VTP vs LWM coordinates continued...

[Guest christian]

I had a bit of a play over Easter and gained the following insights:- The VTP and LWM numbers are identical to pixels! This may have been obvious, but I didn't realise this beforehand. This is why the values don't have to be more accurate, since you can't really be more accurate than a 'ground pixel' in FS2004.- Chris Wright suggested that adding 1.5 / 2 to the VTP values would match the LWM values. I found that 1.7/1.8 works quite well (more testing to be done). This suggests a pixel offset of just less than 2 of VTP compared to LWM. Indeed, I'm pretty confident now that this is a constant offset, and not caused by the 255 vs 256 effect as suggested by Rhumbaflappy. The offset doesn't seem to vary over space. (I'm not saying that this effect doesn't exists, it may very well, but there is a more dominant offset that causes the mismatch). HOWEVER, this raises the question which one is wrong, VTP or LWM? Chris Wright suggested VTP, but my experiments suggest the opposite. When I shift VTP polygons, my runways aren't centered in the airport background polygons anymore. Also, VTP roads along the runways lineup nicely when not shifted. Since my runway coordinates are taken from my GIS system also, everything has to line up correctly in FS2004.- This leads to the suggestion that the LWM coordinates are out by nearly 2 pixles, for whatever reason (This may be FS2004 only, I haven't tested FS2002). One idea I had was that the LOD13 values in the Terrain SDK table are rounded and not 100% accurate. FS2004 could potentially mix and match accurate and rounded values and hence produce the offset. If my 10 sec calculations on the back of a beer cardboard coaster are correct, the rounding error is way too small to cause the offset. Also, I'm not 100% convinced that the offset is the same in x and y. It's very possible that the factor is 1.5 for x and 2 for y.- What I'm suggesting is that all LWM coordinates have to be translated by about 1.7 pixels to be accurate. The translation in x and y may differ and some more experiments are needed to confirm the exact values. (I think it's best to do the translation in lat/long before converting to LWM coordinates. An 'error' in FS2004 will convert the LWM values to wrong coordinates and the translation will fit that. No idea if this is a bug, a desired feature by MS to fix something somewhere else, or if this is actually of significance and necessary).Cheers, Christian

[Guest christian]

Just a final follow-up:I ended up shifting everything about 0.00009 / -0.00005 degrees. This is only approximate and may need adjusting. Then there are still rounding errors and the 256/255 effect.Cheers, Christian

[wsieffert 0]

While all that is interesting, what I find amusing is where you are doing the calculations!!"If my 10 sec calculations on the back of a beer cardboard coaster are correct..."and I believe the last time it was:(sic) my calculations on the back of a cocktail napkin...!All I can say is cheers and can I buy the next round? :-beerchug W. Sieffert :-lol