I was quite happy the last few weeks that hardly any crashes were reported in my bug tracker from my tools. But today I found out why, the bug reporting is actually broken at the moment. It seems the interface into the bugtracker does not like the newer PHP version that i had to enable for using the latest version of WordPress for SceneryDesign.org.
So now that I am aware of this issue, I’ll try to fix it. Probably the best way forward is to update the bugtracker software to a newer version that will hopefully give a working interface with my tools for newer PHP versions. So let me try to find some time for that upgrade soon.
Edit: the bug tracker software has been updated now and is running fine again. So you can report bugs from my tools again. Although I hope you don’t do it too often….
That autogen objects have a distortion in their rotation is something we found out 6 years ago already. But until recently I was not aware that not only the rotation of the objects is distorted, also their sizes can be different then you would expect from looking at the objects in the Annotator tool. I guess most of use work in the zones around N40 and S40, where the distortion is minimal, so it took a developer who works outside of that area to inform me of a bug in scenProc that resulted in objects being rendered in the sim with wrong sizes.
So what I did is make a test photoreal scenery where I burned polygons with a known size in the photoreal. Next I placed autogen objects and tried different settings until I could get them to align correctly for each latitude where the objects are placed.
Below you see a screenshot of this test scenery at a latitude of 60 degrees, where you have quite some distortion already (around 41 degrees latitude the distortion is minimal, due to the way the autogen tile sizes are designed). As you can see for some polygons the rotation is off from what you see in the photoreal, but also the sizes of the objects need to be defined incorrect, to have them show correctly in the simulator. Some of the buildings need to be made longer and more narrow for example.
The good news is that you don’t have to apply this kind of rotation and size corrections by yourself. scenProc has been modified to do it automatically for you now. So with the latest development release you will find that the output generated should align more accurate with your photoreal scenery.
How does it work? In Annotator you are seeing your photoreal shown in a latitude and longitude projection. But the heading of the objects should actually be defined in the AGN file for flat earth projected version of the object. And for the size it is even weirder, the sizes are defined in degrees as if the object has a heading of zero. But even when the object does not have that heading, which means you need to provide an incorrect size in degrees to still get the correct size of your object in the simulator in the end.
And just as the proof of the pudding, below is a screenshot of the test scenery in the simulator, where you see things align well.
The next development release of scenProc will include two new steps for placing points: LineToPoint and PolygonToPoint. In the future these steps will replace the existing steps to place points along lines or inside polygons. As i realise some of these steps are used in many of your scripts, I will not deprecate the old ones very soon. But for new scripts I encourage you to use the new steps.
Why did I make this change? There are a couple of reasons:
When I add the LineToPolygon and PointToPolygon step a while back, the idea was already to have similar steps for other conversions between feature types. So the new LineToPoint and PolygonToPoint steps are in the same philosophy.
Until now there were multiple steps to place points along a line or inside a polygon. Each with slightly different characteristics, but also with a lot of overlap. So it is easier to have all this logic in one step.
The manual has been updated to reflect these new steps as well. Both in the description of the steps and in the sample scripts.
The latest development release of scenProc makes a set of new steps to create terrain vector scenery generally available. This means that you can now use scenProc to make terrain vectors with the shp2vec tool.
The manual contains a sample script that shows you how you can make terrain vectors from OpenStreetMap data, but of course you can also tune this script to use the attributes of your vector data.
You can use this functionality for example to create freeway traffic BGL files for your scenery based on a road vector data.
I have just made available the new development release version of scenProc. This takes the tool from version 2.0 to version 3.0. This new version has been in development for quite a while already, so it includes many new features. To main improvements are:
Polygonal features with holes are supported much better. This means that importing and processing features with holes goes much faster.
It is possible to create XML bridges now in scenProc.
The texture filter editor has been redesigned completely, so this means that the feature detection is much more powerful now.
The scenProc Batch Runner tool has been added, which replaces the old tool to create batch files. This makes it much easier to run a script in batch mode.
And there are many more improvements. I have also updated the entire manual, so please check the manual first if you want to learn about the new features. And else feel free to ask questions in the forum. I hope you enjoy the new version.
I have just released a new stable release of scenProc, this new stable release is version 2.0. It has the same functionalities as yesterdays development release. You can find the download links on the scenProc page.
The main reason to release this new stable release now, is that the version 3.0 development release is just around the corner. I hope to have it released in about a week from now. Since version 3.0 includes many changes I thought it would be a good moment to have a new stable release first.
I’m updating the Assimp library that ModelConverterX uses to a recent version and in the process I am also trying to read skin and bones animations from other formats using this new version.
I don’t think I have it all right yet, but this video was too funny not to share with you. It directly reminded my of Monty Python. Enjoy and hopefully soon I get the skins and bones to read correctly.
Cough, cough, man this blog is dusty. I haven’t written here for much too long.
I would like to wish all the users of my tools the best wishes for the holiday season and for 2020!
I am working on a new update for ModelConverterX that adds support for additional formats. The screenshot below shows a snowman loaded from the glTF 2.0 format.
At the moment the Ground Polygon Wizard in MCX only supports seasonal textures when you export using the FS2002 style code. But inspired by this thread on FSDeveloper I have started now to add support for seasonal textures to P3D v4 MDL files as well.
I have some success already with using visibility conditions to switch different versions of a modelpart, where each then has a different texture. And I plan to experiment with using material scripts as well.
My plan is to support this both in the GPW, but also for normal 3D objects. The user will just have to indicate that seasonal textures are desired and specify the season settings for the specific region. MCX will then on export take care of inserting all the code that is needed (either visibility conditions or script based).
Just wanted to let you know this is coming, I still need some time to change the GPW to support these new options though (underwater I need to cleanup some of the code in the old FS2002 style output as well to keep things organised).
The FSElite website has been running a developers month over the last weeks and I’m one of the developers that has been interviewed. I thought it was nice to give some insight about developers that support other developers as well. You can read the interview at the link below.
SceneryDesign.org 
