Dear IPACS team,
Back to business, downloaded the latest Geoconvert tool version and restarted compilation of a small scenery zone neighbouring the Swiss DLC around Geneva, levels 9 to 14.
Below is a general picture of the treated zone. The greenish Swiss DLC is easily recognizable vs. my orthophotos.
I haven't worked yet on color matching, which isn't optimal, that's definitely not the topic of the test - let's look first at basic scenery feasibility.
As well, one can see the poor lake rendering in the Swiss DLC, which will need to be addressed in a later phase.
Fig1. the global zone of my scenery
The good news is the previously reported stripes issue seems to have been fixed by now.
However I could identify several other issues at compiled scenery interface both with the the Swiss DLC and the default scenery.
INTERFACE BETWEEN COMPILED SCENERY AND THE SWISS DLC:
In some cases like here (aircraft location corresponding to Fig1), the interface between the Swiss DLC and my Geoconverted tool is perfect. Once again, I'm not talking about color matching, but pure phototexture matching. The sub-optimal color matching enhances readability: one easily distinguishes my scenery (right) from the Swiss DLC (left).
Fig.2
Fig3. Looking left, my scenery (left) and the Swiss DLC (right).
In some other places however I see some fancy blurry effects of the Swiss DLC at interface.
Here's a first example at the Rhone river, where the scenery change is easily recognizable.
Fig4.
Fig5. The underlaying Swiss DLC gets locally blurred along the blending line (here, looking left).
Fig6. Looking right
Fig7. Corresponding blending mask after running Geoconvert (as can be seen is images_raw).
Another example at another place, with my scenery
Fig8.
Fig9. The same location without my scenery
Fig10. With my scenery:
Fig11. Without my scenery (Swiss DLC only)
Fig12. Looking left (with my scenery enabled)
INTERFACE BETWEEN COMPILED SCENERY AND THE DEFAULT LOW RES GROUND:
Fig13. On the eastern side, my test orthophoto stops just east of Thonon-les-Bains, on the French side of the lake, exactly at E6.5625°
Fig.14. On purpose for this test I set my TMC file to stop at E6.50° instead, so that the photo coverage is larger than the scenery.
Note that I also tried to round up the NW corner to 2 digits, but it doesn't change anything to the issues.
Fig15. After compilation the zone looks like this when not zooming in too far:
Fig16. If I zoom in just one click, it changes to this.
Let's have a closer look, and draw an overlay.
Fig17. Same zoom level as per Fig15. Note that the ground photo extends to its eastern limit instead of the limit set in the TMC file.
Fig 18. Same zoom level as per Fig16. Note the scenery zone between the eastern end of the TMC file (E6.50°) and the eastern end of my orthophoto (E6.5625°) gets covered with my orthophoto, but in lower resolution (see also Fig.19).
Upper than N46.39° this zone appears untouched (???)
Fig 19. This area seen from the aircraft, looking left
Fig20. Looking right
Fig 21. The same place without my scenery. Note that the N46.39° border can be seen (pixellized zone).
I keep files to your disposal should this be of any help to troubleshoot and fix these bugs. So far (I compiled quite a few times) always the same areas seem to be affected, thus it's pretty much reproducible.
Cheers
Antoine