@D-73:
I might try to organize a fork once I have retired. But I have a feeling that somebody (more competent) will do that before me
@Richard:
I have great respect for you as a person and for your technical knowledge. Even though I do not quite understand some of your points about Ambient Occlusion (most probably my lack of knowledge) and I am not really with you about how easy or not it is to contribute to FG source code, I will therefore politely keep my mouth shut and try to make the best of the situation.
I hope that I am allowed to ask you about some points regarding Ambiend Occlusion, OSG etc. in the future once I had time to look deeper into it.
Already some thoughts:
* Baking AO in maps is ONE technique used by DCS, but clearly not the only one. Wings and Tail cast shadow on the fuselage dynamically and when moving around the aircraft in close distance you can really see the difference between real AO and baked maps.
* Cube maps (esp. inside cockpits) will cast the shadow from the outer shell to the inside. However, in DCS you can clearly see that ALL 3D-objects (well many of them...some are also baked maps, but again, one can see that!!) INSIDE the cockpit cast shadows on neighboring objects/faces dynamically. Also you can clearly see that even nuts and bolts are REAL 3D-objects and not baked into a map (visible if you move your head accordingly).
* I am sure that DCS has more professional livery developers. But -as I said above- simulating shadow by drawing them into the livery (that's what many relatively good looking AC in FG do) and baking bump-maps does not really compare to what DCS does. So I suspect a combination of more professional AC developers AND a better AO-system is used in DCS.
But as stated, I will try to learn more as I go and would be happy to be allowed to ask further questions once my knowledge is up to the point that I CAN ask the good questions....
Help! Why is FG so bad?
Re: Help! Why is FG so bad?
>I am not really with you about how easy or not it is to contribute to FG source code
The hardest part is understanding the core source and being able to write in C++. It took me months to get up to speed with the codebase enough to make changes. The first changes I made were simple - to fix segfaults on closedown.
>I hope that I am allowed to ask you about some points
you don't have to ask this - I'm always open to discussions and questions.
--------------
Realtime 3d rendering is often a bag of tricks to balance performance against what it looks like. If we could do real time ray tracing that would be the best - the new generation of Nvidia cards seem to allow (more tricks) to make this possible, but this technology is in its infancy and it'll be interesting to see where we are in a few years.
Ambient occlusion is a rendering trick to avoid having to do more complex shadow calculations; it generally makes things look better rather than being a real world representation. It can also help to make seams more visible; and is similar in concept to bump maps (normal maps).
One version of AO is screen space ambient occlusion(SSAO)[1] which uses a depth buffer.
The casting of shadows as you see in DCS when the wings have shadows cast upon them could either be a shadow cubemap or deferred rendering, or any of a myriad of techniques[2]. Casting of shadows from a moving object onto static object (e.g. the ground) could also be a cubemap or other techniques.
For the purposes of modelling baking an Ambient Occlusion map is still the fastest. Rembrandt (deferred rendering) can calculate AO (using SSAO[3]); and there are techniques for calculation SSAO in a forward rendering (default, ALS) but these require some sort of depth buffer which would probably be calculated offline or computed in another rendering pass (i.e. expensive in FPS).
>more professional AC developers AND a better AO-system is used in DCS
You're right that it is the combination of professional 3d modellers, texture artist and rendering engine programmers that make the difference. DCS, X-Plane, and P3D all have bettering rendering engine support for shadows and AO than FG does
It would be great to see FG with an improved (or rewritten) rendering engine and I'm fairly certain that none of the core developers would object to this, in fact when IceCodeGL proposed this he was encouraged to do this even if it means dropping deferred rendering. There are other areas that FG could improve; for example the procedural landclass shaders could do a much better job if there was some way of telling the shaders about the neighbouring landclasses - but this isn't easy to do.
Pretty much modelling in FG (and any other sim) is about understanding what techniques are available and how to use them; but this is often a time consuming and difficult process.
-------------------
[1] https://learnopengl.com/Advanced-Lighting/SSAO
[2] https://www.nvidia.com/object/doc_shadows.html
[3] https://en.wikipedia.org/wiki/Screen_sp ... _occlusion
The hardest part is understanding the core source and being able to write in C++. It took me months to get up to speed with the codebase enough to make changes. The first changes I made were simple - to fix segfaults on closedown.
>I hope that I am allowed to ask you about some points
you don't have to ask this - I'm always open to discussions and questions.
--------------
Realtime 3d rendering is often a bag of tricks to balance performance against what it looks like. If we could do real time ray tracing that would be the best - the new generation of Nvidia cards seem to allow (more tricks) to make this possible, but this technology is in its infancy and it'll be interesting to see where we are in a few years.
Ambient occlusion is a rendering trick to avoid having to do more complex shadow calculations; it generally makes things look better rather than being a real world representation. It can also help to make seams more visible; and is similar in concept to bump maps (normal maps).
One version of AO is screen space ambient occlusion(SSAO)[1] which uses a depth buffer.
The casting of shadows as you see in DCS when the wings have shadows cast upon them could either be a shadow cubemap or deferred rendering, or any of a myriad of techniques[2]. Casting of shadows from a moving object onto static object (e.g. the ground) could also be a cubemap or other techniques.
For the purposes of modelling baking an Ambient Occlusion map is still the fastest. Rembrandt (deferred rendering) can calculate AO (using SSAO[3]); and there are techniques for calculation SSAO in a forward rendering (default, ALS) but these require some sort of depth buffer which would probably be calculated offline or computed in another rendering pass (i.e. expensive in FPS).
>more professional AC developers AND a better AO-system is used in DCS
You're right that it is the combination of professional 3d modellers, texture artist and rendering engine programmers that make the difference. DCS, X-Plane, and P3D all have bettering rendering engine support for shadows and AO than FG does
It would be great to see FG with an improved (or rewritten) rendering engine and I'm fairly certain that none of the core developers would object to this, in fact when IceCodeGL proposed this he was encouraged to do this even if it means dropping deferred rendering. There are other areas that FG could improve; for example the procedural landclass shaders could do a much better job if there was some way of telling the shaders about the neighbouring landclasses - but this isn't easy to do.
Pretty much modelling in FG (and any other sim) is about understanding what techniques are available and how to use them; but this is often a time consuming and difficult process.
-------------------
[1] https://learnopengl.com/Advanced-Lighting/SSAO
[2] https://www.nvidia.com/object/doc_shadows.html
[3] https://en.wikipedia.org/wiki/Screen_sp ... _occlusion
Re: Help! Why is FG so bad?
The hardest part is understanding the core source and being able to write in C++. It took me months to get up to speed with the codebase enough to make changes.
Well, if there is nothing newer than this:
http://wiki.flightgear.org/UML_Diagrams
http://wiki.flightgear.org/Category:Software_diagrams
then I do not wonder.
And again: We have nice clouds and a kazillion aircraft, but.....
But probably there IS better documentation and I am too stupid to find it?!?!
Re: Help! Why is FG so bad?
hans05 :
If You don't like FG, do not use them. Now, I'm flying with FSX and PMDG 737. It is far better than FG, all works as declared, LNAV, VNAV, full functional CDU, etc. I need not 500 aircrafts and no one is 100%. For me is enough 1 plane, but perfect. I got it and I'm happy. Check Steam, if You will be lucky, You can catch super action price 9.99 EUR on MS FlightSim X. FSX is dead, but still very good. Check my thread https://forum.flightgear.org/viewtopic.php?f=3&t=35073. There is unbelievable number of the freeware airports, many of them are far better than in FG. There is freeware weather engine, IMHO, again better than FG's advanced weather, many photoreal sceneries, freeware utility for realtime download virtual earth texture for scenery etc. etc etc. And if You will spend 70 EUR for PMDG, or FSLabs A3xx, or some Aerosoft's product, You will get perfect quality for Your money. If You don't miss CDU or AIRAC procedures, You will be satisfied without payware aircrafts.
I have bad experience with opensource or freeware software, nothing works on 100%, everytime I need fight with some troubles or non standard solutions.
But FlightGear is the best opensource known for me.
If You don't like FG, do not use them. Now, I'm flying with FSX and PMDG 737. It is far better than FG, all works as declared, LNAV, VNAV, full functional CDU, etc. I need not 500 aircrafts and no one is 100%. For me is enough 1 plane, but perfect. I got it and I'm happy. Check Steam, if You will be lucky, You can catch super action price 9.99 EUR on MS FlightSim X. FSX is dead, but still very good. Check my thread https://forum.flightgear.org/viewtopic.php?f=3&t=35073. There is unbelievable number of the freeware airports, many of them are far better than in FG. There is freeware weather engine, IMHO, again better than FG's advanced weather, many photoreal sceneries, freeware utility for realtime download virtual earth texture for scenery etc. etc etc. And if You will spend 70 EUR for PMDG, or FSLabs A3xx, or some Aerosoft's product, You will get perfect quality for Your money. If You don't miss CDU or AIRAC procedures, You will be satisfied without payware aircrafts.
I have bad experience with opensource or freeware software, nothing works on 100%, everytime I need fight with some troubles or non standard solutions.
But FlightGear is the best opensource known for me.
Last edited by V12 on Tue Jan 15, 2019 11:04 am, edited 1 time in total.
Re: Help! Why is FG so bad?
I advise not to get any Black box Airbus - apart from FMGC the IDG aircraft are better. Aerosoft looks good but is not great although better than Blackbox and IDG.
This looks interesting and adds many effects to the 737:
https://oldprop.com/sim/737Immersion/
This looks interesting and adds many effects to the 737:
https://oldprop.com/sim/737Immersion/
Re: Help! Why is FG so bad?
Best Bus is probably from FSLabs with price around 100 EUR (FSX) and 139 EUR for P3D version. Too expansive
EDIT :
If You want CDU in boxed aircrafts in FSX or P3D, this one http://c-aviation.net/vasfmc-freeware-fmc-for-fsx/ should be fine. You will need AIRAC data from Aerosoft, or Navigraph and that is again payware.
EDIT :
If You want CDU in boxed aircrafts in FSX or P3D, this one http://c-aviation.net/vasfmc-freeware-fmc-for-fsx/ should be fine. You will need AIRAC data from Aerosoft, or Navigraph and that is again payware.
- Wecsje
- Posts: 167
- Joined: Wed Aug 16, 2017 4:25 pm
- Location: The Closet, Under the Stairs, the Netherlands
Re: Help! Why is FG so bad?
If you want to go the cheap and beautifull way, go xplane 11 (around 60 euros), and use the zibo 737 mod (freeware add-on, but extremely well done). Also, photoscenery is really easy to use for xplane (and free), the amount of freeware airports is huge. And if you want to go airbus, get the FlightFactor 320 (70 euros). Also, model matching and connection to vatsim is superb.
And as an extremely big point, most default airports in xplane are already of superb quality.
C.
And as an extremely big point, most default airports in xplane are already of superb quality.
C.
Twitch Streams: https://www.twitch.tv/wecsjelive
Contact methods: Discord (Wecsje#6351), FlightSims United discord (https://discord.me/flightsimsunited), Steam (Wecsje)
Track me on VATSIM: https://vatstats.net/pilots/1397313
Contact methods: Discord (Wecsje#6351), FlightSims United discord (https://discord.me/flightsimsunited), Steam (Wecsje)
Track me on VATSIM: https://vatstats.net/pilots/1397313
- Wecsje
- Posts: 167
- Joined: Wed Aug 16, 2017 4:25 pm
- Location: The Closet, Under the Stairs, the Netherlands
Re: Help! Why is FG so bad?
A selection of some recent screenshots of Xplane 11 with the FF320
Twitch Streams: https://www.twitch.tv/wecsjelive
Contact methods: Discord (Wecsje#6351), FlightSims United discord (https://discord.me/flightsimsunited), Steam (Wecsje)
Track me on VATSIM: https://vatstats.net/pilots/1397313
Contact methods: Discord (Wecsje#6351), FlightSims United discord (https://discord.me/flightsimsunited), Steam (Wecsje)
Track me on VATSIM: https://vatstats.net/pilots/1397313
Re: Help! Why is FG so bad?
hans05 wrote:And again: We have nice clouds and a kazillion aircraft, but.....
But probably there IS better documentation
Back in 2014 when I got back into desktop simulation I did a complete review of everything on the market (at that time); which was FSX (bought a gold edition for 20eur); X-Plane (tried the demo), FG, YSFlight and a few others that I can't remember.
I ended up with FG because it best suited what I wanted. FSX seemed to me to require a lot of plugins and purchases to become sensible; I didn't like the way that the F-4 flew in X-Plane and at that time possibly the rendering was quite similar. I was mainly flying out of KSFO and environs and it was fun. I'd been following FG since the early versions I even built 0.x on a unix box, and was always impressed with the stars and other 'minor details', but really it wasn't until 2014 that I found it to be polished enough to be flyable; that's not to say that the versions between my occasional tests (probably 1998, 2003, 2007) weren't any better I just didn't try them.
But the F-14 in FlightGear was a lot of fun; although I ignored it's somewhat dubious flight characteristics. It wasn't until my son (aged 14) said "Dad, this just doesn't fly right" (I'd managed to get him onto a level D sim so he had a few hours of 737 time under his belt). After he said this I googled F-14 aerodynamics and just by luck I found the NASA report that made me think that I just had to redo the aero model on the F-14.
For me FlightGear does a lot of things well, there are some things that need improvement, some bugs that are annoying that we haven't tracked down yet; but really the main thing that we need is to do something about the scenery. We need to regenerate the whole world and fixup the landclass handling and terrain rendering to make it look less patchy. At the same time we need better import of the X-Plane gateway airports (and stuff) and it'll be a transformation. But we don't have much in the way of people to do all of this work - and it's difficult to build the terrain model for the world and have it look good. So although it's something that would transform FG I still don't see this happening anytime in the next few years. I think it will eventually happen - but we need some gifted scenery generation people to take it on board and do the work - and these people are few and far between, especially those who'd do it in their spare time.
So if FG isn't fun X-Plane is a good alternative. It's very suited to GA and subsonic, has a really great rendering engine (which keeps getting better) and it's easily worth the $60. I tried P3D developer ($10/month) and that's also really good but I don't qualify for the academic licence so it'd be quite expensive at $200 for me; plus the licencing terms don't really permit home usage.
FlightGear isn't for everyone; there's currently quite a lot of work going on to improve things - but the actually delivery rate is low because we all do this in our spare time.
-------------------
In terms of the FlightGear codebase the code is well structured and for anyone with experience of medium to large C++ projects it's fairly easy to understand. I'm not aware of any detailed technical documentation.
As always if anyone is interested in working on anything specific then I'm happy to provide guidance about where to find things and suggest how to do things.
Re: Help! Why is FG so bad?
Wecsje wrote:If you want to go the cheap and beautifull way, go xplane 11 (around 60 euros), and use the zibo 737 mod (freeware add-on, but extremely well done). Also, photoscenery is really easy to use for xplane (and free), the amount of freeware airports is huge. And if you want to go airbus, get the FlightFactor 320 (70 euros). Also, model matching and connection to vatsim is superb.
And as an extremely big point, most default airports in xplane are already of superb quality.
C.
Definitely - the Zibo is excellent - and in fact it's almost as good as the PMDG aircraft while looking better because of X-Plane's renderer / lighting
Default weather and clouds is still not as good
Return to “Technical Questions”
Who is online
Users browsing this forum: No registered users and 1 guest