Board index FlightGear Development Aircraft

F-20 development

Questions and discussion about creating aircraft. Flight dynamics, 3d models, cockpits, systems, animation, textures.

Re: F-20 development

Postby swampthing » Fri Jul 17, 2020 5:57 pm

The window that opens up when you load the plane has the top cut off. O moved it to the side and it went away once I started the plane. Mu opinion the throttle detent sounds are extremely loud. You may want to lower volume for them. I got into a deep stall nearly hitting the ground could recover instantly with some throttle, that probably should not happen, I think I should have hit the ground. Its nose heavy even above 600 knots even trimming the elevators out quite a bit. I finally let it crash and thats when i see the ground reactions need a lot of work. It went crazy and locked up Flightgear, Ig you need help with that send me a PM. Other than that looking good.

All the best Steve
www.opredflag.com
I have sworn upon the altar of God, eternal hostility against every form of tyranny over the mind of man. -Thomas Jefferson-
swampthing
 
Posts: 591
Joined: Wed Oct 28, 2015 4:10 am
Location: Missouri
Callsign: swamp
Version: 2018.2
OS: multiple

Re: F-20 development

Postby Flying toaster » Fri Jul 17, 2020 8:28 pm

swampthing wrote in Fri Jul 17, 2020 5:57 pm:The window that opens up when you load the plane has the top cut off. O moved it to the side and it went away once I started the plane.

What is the resolution of your window ? I will see if I can move the window default position to improve
Mu opinion the throttle detent sounds are extremely loud. You may want to lower volume for them.

Agreed, I will put in the todo list
I got into a deep stall nearly hitting the ground could recover instantly with some throttle, that probably should not happen, I think I should have hit the ground. Its nose heavy even above 600 knots even trimming the elevators out quite a bit.

Now that is very weird. Once the gear is up, the aircraft should need no trimming at all. The fcs has a g-demand law so if you release the stick in level flight, it should remain around 1-g without any trim.

Also the behavior near stall is as expected. Actually if you yank the stick all the way back but without bleeding too much speed, it will remain stuck at around 30 degrees alpha and just stay there indefinitely. Straight wings have a lesser tendency to pitch up and deep stall than swept wing. This is why the f/a-18 (which configuration is similar to the F-5 family for obvious reasons ;) ) can do the famous "hornet walk" sticking its nose at very high alpha without losing control.
What you did looks actually like a hornet walk. You stuck the nose at alpha max and used power to keep altitude despite the massive drag from lerx lift. This is possible because a clean F-20 is kind of overpowered. It may even be possible to regain altitude in a deep stall at low altitude, but only at full afterburner and very reluctantly (basically, it is a rocket)
Your experience may differ with a full load.
If you want to deep stall the F-20, you will need to do a tail slide and even then, not all configurations will lead to the same results. The AOA will settle around 50 degrees, and the nose will bob up and down, until you get into denser atmosphere where you can try to power out of the stall. You may try to break the deep stall earlier by rolling the aircraft, though it will be difficult.
If you play with loads (fuel tanks mostly) and rudder, you can even start a semi stable spin ( though most of the time recoverable with some work)
Finally, if you really want some more excitement, this new version comes with the controls to switch off the FCS (CAS controls on the left console of the cockpit). You will have the "raw" aircraft, with the ability to pull incredible g's and a fairly nauseating Dutch roll.
I finally let it crash and thats when i see the ground reactions need a lot of work. It went crazy and locked up Flightgear, Ig you need help with that send me a PM. Other than that looking good.

All the best Steve

Yep I did not define body contact points in the FDM... Should go in the to-do list also

Thank you very much for the feedback

Enrique
Flying toaster
 
Posts: 383
Joined: Wed Nov 29, 2006 6:25 am
Location: Toulouse France

Re: F-20 development

Postby swampthing » Fri Jul 17, 2020 11:25 pm

My resolution is 1600x900. For the stall I had power off and let it fall. yes the nose did bob up and down some, it would come back up after pitching down and gaining speed about 115 knots if i remember right. Gear was up so I;m not sure why the FCS did not kick in. I didn't do a low altitude, low speed high AOA pass. that something I like to do and yes the F-18 is great at that. I've seen them do it in person. Its a shame this plane was never put into production I'm a big fan of the F-5 and the F-20 that sadly never got its chance.
www.opredflag.com
I have sworn upon the altar of God, eternal hostility against every form of tyranny over the mind of man. -Thomas Jefferson-
swampthing
 
Posts: 591
Joined: Wed Oct 28, 2015 4:10 am
Location: Missouri
Callsign: swamp
Version: 2018.2
OS: multiple

Re: F-20 development

Postby Flying toaster » Sat Jul 18, 2020 6:17 am

On this simulation, if you have to keep the stick back to remain in stall, you are not in deep stall. If you push the nose down and it does not come back then you are in deep stall. The difference can also be spotted on AOA (around 30 vs 50).
To do a hornet walk, I usually put the engine at idle, bleed the speed off until the aircraft can't maintain altitude, then raise the nose until the stick is full aft back (in general you will be around 90kts and 27/30 degrees AOA). Then increase gradually thrust until the velocity vector on the HUD is just above the horizon, with the stick still full aft. With light loads you can keep there until you run out of fuel. To leave, release the stick, and to leave real fast release gradually as you go to full AB
Flying toaster
 
Posts: 383
Joined: Wed Nov 29, 2006 6:25 am
Location: Toulouse France

Re: F-20 development

Postby Flying toaster » Tue Dec 29, 2020 6:29 pm

Hello all,

I am trying to update the F-20 to play nicely with the latest version of FGFS and in particular the compositor. I am running v2020.3.4 and I have run into a small problem
With the compositor enabled version i the shader paths are broken (not the same name as the old ALS shader names, e.g. model-ALS-ultra.vert vs model-ultra.vert in the ALS folder).
This results in the display being completely broken.

Anybody else found that issue with their creation ?

Cheers

Enrique
Flying toaster
 
Posts: 383
Joined: Wed Nov 29, 2006 6:25 am
Location: Toulouse France

Re: F-20 development

Postby legoboyvdlp » Tue Dec 29, 2020 6:34 pm

FlightGear 2020.3 compositor is more or less obsolete now. Compatibility should be maintained with 2020.4 without any changes being necessary. On 2020.4 Compositor is the only option and is on by default without any changes required. You can try it out on the nightly but unless you use custom shaders, no changes beyond adding <noshadow> should be required.
User avatar
legoboyvdlp
 
Posts: 7859
Joined: Sat Jul 26, 2014 1:28 am
Location: Northern Ireland
Callsign: G-LEGO
Version: next
OS: Windows 10 HP

Re: F-20 development

Postby Flying toaster » Wed Dec 30, 2020 8:28 am

legoboyvdlp wrote in Tue Dec 29, 2020 6:34 pm:FlightGear 2020.3 compositor is more or less obsolete now. Compatibility should be maintained with 2020.4 without any changes being necessary. On 2020.4 Compositor is the only option and is on by default without any changes required. You can try it out on the nightly but unless you use custom shaders, no changes beyond adding <noshadow> should be required.

Thank you !

I am using custom shaders though :wink:
Flying toaster
 
Posts: 383
Joined: Wed Nov 29, 2006 6:25 am
Location: Toulouse France

Re: F-20 development

Postby Icecode GL » Wed Dec 30, 2020 12:51 pm

If your custom shaders do not support Rembrandt, then you should be fine. If they do, please let me know and I'll point you in the right direction.
Icecode GL
 
Posts: 640
Joined: Thu Aug 12, 2010 12:17 pm
Location: Spain
Callsign: icecode
Version: GIT
OS: Fedora

Re: F-20 development

Postby Flying toaster » Wed Dec 30, 2020 5:24 pm

Icecode GL wrote in Wed Dec 30, 2020 12:51 pm:If your custom shaders do not support Rembrandt, then you should be fine. If they do, please let me know and I'll point you in the right direction.

Well I made custom shaders that work both with ALS and Rembrandt in 2019.1.1
I downloaded the nightly build and the shaders appear broken (i.e. they don't work in default pipeline, but I still get the texture map and they appear all black in the ALS pipeline).
I deactivated the rembrandt effect for good measure. I get less messages on the console complaining about the lack of gbuffer shaders, but still no luck :/
Flying toaster
 
Posts: 383
Joined: Wed Nov 29, 2006 6:25 am
Location: Toulouse France

Re: F-20 development

Postby Icecode GL » Wed Dec 30, 2020 7:57 pm

I trIed the F-20 from a Google Drive link on the previous page but FG crashed because what seems to be a JSBSim-related error (FGPropertyManager::GetNode() No node found for fcs/pitch-actuator-rate).

Anyway, after a quick look it seems like removing the Rembrandt Effects (the lacquer-deferred.eff file) should be enough. I couldn't test though because of the error above. Do you get any other console output or did you try something different to what I said?
Icecode GL
 
Posts: 640
Joined: Thu Aug 12, 2010 12:17 pm
Location: Spain
Callsign: icecode
Version: GIT
OS: Fedora

Re: F-20 development

Postby Flying toaster » Thu Dec 31, 2020 9:30 am

Yep, later versions of JSBSIM have become more picky with uninitialized variables (as they should).
here is a link to an updated file in the main directory to fix the problem
https://drive.google.com/file/d/1GKEZ_s ... sp=sharing
And yes I reference directly the lacquer.eff instead of lacquer-deferred.eff, and this is where I get the result I mention above...
Flying toaster
 
Posts: 383
Joined: Wed Nov 29, 2006 6:25 am
Location: Toulouse France

Re: F-20 development

Postby Icecode GL » Thu Dec 31, 2020 11:07 am

Thanks for the fix, it works now.

The problem is that the lacquer.eff uses texture units that are now being used by the Compositor (texture unit 10 is used for the shadow map, for example). In general I don't recommend adding aircraft-specific shaders because things like this tend to happen. I don't think the fix is trivial, the only solution I can think of is rewriting the shader to not use texture units bigger than 7. Also keep in mind that most hardware is limited to 16 texture units.
Icecode GL
 
Posts: 640
Joined: Thu Aug 12, 2010 12:17 pm
Location: Spain
Callsign: icecode
Version: GIT
OS: Fedora

Re: F-20 development

Postby Flying toaster » Thu Dec 31, 2020 1:45 pm

Thank you for your input.

It seems like a promising venue for investigation, but unfortunately, changing the numbers of texture units yielded not result. Indeed the textures are black when ALS is activated. I guess it has more to do with the uniforms used in the new version of ALS.
As a side note, thank you for pointing that the max number of texture unit may be limited. It so happens that ALS uses a LOT of them (up to 8 or more if I remember correctly). So if I don't want custom shaders to interfere I really need to go much further. The good news is that even on my laptop's now completely obsolete core i5 5th generation integrated GPU I have 192 texture units (same for the 940M discrete GPU that I actually use). And with those specs it is not unusual for FG to turn into a slideshow (it is playable most of the time, but it can get choppy). So I guess I will go past the 16 texture units for ALS enabled effects, non ALS will work just fine (albeit without most of the eye candy).

I will keep investigating but thank you for putting me on an interesting track :)

Cheers

Enrique
Flying toaster
 
Posts: 383
Joined: Wed Nov 29, 2006 6:25 am
Location: Toulouse France

Re: F-20 development

Postby Icecode GL » Thu Dec 31, 2020 4:01 pm

It so happens that ALS uses a LOT of them (up to 8 or more if I remember correctly)


Yup, that's proving to be an issue for the Compositor as well. Mac drivers artificially force 8 texture units, so they give errors when we use higher texture units.

So if I don't want custom shaders to interfere I really need to go much further.


Indeed, most devices have a lot of texture units, so you should be fine using texture units higher than 16. I can (kinda) guarantee that core features won't use units higher than 16.
Icecode GL
 
Posts: 640
Joined: Thu Aug 12, 2010 12:17 pm
Location: Spain
Callsign: icecode
Version: GIT
OS: Fedora

Re: F-20 development

Postby Hooray » Thu Dec 31, 2020 4:28 pm

There's a total limit of texture units to be used, but also shader-specific limits, i.e. per shader type: https://www.khronos.org/opengl/wiki/Sha ... imitations

I don't think these limits are currently being honored/checked by FlightGear (?) - if in doubt, these should probably be exposed in the form of properties and also added to about.xml - so that people can more easily report back with proper bug reports.

And obviously, flightgear still uses the compat profile, too.

One option might be introducing "virtual texture units" that only make sense in the context of FlightGear - which are internally mapped to actual texture units, i.e. rather than hard-coding actual TU numbers, use a virtual index that is internally replaced, based on querying the underlying system.
Please don't send support requests by PM, instead post your questions on the forum so that all users can contribute and benefit
Thanks & all the best,
Hooray
Help write next month's newsletter !
pui2canvas | MapStructure | Canvas Development | Programming resources
Hooray
 
Posts: 12173
Joined: Tue Mar 25, 2008 8:40 am

PreviousNext

Return to Aircraft

Who is online

Users browsing this forum: No registered users and 2 guests