Dev Diary 131

Changes we're testing now (they are likely to be included in 2.004 update)

80. REVI-12 display reticle corrected (its angular size increased by 12%);
81. Lift force loss and drag increase of a remaining wing after fragmentation of a wing corrected;
82. First optimization task completed: AI controlled aircraft on the ground can be doubled;
83. 'Jumpy landing gear' issue fixed thanks to improved shock strut model;
84. Bumpiness of concrete runways and taxiways reduced;
85. He-111 won't rotate after an emergency landing;
86. Powered temperature gauges won't work when engine is off;
87. Engine heat model corrected for Ju-87;
88. Oxygen deprivation model corrected (its effects are felt at higher altitudes than before);
89. German breathing devices are engaged at higher altitudes (by German flight manuals);
90. Pe-2 fuel gauge is backlit now;
91. La-5 cylinder head temperature gauge is backlit as well;
92. Pe-2 trimmer neutral position indicators work now;
93. Bf 109 E-7 mass correctly decreases with ammo spending;
94. New technochat message added telling pilot that an engine is deselected.

In the meantime, the work continues. One of our main goals at this stage of development is a significant increase in an amount of AI controlled aircraft acting simultaneously without a noticeable decrease in modeling quality, keeping realistic aircraft characteristics. The idea is to keep the physics modeling accuracy of an AI controlled plane the same as player controlled plane has and optimize only those modules which impact is only noticeable when you pilot a plane yourself. A significant chunk of time of our lead flight physics specialist and the author of our flight model, Andrey Solomykin, is dedicated to this task. We're optimistic since the early results are good.

It is known that one of the physics bottlenecks in terms of the required CPU power was an aircraft behavior on the ground so this was a natural first stop of Andrey's journey to improved performance. And he passed it with flying colors - at this moment we're able to double the amount of AI planes on the ground keeping the same level of performance - simply staying on the ground, taxiing, performing a takeoff or landing. This change will be included in 2.004 update. In addition, as it always happens with a large scale research, this task yielded unforeseen results. By completing it we were able to solve the 'jumpy landing gear' issue.

Earlier we stated that implementing additional landing gear dampening is not possible within the confines of the current landing gear model which was already at its computational stability limit. Increasing the sampling rate even more to raise this limit made the performance impact in the 'aircraft on the ground' case unacceptable. But now, with this optimization completed, the landing gear modeling was improved and it gives more accurate results which are comparable to real landing gear testbed data even at a reduced sampling rate compared to the old one. So the 'jumpy planes' issue will be addressed in 2.004 as well. Aircraft behavior on the ground, which was already the most accurate among ALL existing flight simulators, will become even more true to life. It should be noted that thorough research of documentation revealed that Soviet shock struts were designed for much greater load. This is why Soviet aircraft were much more 'jumpy' than German ones in reality. We even found an interesting phrase in Soviet documentation from 1948 mentioning that some pre-war and war landing gear designs, in which shock struts of a stationary aircraft should have been completely extended, were wrong. You'll be able to feel the difference between Soviet and German landing gear designs in the 2.004 update. Some of the effects are visible in this short video: