Mesh cells shrinking?

[Guest adriano]

Hi,I

[rhumbaflappy 113]

Hi adriano.This has been the way of the world since Flight Simulator began using mesh landclass and photoreal ( FS2000 )... and it portrays the world's sphere-like appearance.The data given to resample is measured in geographic terms, not in terms of linear measurement. So our equiangular data will look odd near the poles or the equator, as it follows lines of longitude and latitude instead of so many meters per pixel. The whole "4.8 meters per pixel" thing is an approximation. It never really existed at all.It's the difference between measuring in fractions of degrees, rather than in meters. There isn't a problem with LWM or VTP data, or with properly projected photoreal, as they are placed or resampled accurately according to longitude. The default terrain textures do look distorted, but there really isn't a feasable cure for that.Dick

[Guest roto5]

Hi Adriano,The tower changing its size along latitude IS expected in FS.This effect occurs because a scenery object in FS has its X, Y, Z coordinates (vertices) defined in an rectilinear space.All the vertices of the objects are defined (by the designer) in X, Y, Z coordinates, not in geographic coordinates. For georeferencing purposes, i.e., to place the object in the right geographic location, this rectilinear space is centered in a geographic coordinate (latitude/longitude), which is defined by the scenery designer as well.Because the visual scenery of FS (i.e. what you see when the simulation is running), is based in a spherical coordinate system, FS transforms the object coordinates from the rectilinear space to a spherical space, while keeping the X,Y,Z coordinates (from the rectilinear space) aligned with the geographic grid. This transformation deforms the object shapes, specially when the objects are placed at high latitudes.Regards,Emerson

[Guest adriano]

hi Dick,well, now I have it clear: landscape IS distorted. Nevertheless, I'm not sure that VTPs and textures are immune from this. I inserted a Cellgrid at the same locations as my towers, and I obtained the same amount of distortion. This is consistent with the fact that all landscape features insist on the same system of coordinates. The exception probably are GMax-like objects, as Emerson points out, but I didn't check that yet.It may be that pre-distorting the landscape the other way round can actually compensate for the effect. But I think one must do that with every element, mesh, textures and VTPs together, otherwise the remedy will be worse than the disease. Will give it a try.Cheers,

[rhumbaflappy 113]

Hi adriano.LWM and VTP polys and lines are still 255 and 256 points per LOD13 Area... the points are just closer together longitudinally near the poles. They only represent geofererenced points.Landclass textures are the same. 256 points in longitude, they are squeezed together longitudinally near the poles.If you take a flat map with gridlines, and try to wrap a sphere, the points of longitude must squeeze together at the poles. That's your distortion. It's very evident with a globe of the earth, with the lines of longitude.Objects, runways, and autogen are made as measured items, so they cannot be distorted. But place objects .002 degrees apart longitudinally, and they will crowd together at the poles... just like in real life.As far as mesh. Mesh is actually measured as elevations of points of lat/long. That is already accounted. The distortion actually is on the flat source, not on the sphere. A mountain peak near the pole is correctly placed and shaped if your source is right... but the landclass textures covering it will be distorted.In order to account for landclass texture distortion, each longitudinal band of LOD13 Area size would need it's own unique texture set. Maybe possible in another 10 years, when we have media and memory of terrabyte size.Dick

[Guest adriano]

Hi Dick,it took me a few days to make some more experiment and get a clearer picture

[rhumbaflappy 113]

Hi Adriano.Photoreal sources must be reprojected to geographic projection ( as you found, equiangular ). Then resample slices them correctly, and they should not be distorted. All this is in the FS2000-FS2002 scenery SDKs.Dick