery update, for several reasons:

1) Color correction. The color is beautiful and true to life. It looks like you are really there. My own models look much better as a result of this new backdrop.

2) Clarity. The imagery is fairly sharp.

3) Angle. These pictures were taken for the most part from a straight-down point of view. This helps in making the models and creating a realistic atmosphere.

4) Consistency. Covering a large area (I haven't found the edges yet) this new imagery is seamless and again, helps to make the area look very real.

Footloose Part 2

Recently I wrote about a new, automatic feature in Google Earth that helps keep your building models aligned to the ever-shifting and updating satellite imagery beneath it.

Sometimes you may get a message on your model preview page in the 3D Warehouse which reads, "Google Earth imagery has been updated. Please check your building position and update as required."

Psst! Hey! What's A LiDAR?

At the Super Modeler Conference last month in Boulder I heard the word LiDAR being tossed around a lot. Not wanting to appear stupid, I whispered to my seated neighbor "what is a LiDAR and where can I get one?" It sounded quite important. As it turns out, I was right. Jonathan Rumohr, one of the Super Modelers also known as WMU3D, offered this whispered response:

"Put simply, LiDAR stands for Light Detection And Ranging. Basically, what happens when somebody wants imagery for their area, they can hire a LiDAR flight. It's just a small plane that goes out with a fancy camera that shoots light rays (laser beams) down at the ground and bounces them back up. No, we're not talking any Star Wars photon beams or anything. Everything is up ultraviolet to infrared range. Of course, these light rays measure the distance from the ground, back to the plane. This, accompanied with the GPS on the camera, allows for the creation of a virtual "cloud" of data points

"Once back on the ground, these clouds are looked at, and combined using software into what is called a DEM (Digital Elevation Map). From there, you can get several different kinds of DEMS. Since there is so much that the light rays can bounce off of, they are able to pick up things from tree heights, building heights, and sometimes even dense clouds. Therefore, it is possible to end up with three different DEMs for an area. One that is a tree-line elevation, one that is a building elevation map, and one that is strictly a terrain map. Those different anomolies are all taken out by software gurus way smarter than me! Just think of someone dumping a bunch of points or tiny lines in SU, and then trying to determine what they mean! Doesn't sound fun to me!

"Once your DEM's are created, that's where the SU fun begins! From those DEMS, we can make contours which we can then model, and submit to Google Earth. Unless the individual user has some pretty pricey equipment, the average SU user won't have access to a raw DEM, even though that would be the format that Google would like most. That's where the cities and municipalities come in. Most cities have a Geographic Information Systems (GIS) department these days. They would be able to submit that raw data for Google to use in large scale terrain maps!

I'm currently working with the University and the surrounding area on getting more of our recent DEMs to Google. Hopefully to them within the next month or so. It usually takes Google a bit longer to upload that data to GE, than the standard one week in the pipeline!"

Thanks, Jon!

NOTE: In some cases apparently the LiDAR scan used in a municipality seems to be a low-resolution building elevations scan. In some cities this appears like gentle lumps scattered across areas that should be flat. Is that correct? Maybe a viewer knows.

What Makes Geo Modeling More Fun?

Just thinking, as a general statement, the more accurate Google Earth is, the more people might be motivated to model. For example, higher resolution topographical data; more recent, and clearer, satellite imagery (especially "parallel projected" or high altitude imagery). Also better Street View images-- the clearer the better. I find myself avoiding modeling in areas where there may be an upgrade anyway. The more tools (and better tools) we have to model, the more it makes it worth while to model. And the more fun it is too.

There Is A Cloud Over Lima, Peru

This week many parts the world woke to a pleasant surprise: the latest Google Earth satellite imagery push. The folks at Google were hard at work on that, so much so that they had to delay the normal 3D buildings release this week (but then they surged ahead and we'll have that push anyway!)

Argentinians blogged about how 80 percent of their nation was now available in luxurious high-res. Americans anxiously scoured the corners of their nation to see what was updated. But for an unfortunate few the imagery push was not what they hoped for this time.

Modeler and fellow SketchUp Islander Auquicu informs me that parts of Lima, Peru, were updated with older and lower res images. There was also an imagery shift that left many of the already-modeled 3D buildings slightly off to the side of their foundations, and last but not least... parts of the beautiful Peruvian coast which were previously clear, now include cloud cover in the satellite photos.

Sorry to hear you've had a bad day, Auquicu. I'm sure that updating imagery for Google Earth is a very complex process, and these things are bound to happen. Hopefully the kind people over at Google will notice soon and fix it!

Building An Observatory: Observations

Time for some good, old-fashioned SketchUp behind-the-scenes stuff. We're going to have a brief look at my model of Griffith Observatory in Los Angeles, California. Calm yourself. I know it's exciting but get a grip. You may have noticed the Observatory in the first Transformers movie with Shia LaBeouf:

For this project I was lucky enough to find some floor plans online.  At the time, the Google Earth terrain in Los Angeles was very wonky and it would have been hard to get a clear "footprint" of the building. The actual hilltop site has been leveled off since 1930 or thereabouts, but our Google Earth hilltop was very rounded like... a hill (now the GE topographical data is very high resolution and the hill is manicured into a flat lawn, just like in real life, but not at the time the model was made).

I began the project by importing the floor plan into SketchUp as a flat image. I was then able to easily trace the foundation and extrude the shape to the correct building height and add other things like turrets, domes, and copper roofs.

Model was built right on top of real floor plans

As the model progressed, the original floor plan disappeared bit-by-bit, to be replaced with glorious photo textures which add so much realism to the final product. One thing had me stumped-- how was I to texture the three large domes on the building? In real life they were copper with a beautiful patina and other signs of aging and character. (One houses the sun scope; another the telescope; and the biggest of all looms over top the planetarium). How would I take a photo of a curved dome and apply it to a bunch of flat surfaces (because, if you don't know, everything in SketchUp is flat-- even "curves" are really just a bunch of flat chips put together).

My solution was to create a completely false texture, patina and all, and project it downward over the domes. Imagine breaking an egg and watching it run down the sides of someone's head... or shining a flash light down on a ball and watching its light touch all parts of the curve right down the sides... basically that's the idea. Here we can see the texture that I created in an image editing program, and the way it was used to create a fairly realistic-looking aged copper dome in SketchUp:

My own flat painted texture on the left; finished 3D dome model on the right

The copper joints running around the dome were made simply by creating additional, horizontal bands of geometry and painting them with a flat dark brown color. But, this created a lot of excess polygons. I have some other solutions to copper dome-making that I want to try out. I'm actually re-doing the Observatory model, and what I have so far (thanks in part to much more accurate topographical data from GE) is about one-and-a half-thousand times better than this.

Plus, practice makes perfect. This was my first really complex model, and I did it within a couple months of ever having tried to use SketchUp. I was pretty proud of it and felt in part that I was making a little piece of history. I am anxious to see how the improved one will turn out...

EDIT: I had used SketchUp as early as a couple years prior to this, but I made this model within a few months of using SketchUp in earnest.

BAD Terrain Data! NAUGHTY Terrain Data!

What happens when the Google Earth terrain data doesn't cut it?  The satellite picture is fuzzy.  The skyscrapers seem to be lying flat across the ground.  A hill appears when there is no hill in real life.

You want to make a 3D model that integrates with the terrain nicely, but you can't! What do you do? Do you surrender? Or do you FIGHT?