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 Thorsten » Fri Dec 30, 2016 7:53 am

BUT if a user is using Rembrandt he could discover some ugly side effect, under some conditions the developer can compensate it .


Again - the aircraft should not touch any settings which are configured by the rendering GUI dialogs unless the user specifically requests this. That includes quality settings, filter parameters, the number of passes done for Rembrandt shadows etc. If the user discovers an ugly side effect, the user has to compensate for it (for instance because the developer has no idea of how something looks on a differently calibrated screen and on different hardware).

The aircraft should only set the effect parameters which are not configured by the rendering dialog (such as, for instance, enable normal mapping on if there is a normal map provided - where the user has the option to not use the normal map by selecting low rendering quality).

The whole system is intentionally designed the way it is, I've tried to explain to you a couple of times why (settings being applied to all scenery elements, not only your plane, user confusion with auto-saved settings, multiple systems fighting for control over the rendering settings, configurations which can't run on weaker systems,...) , if you still don't understand why, then please do whatever you feel you need to do in your own hangar, but don't give bad advice in this forum. Thanks.
Thorsten
 
Posts: 12490
Joined: Mon Nov 02, 2009 9:33 am

Re: F-20 development

Postby Thorsten » Fri Dec 30, 2016 9:00 am

Somewhat off-topic (feel free to move):

The problem is that, the interior lights are very static, therefore can not have complex effects like with Rembrandt.


In terms of complexity, the ray-tracer used for the Shuttle cockpit lighting beats any real time technique by orders of magnitude (it has multiple secondary and tertiary reflections on materials with different properties, shadows, near exact ambient occlusion for the reflected light,...)

The main lesson here to understand is - rendering is about making choices between techniques, nothing is perfect in real time. Asking for the best of two worlds is denying that basic truth.

If you can compute something offline, you can commit half an hour of computation time to it. It's going to have all the intricacies of the real thing. If you want to do it runtime, you have 1/60 of a second for the job. it's going to be coarse.

If the scene you render is very dynamic (think of cars on a busy street), then you need to do all computations runtime and so it has to be coarse. If the scene is near static, then you can do a much better job computing everything offline.

Thus, to the degree that a cockpit interior is a static environment, a raytracer will give the best results for illumination by far. But it will not treat moving parts correctly, some special treatment is needed there. A real time technique will treat moving parts correctly, but in a very basic way - never ambience generated by multiple reflections or multiple shadows.

Likewise, pre-computing ambient occlusion is far more exact (and performance friendly) than doing it runtime - so if the nooks and crannies of your environment never change much, pre-computing AO is the much better way to go. Only if they change you want to have the computation runtime.

So in many situations the choice is not between forward and deferred rendering, it is between high-fidelity offline techniques and much coarser runtime techniques. You can never get a lighting computation in raytracer quality by doing it real time - you can have one technique with it's advantages and disadvantages or you can have the other, but not a merge of both.

(Personally, I believe approximating cockpits as static scenes is very good and that things you actually see from inside the cockpit like cloud shadows are far more important than things you only see from exterior views like aircraft shadows - but e.g. Enrique thinks differently with regard to the static approximation, so it's a matter of taste).
Thorsten
 
Posts: 12490
Joined: Mon Nov 02, 2009 9:33 am

Re: F-20 development

Postby Flying toaster » Fri Dec 30, 2016 9:48 am

abassign wrote in Thu Dec 29, 2016 11:05 pm:
For some time I would like to merge the two projects because each of them has its advantages, but also disadvantages.
When I saw the shadows, the effect of rain, the position lights, the sky etc ... ALS is certainly great, but Rembrandt, looked very well the emission lights, especially the interior lights. The solution of external shadows adopted by ALS is still very poor, in that context Rembrandt is considerably higher. For the interior lights, Thorsten has demonstrated that it is possible to achieve very significant effects. The problem is that, the interior lights are very static, therefore can not have complex effects like with Rembrandt.

Therefore I would focus on the affects ALS, which are still not simple, hoping that in the future someone will maintain Rembrandt and integrate it with ALS.


You summarise quite well my feelings. Merging ALS and Rembrandt would mean fixing a couple of things to merge both lighting models. Plus Rembrandt would allow a much more credible heat blur effect! Also antialiasing is possible under Rembrandt, you just have to be ready to pay for the frame rate expense... As you point out it is a problem of lack of maintainer.

The difference of performance when I changed the MOBO and processor would be more related to memory/bus bandwidth than CPU (my new CPU is kind of idle now). That would hint to memory transfers back and forth from main memory to GPU... If true, that's not a good thing. Since I have to tweak the C++ code anyway for dynamic lights, I will look at the rembrandt part to see whether all the buffers are declared in GPU memory, or any other easy to spot bottleneck.

As far as the F-20 is concerned, I will try my best to keep it framerate friendly on a reasonnable rig (hence my request for feedback), and yes I will tweak it to look better on any renderer (within the limits set by my other RL activities).

Cheers

Enrique
Flying toaster
 
Posts: 390
Joined: Wed Nov 29, 2006 7:25 am
Location: Toulouse France

Re: F-20 development

Postby Thorsten » Fri Dec 30, 2016 10:17 am

You summarise quite well my feelings. Merging ALS and Rembrandt would mean fixing a couple of things to merge both lighting models.


I give up. I'll just ignore any such remarks in the future rather than trying to explain anything.

Plus Rembrandt would allow a much more credible heat blur effect!


May I ask - have you even tried using the hires flame shader for heat blur before writing this?
Thorsten
 
Posts: 12490
Joined: Mon Nov 02, 2009 9:33 am

Re: F-20 development

Postby Flying toaster » Fri Dec 30, 2016 10:20 am

And to reply to Thorsten, I think the key there is what causes "suspension of disbelief" for each individual
The approach to have everything precomputed makes for beautiful screenshots (and I mean stunning lifelike sometimes).
Also the clever scheme for interior shadows in the direct rendering pipeline is a case of ingenuity, and does alleviate the issue of static images.
That being said, having a shadow at exactly the same location when I throw a switch or when I am making a 360° turn in a sun bathed cockpit is really a spoiler for that "suspension of disbelief" to me (to the point I'd rather have no shadows, rather than fixed in some instances). Hence my preference for simpler but geometrically right shadows.
But as you point out it is a matter of personal taste, and in that department everything is right.
I will try to accomodate as much as I can to cover for various tastes, and release the sources (i.e. blender and source SVG files) for people to get the best for each method according to their individual taste (the F-14 has gone on to have a successful life long after its release after all).
Hopefully that will make for a nice and peaceful FG community :)
Cheers

Enrique
Flying toaster
 
Posts: 390
Joined: Wed Nov 29, 2006 7:25 am
Location: Toulouse France

Re: F-20 development

Postby Thorsten » Fri Dec 30, 2016 10:32 am

That being said, having a shadow at exactly the same location when I throw a switch or when I am making a 360° turn in a sun bathed cockpit is really a spoiler for that "suspension of disbelief" to me. Hence my preference for simpler but geometrically right shadows.


Rembrandt delivers no shadows for secondary light sources, not simple but geometrically right shadows, just plain nothing - any cockpit light you implement passes through any surface unhindered. And no surface with an alpha channel ever casts a shadow on anything.

And the lightmap is agnostic to whether you render it with switch shadows or not.

It's in how you use these things whether you get to see what you want to see or not.

Please don't confuse what the technologies actually do with a particular implementation - the particular implementation always requires manual attention to make sure you don't get any undesirable visuals no matter the technology (in Rembrandt, you might want to fiddle with the light volumes to prevent light from leaking out for instance) - there's no world in which you can get highly realistic visuals with just a few lines of XML.

Edit: I guess what's bothering me here are things like offhand declarations of offline technqiues like raytracing as 'less sophisticated' in comparison with a technique in which light can shine through a solid wall. And also the notion that any of the obvious weaknesses of the Rembrandt lighting technique can easily be worked around, whereas equally obvious but different weaknesses of the ALS lighting techniques are perceived as fundamantal flaws of the framework which can't possibly be mitigated.

I think that's a very wrong picture. I see two techniques with different strengths and weaknesses, the weaknesses can always be mitigated, and what's better really depends on the situation, on available performance and last also on taste.

But you can't have the ease of a shadow map and the semi-transparent shadows and caustics of the ALS interior effect at the same time - you have to make a choice how to do it.
Thorsten
 
Posts: 12490
Joined: Mon Nov 02, 2009 9:33 am

Re: F-20 development

Postby Josh_grtuxteam » Fri Dec 30, 2016 3:27 pm

The whole system is intentionally designed the way it is, I've tried to explain to you a couple of times why (settings being applied to all scenery elements, not only your plane, user confusion with auto-saved settings, multiple systems fighting for control over the rendering settings, configurations which can't run on weaker systems,...) , if you still don't understand why, then please do whatever you feel you need to do in your own hangar, but don't give bad advice in this forum. Thanks.


Thorsten,
Do you mean , that forum is reserved ONLY for peoples who are official, granted by the FG development team, same process than for getting access to FGAddon.
Considering that any external advice is bad.

I very surprised, you or any moderator should have said that, and written within the forum rules before i start to present the Aircraft by GRTUX updated.
I don't mind to retire from that forum.
However what about the users if there is no way to communicate, freely.

Are you moderator ?
If not i would like to get a real moderator advice.

Again polluting the great F-20 development topic
please Thorsten next time open a new topic, for your remark against me, that's would be more convenient.

Josh
User avatar
Josh_grtuxteam
 
Posts: 106
Joined: Mon Oct 17, 2016 3:58 pm
Location: France
Version: lastStable
OS: Linux OpenSUSE

Re: F-20 development

Postby Thorsten » Fri Dec 30, 2016 3:38 pm

Do you mean , that forum is reserved ONLY for peoples who are not official, granted by the FG development team, same process than for getting access to FGAddon.


I told you what the problem is, I gave explanations why this is so, I made a request that you stop creating additional workload for me (as the person most likely to deal with rendering bugs in this forum) and for the rest of the devel team, I said 'please' making that request - kindly where do you read any hostility into that?

Where did I say 'any external advice is bad' - your particular piece of advice is bad for the reasons mentioned, and I'd say the same thing if Curt as forum admin had given it - because I believe the reasons cited are valid. So please keep this to the point and don't construct anything against you personally from nothing.
Thorsten
 
Posts: 12490
Joined: Mon Nov 02, 2009 9:33 am

Re: F-20 development

Postby wkitty42 » Sat Dec 31, 2016 11:36 pm

Flying toaster wrote in Fri Dec 30, 2016 9:48 am:You summarise quite well my feelings. Merging ALS and Rembrandt would mean fixing a couple of things to merge both lighting models.

do you guys not understand the basic difference between ALS and Rembrandt?
one is like a vehicle where you sit out in front of the tires to drive it... the other is like a vehicle where you sit all the way in the back to drive it...

one effectively draws the scene from the back to the front and paints over items that could have not been painted at all... the other effectively paints the scene from front to back and doesn't paint what can't been seen from the current point of view... they each have their uses but they cannot be "merged" so simply to become one...

that's my simple layman's understanding... hopefully it is fairly accurate and helps others in some small way...
"You get more air close to the ground," said Angalo. "I read that in a book. You get lots of air low down, and not much when you go up."
"Why not?" said Gurder.
"Dunno. It's frightened of heights, I guess."
User avatar
wkitty42
 
Posts: 9146
Joined: Fri Feb 20, 2015 4:46 pm
Location: central NC, USA
Callsign: wk42
Version: git next
OS: Kubuntu 20.04

Re: F-20 development

Postby abassign » Sun Jan 01, 2017 11:23 am

Image

For this picture, which I used as my cover image on FB for the new year, I must thank least three subjects:

    "Flying toaster" for the beautiful F20
    "Thorsten" for ALS, and many other things that is putting into the project
    "The FGFS community" to continue to believe in this project

Happy New Year!
Developer of the program https://wiki.flightgear.org/Julia_photoscenery_generator
FDM developer of the G91R1B aircraft https://wiki.flightgear.org/FIAT_G91R1B
JSBSim collaborator
abassign
 
Posts: 947
Joined: Mon Feb 27, 2012 6:09 pm
Location: Italy (living 5 Km from airport LIME)
Callsign: I-BASSY
Version: 2020.4
OS: Ubuntu 20.10

Re: F-20 development

Postby bugman » Mon Jan 02, 2017 8:33 am

wkitty42 wrote in Sat Dec 31, 2016 11:36 pm:
Flying toaster wrote in Fri Dec 30, 2016 9:48 am:You summarise quite well my feelings. Merging ALS and Rembrandt would mean fixing a couple of things to merge both lighting models.

do you guys not understand the basic difference between ALS and Rembrandt?
one is like a vehicle where you sit out in front of the tires to drive it... the other is like a vehicle where you sit all the way in the back to drive it...


I would say it's more like languages - German and French for example, both do the job in quite different ways. Good luck with that merging ;)

Regards,
Edward
bugman
Moderator
 
Posts: 1808
Joined: Thu Mar 19, 2015 10:01 am
Version: next

Re: F-20 development

Postby Flying toaster » Tue Jan 03, 2017 7:27 pm

Happy new year everybody

Just a faint doubt crosses my mind...
When I output the transformation matrix for the aircraft, I consistently get the identity matrix ...
Could it be that the aircraft model is the root of the scenegraph in FG ? Or is it just my code which is wrong ?
Thanks in advance

Enrique
Flying toaster
 
Posts: 390
Joined: Wed Nov 29, 2006 7:25 am
Location: Toulouse France

Re: F-20 development

Postby swampthing » Tue Jan 03, 2017 11:56 pm

Flying toaster wrote in Wed Nov 30, 2016 8:46 pm:
swampthing wrote in Wed Nov 30, 2016 8:04 pm:Off topic. I see screens. Do you plan on putting radar in this model?


I don't plan to have a weapons system on that thing (the stores you see are only ballast). So the radar would be a lot of development for little benefit. I won't prevent anybody from doing it though.

Cheers

Enrique


Just to clarify, I understand you have no problem with someone making a working radar. Would you also have no problem with someone making a working weapon system?
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 5:10 am
Location: Missouri
Callsign: swamp
Version: 2018.2
OS: multiple

Re: F-20 development

Postby Flying toaster » Wed Jan 04, 2017 12:39 am

swampthing wrote in Tue Jan 03, 2017 11:56 pm:Just to clarify, I understand you have no problem with someone making a working radar. Would you also have no problem with someone making a working weapon system?


I can't prevent other people from doing it, as I can´t prevent people from doing the real thing no matter what my opinion on the subject. I won't even try to (the licence in the end will be open source, most likley GPL), but, as I said, I won't assist either. I can only be responsible for my own deeds

Cheers

Enrique
Flying toaster
 
Posts: 390
Joined: Wed Nov 29, 2006 7:25 am
Location: Toulouse France

Re: F-20 development

Postby swampthing » Wed Jan 04, 2017 1:18 am

I guess what I was looking for was whether it would cause bad blood. I don't know your opinion on the subject is, I just got the impression thats its not something you were interested in putting time into.
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 5:10 am
Location: Missouri
Callsign: swamp
Version: 2018.2
OS: multiple

PreviousNext

Return to Aircraft

Who is online

Users browsing this forum: No registered users and 16 guests