you are definitely right 
After I think IPACS was not legally obliged to give access to a free SDK... wrong ? for now, we have access to a basic SDK for "free" and maybe there is a kind of "Premium" SDK for money (specific tools, compiler, etc.) so for those who can afford.
After, I did not inquire further with IPACS if a service like this is available.... Anyway I can not afford
I agree with you Trespassers, I look forward to having more tools and info to accellerate and increase the quality of our work ...
But on the other hand I can understand IPACS...
Subtle balance between "Openness" and "Exclusivity"...
the opening, allows the creation of a passionate community, to increase at a lower cost the popularity of a product and thus to ensure its development. On the other hand exclusivities with commercial partners assure the financing of the development and the perenniality of the product in the time ... if the product is not yet sufficiently popular and therefore widespread, to remain in the project the partners ask to being able to stay one step ahead of the freeware to be sure to have a return on investment on the sale of their DLC.
I think that IPACS considers that their product is not yet sufficiently mature and widespread and does not yet have sufficient support in the simulation market to let the freeware become a serious competitor for their partners. The risk is that partners like Orbx are going away if they have to increase development costs to stay over freeware offer and in the end they only sell a few copies of their DLC ...
I think that the concurrency Freeware vs Payware is healthy and allows to pull ever higher the quality of the DLC, but I can understand that IPACS make choices not always popular because they want to secure the future and the financing of fevelopments of their simulator at longer term. Indeed put in by the sale of the starting product, the sources of income can only be on the DLC, otherwise it would be necessary another mode of financing by means of monthly subscription or annual seasons like other games. But I am not not sure that this would be very popular either
Too much openness is also dangerous, as your reaction, like those of other first-hour users who become very impatient over time, proves it.
The risk is to demotivate the community, to scare away enthusiasts and consequently to lead to a decline in the popularity of their product...
For futur development, Aerofly needs both: strong community and strong partners. In short, good luck at IPACS to find the right balance. I prefer to be in my place than theirs 
Sorry i read one more time your script and i'm not sure it would change something... there
Is not so much object to create in your TOC file. I don't understand why it don't work.
Not at home this week but may be someone could test your to file on his computer !?
same problem if you enable splitgrid step in your script ?
Need to see CREATAFS2LIGHT script part please...
great result Thomas !
no tested yet but it seems to be the more accurate free data source for DSM for world coverage (pixel:30 meter but about 5m height accuracy):
it's DSM: Digital Surface Model so building & vegetation are not removed from data.
ALOS World 3D is a 30-meter resolution digital surface model (DSM) captured by the Japan Aerospace Exploration Agency’s (JAXA). Recently, this DSM has been made available to the public.
The neat thing about is that it is the most precise global-scale elevation data now. It uses the Advanced Land Observing Satellite “DAICHI” (ALOS) based on stereo mapping from PRISM.
you need to register to download data...
if you need 1.03 source may be ask directly to propkop2k2 here -> https://www.avsim.com/forums/topic/3…pdated-version/
If you need to convert from fsx, may be soon Bgldec will help you ? But it’s not finalised at that time...
Display MoreThe hardest part is defining the problem areas. They don't fall neatly into image tile boundaries and often have blended edges.
Once you know that the rest is relatively straightforward:
1) Create an image with the problem areas removed, transparency in their place
2) Create a transparent background image with just the problem areas
3) Measure the RBG / HSL of image (1)
4) Adjust RGB / HSL of image (2)
5) Put the images back together
This library would do all those steps, http://www.aforgenet.com/framework/features/
Maybe its edge recognition classes can find the problem areas too, we'll have to give it a try.
Exactly! This is the first step that is the most complicated: Segmentation Step
This is where Computer Vision comes in, you have to perform one or more treatments (Canny edge detector, or other...) on the image to identify two groups different by colorimetry. The human eye gets there easily, but that are necessary algorithms for a computer that I do not know unfortunately...
But they rely surely on the same algoryhtm principles as those that allow to detect a shape on an image (a face for example or building or road line for another, etc.) ...
some research to do again
If I'm undestanding this problem correctly, there is a color balance issue with existing and new imagery. I'm thinking of processing the color of the downloaded tiles with 2D linear interpolation between the color balances of the bounding areas. Then the new tiles should no longer 'stick out', no 'steps' in color balance of neighboring tiles of different sets.
Just give me some time, I have a ton of things to read/comprehend about the workflow/tools.
Balancing colors on grabbed sources like google or bing is the major problem!
Here are the results of my research on "How to" to follow if you wish to automate all this. it may not be the best solution but it could work..
for more info, do some search on google: "segmentation", "Color Matching", "Computer Vision", "OpenCV library"
Here is a simple example where the delimitation between the different colorimetry is obvious to the eye, it is very schematic and simplified:
Source image:
[Blocked Image: https://tof.cx/images/2018/05/02/8e29c8c2d63bc834ece921791bee160e.jpg]
first step: "Segmentation"
[Blocked Image: https://tof.cx/images/2018/05/02/6c296b169ae975eb668468571a9ff67a.jpg]
[Blocked Image: https://tof.cx/images/2018/05/02/60e7e56bd6e068735aad9de19e579e47.jpg]
Second step: "Color Matching"
[Blocked Image: https://tof.cx/images/2018/05/02/31610cff0e5523b0aaa0a2bef11e6d32.jpg]
[Blocked Image: https://tof.cx/images/2018/05/02/6eac839649f96f11f23bb4689781e77a.jpg]
[Blocked Image: https://tof.cx/images/2018/05/02/8ded261919f7e5d78e9c65bd58535a59.jpg]
my AFS2 class... not a tmc file parser sorry (not needed at that time)... if it could helps you, it's a little (or a lot) rough but maybe things to help...
Hi Nickhod !
Great news, great idea
I will try to give some answer:
- What file format are the TMC / AID files? It's kind of XML but not valid. Wondering if I can get a prebuilt parser.
... no really sure but I think a parser would be better.
- The readme in Geoconvert mentions that each downloaded image needs an AID file. The tutorial says it needs a TFW file. Which is correct? Does it need both?
... Geoconvert create automatocaly AID file from tfw, but finaly just AID file is necessary for geoconvert
- Do the downloaded files need any special names, or will each just be handed in filename order by Geoconvert
... any name, any number of files but may be better to respect AFS2 special GRID --> look information here: Image tile coordinates
- In the AID file, [steps_per_pixel] refers to the increment in latitude and longitude for each pixel from the [top_left] lon / lat value?
... ?
- I assume that Geoconvert is handling everything regarding topography data, or that's already in the simulator?
... you find already the answer...
- Anyone know what size tiles Ortho4XP uses? What size should I use? (Not the image tiles it downloads, the whole region you click on to download in one hit).
... look at here: Image tile coordinates
very nice screenshot 
... Why do you have problems with textures? These are self-generating buildings (AFS2 object) or generic buildings (AFS2 autogen) ?
humm interresting ! I thought I'd get rid of that but finally, after further tests I still have buildings flying on my scenery with a custom elevation 
"Don't forget your vomit bag please !"
A short video to see the advances on the autogen of my "Martinique" scenery... Sorry for distorsion & bad quality of video, I had trouble with the capture settings and I drastically reduce the amount of tree (directly on the .toc files) to avoid burning my graphics card (GTX 1060) ! but despite that, it still draws a lot more and more when I turn on the VR on-screen display to allow video capture ....
And don't care too much about my way of flying, I'm not a pilot and the more important for me is "flight feeling" and landscape watching
sit comfortably in the back, fasten your seat belt and don't forget your vomit bag please !
Good idea ! not much free time to spend. but I would be happy to help from time to time ...
thanks Thomas
I could not have said it with more poetry... 
as usual your advice was very good ... As advised I generate elevations for low level (7, 9,10). And after a few first tests it seems to work very well 
so a big thank you to both of you 
[Blocked Image: https://tof.cx/images/2018/04/20/654e22bfcb399e2757666df2e68aec6d.jpg]
