1 hour ago, tweekz said:

It looks very amateurish

What are you smoking and can I have some?!

1 hour ago, LHookins said:

I've never seen a "lookup table" in anything other than pre-calculated trig functions back before math coprocessors.

I just remembered that lookup tables (LUT) are used in color work on a computer.  OK, that's TWO things that use lookup tables, with only one still in use. 😄 

Hook

11 minutes ago, Bottle said:

What are you smoking and can I have some?!

I think Tweekz was just referring to (when he said "looks amateurish" ) when companies like MS fail to perform an easy fix before release date. Yea it may look unprofessional but there is good reason why it happens in a project of such scope as MSFS.

I don't think he was referring to the entire project "looking amateurish", in other words 🙂 (if that's what you were thinking)

But if I'm misunderstanding anything there please excuse the foot in my mouth.

Edited August 14, 20205 yr by hangar

6 minutes ago, LHookins said:

Polynomial function:  A x squared plus B x plus C.

A, B and C come from tables.  Please, Please, PLEASE don't tell me those are hard coded in any sim! 😄 

I think MSFS still has some fine tuning to do.  If I'm not happy with what we get out of the box, I'll do my own fine tuning.

Hook

That's actually a quadratic (second order) function, which is but one type of polynomial function.  😜

I'd bet that the professional Level-D type sims actually do have many of those functions hard-coded for the acft they are intended to simulate. 

I wrote a custom device driver for my modified PFC flight controls.  On several axes I replaced the pots with a Hall Effect sensor and magnet(s) in various configurations that produce non-linear output values that need converted back to a linear set of values that closely represent the position of the axis control.  I measured the position and outputs across the control's range, and I first tried coding those measurements into a table and interpolating the results.  As I suspected going in, I discovered that linear interpolation of a non-linear dataset was underwhelmingly inaccurate.  I then did curve fits and hard-coded the resulting quartic (4th order) polynomials, and that produced a *much* smoother, linear transform.

As to fine tuning...let's hope that the capability to do that is built into the sim. 

13 minutes ago, LHookins said:

I just remembered that lookup tables (LUT) are used in color work on a computer.  OK, that's TWO things that use lookup tables, with only one still in use. 😄 

Hook

no actually theres' more...lots more...I use em every time I need an excuse when my wife wants to know why taking out the trash after dinner always has to wait awhile 🙂

Edited August 14, 20205 yr by hangar

21 minutes ago, w6kd said:

"Wrong" implies a binary set of possibilities..."right" or "wrong". 

Better to say that all models are inaccurate to some extent.  The question is whether that inaccuracy, or deviation from reality, is significant enough to render it materially nonrepresentative of what the model is intended to emulate.

It’s a famous quote from the statistician George Box (which should have been attributed by the OP).

The Asobo model is immediately wrong if it doesn’t have all the variables that the real world will have, which of course it can’t have so by definition is “wrong”. But 1000 points is certainly a vast improvement in parameter space over previous sims and as long as it doesn’t fall into the trap of over paramertization, should do more than adequately considering it’s just a graphical representation of reality that it’s giving us. The term model makers use is “goodness of fit”. I’m confident that the Asobo FM is sophisticated enough to give high goodness of fit once the correct parameter values are arrived upon, which arguably (and I use that term advisedly) it needs a little more finessing in.

I find overall these forums suffer significantly from “glass half empty” personalities.

33 minutes ago, LHookins said:

Please, Please, PLEASE don't tell me those are hard coded in any sim!

Pure speculation here:

What it could be is a function associated with each of the 1000 points. Each is fed with the input state of the cell it is in at each time point and then spits out the appropriate value that gets fed to the appropriate bit of the code. The “flight model” could then be “hard baked” into each function via such things as scaling factors or even redirection to secondary transforming functions. To make it generalised, these FM elements could be variables themselves. This would be a much more complicated system than a more traditional look up table, but has the advantage of being generalised. It would be enormously difficult to do but if they have done it that way, I take my hat off to the team and hope they eventually publish it in an academic journal and/or patent it.

24 minutes ago, tweekz said:

Doesn't BET do both of it? Reading in geometry of surfaces and creating data for the sim how to interact with contact points in certain situations?

Yes...the difference I was trying to highlight between the apparent significance of the terms "surfaces" and "points" is that BET (surfaces) actually computes a large set of moments based on the physics of a fluid flowing across a small section of airfoil..."points" might well be computing those moments (and many fewer of them) based on an approximation derived from something else, like backing-in the data based on observed or expected behavior of that or a similar type of acft.

13 minutes ago, w6kd said:

As to fine tuning...let's hope that the capability to do that is built into the sim.

According to my tests, we can.

Hook

Has there ever been any clarification between the fact that the aerodynamics preview video text says 1000 points, but the developer says "thousands" verbally?

All of this discussion brings us down to the only question that matters:  How well does it simulate real flying?  

Or what may be even more important, how well does it simulate what YOU think real flying should be like?  

Hook

2 minutes ago, ShawnG said:

Has there ever been any clarification between the fact that the aerodynamics preview video text says 1000 points, but the developer says "thousands" verbally?

This demonstrates how much the whole thing is fractal in nature.  The deeper you go, the deeper and more complex things become.

Would your girlfriend be just as beautiful on the cellular level?  On the molecular level?  Why are we gazing into soulful eyes with an electron microscope?

Hook

3 minutes ago, LHookins said:

All of this discussion brings us down to the only question that matters:  How well does it simulate real flying?  

Or what may be even more important, how well does it simulate what YOU think real flying should be like?  

Hook

Or even more importantly: how well does your cfi think it simulate real flight

A flight sim doesn't even need a flight model to teach you the most useful things you will learn from it.

Hook

41 minutes ago, LHookins said:

I just remembered that lookup tables (LUT) are used in color work on a computer.  OK, that's TWO things that use lookup tables, with only one still in use. 😄 

Hook

My airplane uses look up tables to figure out what intial pitch attitude should be to fly V2+20 after takeoff.

There's another one for your list! 🤣