Thanks for the screenshots Steve.
Yes it's a replacement .x model that's falling foul of an enhancement in the 1.12 betas that you & Bounder both have. (this doesn't affect 1.11)
Some child models are meant to be scaled to different sizes, and previously this wasn't implemented in the FFED3D code - so the crater would just have drawn at a fixed size.
Another example of such a model is the rotating radar dish that's used on ships when a scanner is fitted, as well as on some Coriolis stations and at some military bases.
This one used to look OK when drawn at a fixed size on ships, but when it was used at stations/bases it was tiny - so I'd forced the drawing of the polygon model instead in previous builds. This 'fudge' has now been removed as the radar dish model will scale correctly AFAIK - but feel free to look out for it at military bases on bombing/photography missions and let me know if you spot any that don't!
You can easily fix the crater problem yourselves -
Open up the folder for the crater model (<FFED3D path>\Models\162) and create a new text file in there.
Edit the text file with notepad, paste the following text into it & save it:
Code:
[MODEL]
scale=0.065
scale_y=1.0
scale_z=1.0
offset_x=0.0
offset_y=0.0
offset_z=0.0
finally rename the text file to tris.ini
The value for the scale was arrived at by a quick bit of experimentation, hovering my ship near a few of those craters with the game paused.
The tris.ini file can be updated and saved - and the effect quickly seen by toggling between replacement/polygon models mode (CTRL+F6 key) in game - quitting and reloading isn't required.
This seems to give a very close match to the original large and small craters at that location. If you could check the resulting crater after a bombing mission too that would be great. (as would a save game for a test case if you have one)
I'll include a tris.ini with the core files in future to fix this.
Regarding the model preload failures - I've taken a quick look at memory usage between 1.11 and 1.12 betas and I don't think that they are significantly different from the application point of view. However with the new models that we've had since 1.11, these do take memory usage up above 2GB even with caching textures onto the video card.
I don't have all of Gernot's latest installed and I've chopped a few textures elsewhere - but without loading textures onto the video card, total memory usage was at/over 3GB.
Now, the app is 'large address aware' and ought to be able to use "up to" 4GB on 64bit Windows, but I think that we're definitely into unknown territory now and there's a good chance of hitting issues again pushing it this far.