The colors are much better than some of the online aerial stuff I've seen. One Question -- have you tried this at an airport? Is it feasible to do away with airport ground-level scenery (runways/taxiways)? One problem might be "pictures" of aircraft on the ground.scott s..

Airports are a problem zone. Not because of aircraft images captured as part of the ground texture, but because of MS FS overlaying a gray runway on top of the photoscenery, hiding the realistic background below. Often this overlay is off by a few dozen or hundred feet, which looks awkward.And... oh.. terrain flattening sometimes has a really funny effects on airports, especially after installing a 3rd party mesh.So I guess I will have to create excludes for all of this using the SDKChristian

Matt, If my comments are not realistic, then please share your data? I haven't seen you post any concrete streaming numbers?Some critical things you need to post:1. Data rate for single hi-res request2. Routing3. Number of estimated concurrent users/connections4. Frequency of data requests (Aircraft @ 400Kts vs 120Kts)5. backbone/line capacity going to these serversYes the US internet is progressively getting slower and slower and slower (as demand is exceeding capacity) -- this is a very good article: http://web.media.mit.edu/~aggelos/861.html -- it's pretty technical but goes into the details of the problem that I'm being VERY realistic about.There are plenty examples of "innovative technology" that has gone no where for many reason -- should we stop trying, absolutely not, but ignorance of existing limitations doesn't help anyone.Rob.

A lot of people are looking into streaming, including Apple's HD/DVD on demand -- but that's exactly the problem, a lot of companies have bigger plans to consume more bandwidth -- hence the involvment of congress and the so called "free internet" debate. Bandwidth isn't some infinite resource, but the real issue is routing of data.This explains the limiations (older article, but the content remains valid to this day):http://web.media.mit.edu/~aggelos/861.htmlBut like I said, I think we'll solve this problem other ways as they can and will be better performing solutions with a more realistic experience. The distributed process/data model is still very much the better method of implementation -- continuous streaming data is a bad long term concept.Rob.

Not trying to place blame on anyone, just the fact that demand for bandwidth and data routing (which is the real issue) is out pacing our ability to provide it. So much so that Congress wants to get involved and the so called "Free Internet" debate -- which in itself is sorta silly.good readhttp://web.media.mit.edu/~aggelos/861.htmlRob.

>Matt, >>If my comments are not realistic, then please share your data?> I haven't seen you post any concrete streaming numbers?>>Some critical things you need to post:>1. Data rate for single hi-res request>2. Routing>3. Number of estimated concurrent users/connections>4. Frequency of data requests (Aircraft @ 400Kts vs 120Kts)>5. backbone/line capacity going to these servers>No need. Google Earth proves that it's possible to stream high resolution imagery plenty fast enough for 500 kts.>Yes the US internet is progressively getting slower and slower>and slower (as demand is exceeding capacity) -- this is a very>good article: http://web.media.mit.edu/~aggelos/861.html -->it's pretty technical but goes into the details of the problem>that I'm being VERY realistic about.Who cares? A few thousand FS users streaming imagery would probably use 1 billionth of 1 percent of the total internet capacity.Matt

Matt,You really should gather up your facts -- 21 million copies sold with some of the more popular 3rd party add-ons selling 60,000 - 100,000 copies. That's a little more than a few thousand.I can do over 185 mph from home to work on my motorcycle at 2am, I can do maybe 30 mph at 5pm. It takes more than one product demonstration of feasability to make it a working reality.Rob.

If you fly at high altitude (typical 11km or 33000 ft) then my program would stream lower resolution tiles, as it uses the euclidean distance to the ground tile as a measure for the required resolution. No need to pull 1024x1024 pixel tiles when flying high.But I admit the streaming performance is not yet good enough. My program lacks multithreading capabilities and reads tile by tile sequentially. There is no look-ahead streaming along the flightpath YET...Christian

>>The whole point is that this is an innovative technology and>I'd bet money that we'll see it implemented (either directly>by MS or by a 3rd party) within the next 5-6 years, if not>much much sooner. >>MattI even use Google earth to develop and add some realistic beaches to my country :-lolGreat stuff...Andr

Do you cache the images? I know the images for my area are pretty static, unfortunately some images are 6 years old while others are 2 years old which leaves some odd connections (roads that suddenly appear, housing developments truncated).The problem with swapping from low to high res textures is the obvious switching that can be distracting from the realism -- even with several layers the switch is still pretty obvious. How many texture layers are you using 1024, 512, 256, 128?I have been working on 3D object building based on 2D sat images, but without good matched elevation data it is a low success rate to extrapolate the object. These folks have what I need for accurate mesh: http://www.iracing.com/technology_scanning.htm -- good down to 1 inch! But this is focused on auto racing simulations where 1 inch can make a difference.Rob.

>You really should gather up your facts -- 21 million copies>sold with some of the more popular 3rd party add-ons selling>60,000 - 100,000 copies. That's a little more than a few>thousand.>Are you sure about 60,000 to 100,000? I've always heard that it's about a 10th of that amount.

...Danny

>Matt,>>You really should gather up your facts -- 21 million copies>sold with some of the more popular 3rd party add-ons selling>60,000 - 100,000 copies. That's a little more than a few>thousand.>>I can do over 185 mph from home to work on my motorcycle at>2am, I can do maybe 30 mph at 5pm. It takes more than one>product demonstration of feasability to make it a working>reality.>>Rob.How many people are playing flight simulator at any one moment in time??? I have no idea. Even if it's 100,000...or 500,000, it's still not going to have any kind of noticeable effect on the internet as a whole.This whole arguement is irrelevant to the point of the post. So to sum it up. I think this is a cool technology and you think it's going to crash the internet. Fine. Lets just agree to disagree an move on to something more interesting. If you want to have the last word, feel free.

Anything from 4x4 textures to 1024x1024 - the choice is up to FSX which one it loads. My project tries to provide textures in the resolution which it estimates that FSX will load.There is some popup effect when a blurry texture suddenly becomes crisp. I wish there was some blending applied (like e.g. for 5-10 video frames).I also think that realism could be improved by applying bump maps or per pixel displacement maps (possible with DX10). Maybe then houses could even be "stamped out" of the background. And small scale terrain variations would become more three-dimensional. The shadows in the satellite data could be used as an indicator of threedimensional structure to generate those displacement masks. Ideas, Ideas...

As far as what I've been working on and developing (not really to be sold, just as an experiment to see what I can come up with).Critical to know when (time, day, year) and where (orbit) the sat image was taken. Using this data can help with the accuracy of the object's shadow and produces better results for the final 3D object.1 pass is required to just "normalize" the image data (color, saturation, etc. etc.) and produce an indexing file which can be used to match up elevation data but I'm finding elevation data to be very unreliable or just not accurate enough and there is no way to obtain Military spec data -- unless someone has a source?Sat orbits:http://science.nasa.gov/Realtime/JTrack/Spacecraft.htmlSat data from:http://www.globexplorer.com/?gclid=CKzCurG...CFQvlYAodLhru7wBut with over 450TB worth of satellite imagery it will require some serious processing power to go thru that much data to produce a detailed and truely accurate 3D world.But keeping the polygon count simple should still produce a flyable sim (be it FSX or XPlane or whatever). What I don't know is texture load and what is realistic, the 8800GTX has some impressive abilities for compression that "might" just make it possible to retain a moderately high quality STATIC texture map. Actually probably two 8800GTX sharing the 1.5GB (of video memory) which I "think" can be done. But ATI's next DX10 card looks to be even more impressive and programmable.Unfortunately the FSX engine is pretty weak/dated, this is where XPlane engine is considerably more robust (in some areas, not all). Of course I'm exploring both but the look ahead texture compression and object modeling will need more than one CPU and hence why Xplane is more attractive for what I'm trying to accomplish. But who knows, if the Ace's dev team re-work FSX it might be work better with FSX.Rob.