GCV - Cheeky Curry Edition

ACG Fall of France and Battle of Britain Campaigns

Moderators: Thaine, Falcon, Administration Staff, WebDevs

User avatar
Donkey
Pilot Officer
Pilot Officer
Contributor
Posts: 1881
Joined: Thu Jul 03, 2014 9:21 am
Location: Bomber *Command* County :D - Buckinghamshire, England

Re: GCV - Cheeky Curry Edition

Post by Donkey » Wed Aug 02, 2017 8:23 am

So thinking about this further, the issue will be with things like inverted flight and barrel rolls. In order to account for such situations I will need to derive an "up" vector for each step on the track, where up is from the aircraft's perspective as opposed to "global" up (i.e. away from ground.) Now I can safely make the assumption that aircraft start "up", and I can probably then write some kind of algorithm to determine the most likely "up" on the next step based on other bits of info ... But that would fall more in the General Aviation / Sunday flying category, and I suspect it will fall apart pretty rapidly when we start to dogfight. And then I just know there'll be moans that I'm not properly capturing someone flying inverted or pulling snap rolls etc... So unless Bonkin has a magical way of deriving further information from the track (which, as we know, is sampled at a crappy once every 2.6x seconds) I can't see how I can get this to be anything other than an approximation that falls apart at the first evasive maneuver. I don't have precise figures to hand but would imagine the fighters can do a full roll in 4/5 seconds - so less than two samples - and the change in heading over two timesteps associated with a full roll won't be enough to generate a turn radius (and hence bank angle) anywhere close to reality. So I'm going to freeze this idea pending my TS chat with Bonkin.

User avatar
Frederf
Oberleutnant
Oberleutnant
ACG Staff
Posts: 133
Joined: Tue Feb 16, 2016 11:00 pm

Re: GCV - Cheeky Curry Edition

Post by Frederf » Wed Aug 02, 2017 11:32 pm

It's mathematically impossible to derive bank angle from position-time data perfectly. The airplane simply has more than one bank angle (or in general orientation) that may pass through the given positions in the given times.

I think the first thing you do is turn a 3D problem into a 2D problem by constraining the AOA to be zero. The lift vector is then always normal to the flight path. The flight path at N is probably best determined to be the circular path that passes N-1, N, N+1 points. That's easy enough. Ignore gravity for the moment and simply consider that at every N, the lift vector is pointing at the center of this turn circle. With a bank value defined at every N, interpolate smoothly for points between each N.

In the pure vertical plane this is fine, but assuming no gravity that would show all turns with horizontal component would be overbanked (e.g. level turns would bank 90). So how do you know how much to bias the lift vector anti-gravity? You need the turn radius, tangential speed, and turn plane orientation. With radius and speed you get the radial acceleration. Given centripetal (opposite sign radial) acceleration vector (inclined as turn plane is inclined) and gravity then lift vector can be reconstructed. Lift vector orientation gives bank angle. E.g. 2g radial, level turn bank must be 63 degrees. If this turn plane is inclined 45 degrees then vertical lift component is -(sqrt(2)-1) {vertical centripetal and gravity}, horizontal is sqrt(2) {horizontal centripetal} and bank is therefore 106° degrees. Turn downward in the vertical plane is no horizontal centripetal or gravity and vertical centripetal-gravity>0, 180° bank. Turn upward is the same but lift must be upward so 0 bank.

Summary, top vector to center of instantaneous turn radius center biased for gravity.

===

I thought a programmatic interpretation of the raw file name, not changing the file name itself. Understand the sun slider, just thought it would be natural if the slider when auto-controlled by file would update slider position so slider position was always truth in manual or auto-controlled mode.

The block was an example of how the data block is and how it might be (unified, color themed to ball, abbreviated labeling).

Didn't know about the font color. I wouldn't change it as a user. Yellow isn't generally readable against many backdrops. Even an auto-color-picker might struggle as different parts of the screen can be totally different colors. Is there a way to have a non-changing universal text like black with white outline or white with black outline?

User avatar
Donkey
Pilot Officer
Pilot Officer
Contributor
Posts: 1881
Joined: Thu Jul 03, 2014 9:21 am
Location: Bomber *Command* County :D - Buckinghamshire, England

Re: GCV - Cheeky Curry Edition

Post by Donkey » Thu Aug 03, 2017 7:54 am

So yes, I am fully aware that it is impossible to derive bank angle from position perfectly - I was going to approximate it by determining turn radius from step to step and then applying the old r=((v^2)/g.Tan(theta)) where theta is bank angle, and then go from there. But all of this falls down because of the slow sample speed that we have. We will never be able to pick up what really happened, simply display a possible solution for a path that hits all the points. And as I said, I just *know* that a lot of folk won't read this and will then say "hey, I was inverted for the whole flight, but you rendered me right way up - GCV sucks." or something like that and, frankly, I don't feel like going down the path of having to defend it on a regular basis. If our sample speed were higher I would reconsider, but as it stands we just don't have the data.

Oh, so "flight of 22nd July 17 @ 20:01" as opposed to 201707222001? Yeah, can do.

Ok, sun slider - so I think I get what you want. Difficult to achieve really, the sun is a single point light at distance, and the ambient lights are four floodlights (I think, would have to check - they may be points), one at each corner and much closer. So if I understand you correctly you want the toggle to pretty much work out a "best guess" light intensity that corresponds to the light given off by the sun, and apply that on toggle time? I wouldn't even know how to proceed to do that other than to have some enormous table of time v intensity that I pre-calculate (by eye). So ... No! The only other way (off the top of my head) would be to have multiple secondary cameras looking at different fixed points on the map (maybe the water?), reporting back an average colour, and then using that. But again, I would need some intensity-to-RGB function. Which again I would probably end up doing by eye as I don't really know much about colour models.

Will look again at the data block

Font colour, yes I agree the yellow can be hard on the eyes. As can black, as can white. In my own entirely subjective tests (aided by 9 year old daughter) we felt yellow was best. But yes, it's kinda nasty at times - hence why I added a click cycle. And yes, an outlined text would be great, but I haven't found a way of doing it yet (not, I must stress, that I've spent a long time looking). It will probably require a different shader, maybe I can use the textmesh. Don't know. But I'll have a bit of a Google this lunchtime as it is a better solution. Well, apart from rendering a background like I do for all the buttons. I don't like backgrounds though as they obscure the actual view. Mind you, I suppose I could add a hot key to toggle all GUI elements? That may be a nice to have. But I would not persist that state, obvs, as you need the GUI to choose the data file...! ;)

User avatar
Donkey
Pilot Officer
Pilot Officer
Contributor
Posts: 1881
Joined: Thu Jul 03, 2014 9:21 am
Location: Bomber *Command* County :D - Buckinghamshire, England

Re: GCV - Cheeky Curry Edition

Post by Donkey » Thu Aug 03, 2017 11:19 am

Right, latest version (2.14) up
Changes:
*Second hand in clock
*Nicer file names in file chooser
*Some buttons renamed
*Further speeds in speed slider
*Version number in help
*Outlined text for pairs block and Mission info / following (and as a result I have removed the cycling colours yellow/black/white)
*Hotkey to toggle GUI on/off ("G")

Plus a few ancillary under-the-hood changes

That's it

Let me know of any bugs, don't expect further changes.

D

Ginger
Contributor
Posts: 1798
Joined: Tue Sep 01, 2015 9:43 pm
Location: Nottingham, England

Re: GCV - Cheeky Curry Edition

Post by Ginger » Thu Aug 03, 2017 12:08 pm

All's good my end Donkey. Thanks again for all the effort. :nice:
"I saved you," cried that woman "And you've bit me even, why? You know your bite is poisonous and now I'm going to die"
"Oh shut up, silly woman," said the reptile with a grin "You knew damn well I was a snake before you took me in.

User avatar
Donkey
Pilot Officer
Pilot Officer
Contributor
Posts: 1881
Joined: Thu Jul 03, 2014 9:21 am
Location: Bomber *Command* County :D - Buckinghamshire, England

Re: GCV - Cheeky Curry Edition

Post by Donkey » Thu Aug 03, 2017 12:50 pm

Balls, there appears to be a bug in that the tail smoothing option seems to be reversed - i.e. it is off when toggled on and vice versa. I'm not sure if this is tied into persistency when there is no prefs config file on initial load, will need to investigate. I guess there will be at least one more version :|

User avatar
Donkey
Pilot Officer
Pilot Officer
Contributor
Posts: 1881
Joined: Thu Jul 03, 2014 9:21 am
Location: Bomber *Command* County :D - Buckinghamshire, England

Re: GCV - Cheeky Curry Edition

Post by Donkey » Thu Aug 03, 2017 2:12 pm

2.15 is up. It fixes the out-of-sync toggle issue on first load with no pref config file set. Also improves early morning sun lighting. Please God let that be it...

User avatar
Dickie
Air Commodore
Air Commodore
ACG Staff
Contributor
Posts: 12953
Joined: Sat Jul 14, 2012 12:15 pm
Location: Gloucestershire, England

Re: GCV - Cheeky Curry Edition

Post by Dickie » Thu Aug 03, 2017 2:13 pm

Far more than I ever thought we'd have. Amazing tool Donkey!
Image

User avatar
Broadsword
Flying Officer
Flying Officer
Contributor
Posts: 2751
Joined: Wed Sep 03, 2014 7:54 pm
Location: Cheshire, UK

Re: GCV - Cheeky Curry Edition

Post by Broadsword » Thu Aug 03, 2017 2:29 pm

Still no flake?
Image
I must go up to the skies again, to the peace of silent flight, To the gull’s way, and the hawk’s way, and the free wings’ delight;
And all I ask is a friendly joke with a laughing fellow rover, And a large beer, and a deep sleep, when the long flight’s over.

User avatar
Donkey
Pilot Officer
Pilot Officer
Contributor
Posts: 1881
Joined: Thu Jul 03, 2014 9:21 am
Location: Bomber *Command* County :D - Buckinghamshire, England

Re: GCV - Cheeky Curry Edition

Post by Donkey » Thu Aug 03, 2017 3:27 pm

Broadsword wrote:Still no flake?
Dull sound of explosion emanating from central London...

As it happens Broady, yes, I'll make sure to give it to you next time I see you. Don't be alarmed that it looks a bit like a cudgel...

Post Reply