Hello Michael!
Robsony made a very good question. Since loading your latest alpha version FMS_A.7.5.1 - which contained only a few changes - obviously was created in february 2002. So I am impatiently waiting for further releases which perhaps finally would fulfill my special (?) expectations.
I am very enjoyed that you decided to continue developping under Direct3D (instaed of OpenGL) because OpenGL does not run with my graphic card. Since Microsoft did its great flightsimulator also with Direct3D, it MUST be possible to develop another good simulator without OpenGL (even if some other programmer may say, Direct3D is a bad platform). Thanks for this decision.
I do not have any problems with the graphic performance. I do not conditionally need a smoke trace or similar trifels, which probably may cost a lot of calculation power, in relation to its use. I am very pleased with all details of landscape, sky and models, because these things are not so important when training or optimizing self-constructed models as I do.
In contrary, I find it much more important to improve the aerodynamic behavior of the models under FMS to fly finally as realistic as in reality. Until now, under FMS this expectation is not fullfilled. In the following I listed some of those weak points which I think they have to be improved urgently.
1. Running the model on the ground:
1.1
Problem with starting low powered models:
Until now there is a great discontinuity of behavior between running the model on the ground and free flying, especially with light and low-powered models: when starting, the model "sticks" to the ground although it may have enough speed. In this situation slowly pulling the elevator will only lower the tail but instantly reduce speed (perhaps below minimum speed to fly), but will not lift off (even if the model would fly at the same speed with lower power, once it is in the air). Only fast, strong (and short) pulling of the elevator can "tear up" the model from ground - but this is not a realistic way to start from a smooth runway (where ground effect even will help to lift off)!
1.2
Steering with the rudder on the ground:
Steering the model on the ground is still hardly possible. Especially with a slowly rolling model (where aerodynamic forces are nearly zero and inertia forces are still low) it should be possible to move the model just like a car. It should move just as if a wheel is mounted under the rudder (or under the bow) and steered with the same angle as the rudder. At higher speeds this slowly should change into the normal aerodynamic behavior, calculated from aerodynamic forces, inertia und the rest of lateral friction between wheels and ground. It should be regarded that in most cases the angle of the rudder-mounted wheel is much higher than the effective rudder-angle calculated from the .par-file, because in most cases rudder is combined with a fixed damping area.
By the way, in the air a model with an actuated rudder will not only slide (put the nose a little bit to one side - but continue flying into the same direction as before) as it is implemented under FMS until now. The model REALLY will slowly bank (change the course), even without any coupling across the V-form-Faktor (in the .par-file). This is a consequence of the angle between the fuselage's side-areas and the rudder which forms a vaulted profile and causes some "torque coefficient".
1.3
Different friction coefficients depending on the kind of ground surface:
On the smooth surface of a runway there should be only low frontal friction (but fully lateral, to enable steering on the ground as mentioned before). In contrast to this, on all green ground surfaces, frontal friction should be higher and lateral friction should be lower (because wheels normally are jumping over the grass). On water only models with floats should be able to land (what about to let slowly sink any other models?). Landing on ice or snow should be followed by crash with any model without skids.
1.4 Different speed limits on different ground surfaces:
Until now there is a speed limit when running the model on the ground. Instead of this I would like to have different speed limits depending on the kind of surface (especially on the runway it should be possible to run with unlimited speed to train touch & go; only vertical speed should lead to crash, if exceeded). Speed on other surfaces should be limited, perhaps to a constant factor above minimum speed, which can be easyly calculated from weight, wing area and maximum lift-coefficient.
2. Improvements in aerodynamic behavior:
2.1
Reaction of tail unit only statically caculated?
I have the impression that the reaction of rudder, elevator or aileron is only a function of the primary angle of attack (before starting rotation) and not calculated under dynamic conditions. When any rudder surface is turned against the actual air-stream, at first it causes a force, depending on speed, surface area and angle of attack. This force works against the inertia of the actual axis and then starts to accelerate the model around this axis. But as soon as this rotation starts, the effective angle of attack to the rudder is reduced because of the rotating speed and therefore a new resulting force must be calculated, and so on.
Now it seems to me, that within FMS this continued calculation (depending on the actual rotating speed) is not implemented. Otherwise I cannot explain some very strange behaviors when "playing" with coefficients in the .par-file. For instance, with a very strong engine it is easy to produce loops which are very much tighter than they physically could be, limited by the angle between wings and elevator and the distance between them (this construction describes a circle with the minimum diameter, which never could be tighter; any inertia force only could widen the resultant loops). Nevertheless, under FMS I had models with looping circles smaller than the models themselves - they could rotate around the wing's axis as if the wingtips are put in a bearing!
2.2
Flying on knife's edge (Messerflug):
Once, I read in this forum, flying on knife's edge could be performed by hanging the model onto the propeller and using a very strong engine (adjusted within the .par-file). I find, this is not a realistic simulation. In reality when banking at high speed there is a lot of lift by the fuselage's side surface, used to keep the model in the air without hanging on a propeller. Under FMS, it should be calculated in this way, too.
2.3
Rudder and elevator reaction with regard to the propeller's beam:
Although I am not a friend of extreme acrobatics, e.g. torque rolls, I think, they should be possible, too. Is it only my awkwardness or is it simply impossible to perform torque rolls under FMS? For me it seems that rudder and elevator are not functioning at all as soon as the model's speed becomes nearly zero (as it is normal when hovering under the propeller). Not only therefore the simulator should always calculate with the real speed of the propeller's beam (instead of a simple thrust). In reality great differences in reaction to rudder and elevator can exist depending on the engine's actual power. This generally should be implemented under FMS (and torque roll will run, too). The engine's torque (power, devided by the engine's speed) and the damping area of the wings can be used to calculate the rotating speed at torque roll and the small torque around the length axis at normal fligth.
2.4
Using real propeller's jet data instead of simple thrust:
In reality the propeller's thrust is not a constant force as it is written in the .par-file! Somewhat simplyfied (and normally with a propeller which has a gradient not too high), the function looks like a negative square function with maximum thrust at zero speed and zero thrust at its "jet speed" (which can be calculated from the propeller's gradient and its rotation speed). Similar to this under FMS the thrust for all propeller models at least should be calculated depending on the actual model-speed. This would help optimizing the propeller dimensions for a real existing model. Even with jet-enginges it would be useful to calculate thrust from the difference between jet-speed (which is not unlimited) and the model's speed. The calculation should be done from the actual engine's speed which results in lower jet-speeds when the engine's power is reduced.
3. Pilot's view?
Sometimes, I wish to sit within the model (oh no - I only wish to have the pilot's view

). This can have some advantages against simple following the model, especially within loops (where the graphic sometimes is confusing) and tight turns (where the graphic simply is too slow). Therefore this is not only an important "fun-option" to me.
4. Repeat function and free setting of any start-situation?
Sometimes it is useful to repeat the last seconds (up to 60) of the simulation and - perhaps - during this stop repeating and continuing with normal flight. Additionally I would like to have free and easy programmable start situations (Model speed, direction, height and pilot's position) as this is implemented in Microsoft's flight simulator. Is this possible?
Dear Michael, I would enjoy if you work on some of these points, too, and give us a positive answer.
And thanks to you for all the good work with FMS you have done for us until now!
Best regards,
Norbert Graubner