WOW, now I'm jealous.. Happy christmas :) :SPPL - Algoa Flying Club (Port Elizabeth, South Africa)FS5, FS98, FS2000, FS2002, FS2004, FSX user :)

One more update: I've been testing my project inside a virtual machine (VMWare Player) running on the second CPU locally on the flight simulator PC. The performance is not all that great, I think that is because VMWare does not find any hardware-assisted virtualization feature in my Dual Athlon MP 2600+ CPUs. So a lot of the virtualization has to be done through emulation in software. The solution with Mac Mini attached externally appeared to run faster.I will try Microsoft's Virtual PC solution as well when I find some time.But I'll have to get married first. ;-)I also tried to boost speed by creating a cache for decoded JPEG tiles, such that a tile does not have to be decoded multiple times when used more than once. This is the case for tiles further away from the plane, where a single 256x256 pixel tile from the Internet is enough to supply data for hundreds of Flight Simulator BMP tiles in the far distance. But the benefit of this optimization did not quite cut it - I think my main problem at this time is Samba performance. Accessing tens of thousands of small files is simply going to take time. I may have to abandon a Samba based solution if I can't get this any faster.

I've got a laptop with internet connection networked to my desktop PC which runs FSX.. Would that be benificial to having the whole load running on the single PC?PPL - Algoa Flying Club (Port Elizabeth, South Africa)FS5, FS98, FS2000, FS2002, FS2004, FSX user :)

The more I experiment with the Samba based solution of the tile server, the more I see that this may NOT be the way to go for a "professional/commercial quality" version of my hack. The performance is simply not satisfactory.I will try to looking into writing a Windows File system driver, ... if time permits. The part that interacts with the Internet will be running as a user space driver to lessen the impact of bugs (no bluescreens).I also tried installing Microsoft's Virtual PC to compare performance with VMWare Player. Both the Virtual PC 2007 Beta as well as the Virtual Server R2 SP1 Beta2 screwed up my Windows system so badly that Internet networking was gone and Windows required a reactivation. Even after a system restore, Windows *still* wanted to be reactivated. If that reactivation had failed, it might have killed my entire project.

####, good luck with things, keep plugging away!And why isn't this generating more interest? The most exciting breakthrough since AI aircraft and ATC for mine..PPL - Algoa Flying Club (Port Elizabeth, South Africa)FS5, FS98, FS2000, FS2002, FS2004, FSX user :)

Everytime I post new images in the main forum, people get pretty excited. For now, I'd rather like to keep a low key approach ;)

Well when something is that amazing of course we are going to get excited! :)

Any New News??? I'm so Excited about this.... The Pictures are Awesome!!!!!!

Yes, there are some news.There will be a semi-public showing of this project at a closed user group at the University of Applied Sciences in Karlsruhe from the 12th14th of January. The gathering is called MeKa 2007 and privately organized.I have done a lot of work to better support MSN Virtual Earth. As I have mentioned earlier, starting at a certain zoom level the Virtual Earth falls through to black & white USGS aerial survey data. I have created an algorithm to colorize this data in realtime with the color information taken from lower zoom levels. I convert the color information into HSV color space and copy the Hue and Saturation components to the higher USGS aerial data zoom levels.Some VERY obvious color artifacts occur when the tile carrying the color information has a low brightness (in dark shadowy areas). That causes that the color copied to the higher zoom levels appear mostly random - especially at places where the zoom levels don't align perfectly align with each other. So the the shadow in the lower zoom level may fall on some brighter area in the high zoom level causing that part to appear in weird color.The overall effect is that the Grand Canyon looks like one of the Sony Bravia TV ads - Lots of random color specks all over the place! Duh. But overall the detail level that I obtain from MSN Virtual Earth can rival that of Google Earth, except for these color artifacts which I have to keep working on.I am not really satisfied with the performance overall (loading times and keeping up with faster airplanes), so I got myself a download of the Windows Driver Kit (2.3 GB!). I will be looking into creating a driver that sits on top of a Windows file system instead of using Samba. I cannot promise anything at this point - it would appear that this driver project alone is a 2-3 month thing due to the complexity involved.Oh, I finally got Microsoft's Virtual PC 2007 Release Candidate to work, but the I/O performance on the virtual hard disk is abysmal when running Suse Linux as a guest OS. I will do a speed comparison with VMWare Player to see which one is better suited to hosting my Samba based tile server project.New screenshots should be up soon. I have created some screenies of the default flights, e.g. Friday Harbor and some of the missions.Cheers,Christian

I solved weird discoloration problem when coloring black and white USGS data! The Grand Canyon looks much more sane now. Remaining issues with the USGS data are:- it is brighter in most places than the lower zoom level data: so we need luminance adjustment- contrast varies from USGS tile to tile (the pictures were taken at different time of day)- vignetting effects are visible (different brightness near the edge of the lens used for the photo operation)- the USGS data seems to be dated (1989 maybe?)I plan to counter these effects by applying a brightness and contrast adjustment quite similarly to copying the color information. Of course the old age of the photos can't be changed by image processing...Because of performance considerations I have started to begin phasing out Samba in this project, however Samba based development will continue in parallel:I just got some samples from the Windows Driver Kit to compile and run. I can now monitor file accesses to any Windows drive in realtime. And that would include accesses to Flight Simulator's terrain tiles. However Windows kernel mode driver development can be dangerous. One mistake and you bluescreen the system.I cannot (yet) cause the any I/O operation to block to wait for an Internet download to complete - this would require an established two-way communication between the kernel component (minifilter driver) and the monitoring application (a userspace program). Right now this developer sample only provides for a one way communication between filter driver and monitoring tool. I do not know if a Minifilter driver is even allowed to block an I/O operation for an extended period of time.Anyone willing to sponsor this work commercially? I am in the process of changing jobs anyway. I might as well turn a hobby into a profession.Christian

Regarding the Windows Filesystem filter driver: Overall the Windows NT I/O model and filter model has an awful level of complexity to it. I begin to see why Microsoft employees are paid so well - and why they burn out after 10 years. ;)But there is a simple "virus scanner" example driver available in the WDK that should get me started pretty quickly. This driver passes control to a user application when opening particular types of files (e.g. BMP). This is ideal! Instead of scanning for "viruses" in these files, my user space program would try to download the terrain tiles from the Internet instead. And I can reuse a lot of my existing code for this.This could all be done much faster than I originally thought. I expect this to work within 2 weeks and performance should be an order of magnitude faster than that of Samba.There are a few open questions still, but I've got an plan of attack already.Christian

I can't wait to hear more about it!

>Anyone willing to sponsor this work commercially? I am in the>process of changing jobs anyway. I might as well turn a hobby>into a profession.>>Christian>I'm a poor South African student, but I'd sure as #### pay money for your product if it was released. I think it would be a tragedy if all your time and effort A) Was never made available to other FSX users or :( You ended up being out of pocket because of the project..Good luck with your searchPPL - Algoa Flying Club (Port Elizabeth, South Africa)FS5, FS98, FS2000, FS2002, FS2004, FSX user :)

Thanks for the kind words.Unfortunately, for Africa the satellite data which is published online probably covers only a few major cities in good resolution, the rest will look good only from a very high altitude.One major problem remains: I do not own the rights to the satellite data and making this project available publicly without prior authorization by the rights holders would make me liable for ... probably more than the the MPAA would ever ask for if I had put their entire music catalog on filesharing. They might make me pay for an entire satellite launch I suppose ;)So let me just spend a few more weeks of my time on this until I've got something so convincing that it may get the big players interested.Christian

Obviously this is an awesome project but it tickled my curiosity in regards to how it interacts with the other FS data?Do runways and airport buildings still show? Does the 'freeway traffic' still drive along across the textures etc? (i.e. are the things that are normally kept when adding a phototile using more traditional methods still visible using this method?)Are you streaming the google/V earth terrain data as well or just the images?Just curious :)Best regards and best of luck with the legal aspect of it.Tim