StG2_Manfred Posted March 3, 2015 Posted March 3, 2015 As the other thread is closed now, I want to answer here to LukeFF's video about airflow... Because if we have this in game I'm wondering why the smoke column here (0:40) for example is not affected at all https://www.youtube.com/watch?v=MmrMapANQ5M One thing is a case study what's an engine can do in principal and the other thing is what we actually have in game.... 1
6./ZG26_Custard Posted March 3, 2015 Posted March 3, 2015 I think if you look carefully at the video at the 40 second mark the aircraft are in the foreground and the smoke is in the background and the planes don't fly through the column of smoke.
Finkeren Posted March 3, 2015 Posted March 3, 2015 The smoke collumn is a particle effect. Just because the effect of air flow isn't modelled on particle effects (talk about a performance hog) doesn't mean that it isn't modelled for the airfoil, where it truly matters.
StG2_Manfred Posted March 3, 2015 Author Posted March 3, 2015 Correct, the video is not perfect, admitted. It was just the "best" I could found in a quick search. But I repeat my statement, that there is not a single molecule of air in this sim or in any other. It's called object oriented programming with classes and instances and each instance has it's own attributes and if you want to simulate air and it's flow you have to generate for each molecule an own instance. As soon as you would have generated a couple of cubic meters of air your PC is at its limits. The external model of the planes have absolutely no effect on how a plane fly. Actually they are exactly those instances I've mentioned above, derived from there classes. And the attributes of the particular class defines the flight behaviour. Everyone who ever used an object oriented programming language knows what I'm talking about. You also can create a rectangular cube with the same flight performance of the current planes.
1CGS LukeFF Posted March 3, 2015 1CGS Posted March 3, 2015 (edited) And to just add something else as well from the other thread: Sure, nice to see Luke admit it is the same engine after 2 years of him propagating the fact that the engine is different. I've never said such a thing. It's the same base engine (Digital Nature) powering both BoS and ROF. Edited March 3, 2015 by LukeFF
SharpeXB Posted March 3, 2015 Posted March 3, 2015 What Ive heard about DN is that the effect is there, you can see the balloons react to it. But to make the smoke do that would mean millions of little particles doing the same, instead of a few hundred ballons, so the PC can't show that. It would be cool someday... But not today.
GOAT-ACEOFACES Posted March 3, 2015 Posted March 3, 2015 Because if we have this in game Glad to see your brought this back up.. So, are you willing to admit that I was not as ridiculous as you first thought I was?
=362nd_FS=Hiromachi Posted March 3, 2015 Posted March 3, 2015 This air-flow is pretty nice, but I'd see vortices rolling on the wings ...
migmadmarine Posted March 3, 2015 Posted March 3, 2015 You do understand that each type of object or effect has to be coded to respond to airflow, so just because smoke is in the air, and the air moves, in a video game that does not mean the smoke moves too as it would in the real world, unless it's coded to. Similar to object clipping, while the object's visual model can appear to be there, it's collision or physics model may not be, meaning objects around it don't interact with it.
1PL-Husar-1Esk Posted March 3, 2015 Posted March 3, 2015 Well try fly behind and little higher of pusher plane like ROF D.H 2 - you will notice prop wash or else said distrubted airflow behind pusher.
SharpeXB Posted March 3, 2015 Posted March 3, 2015 I think I've been hit by propwash in RoF from another plane zipping by.
StG2_Manfred Posted March 3, 2015 Author Posted March 3, 2015 What Ive heard about DN is that the effect is there, you can see the balloons react to it. But to make the smoke do that would mean millions of little particles doing the same, instead of a few hundred ballons, so the PC can't show that. It would be cool someday... But not today. Exactly, you've got it! You do understand that each type of object or effect has to be coded to respond to airflow, so just because smoke is in the air, and the air moves, in a video game that does not mean the smoke moves too as it would in the real world, unless it's coded to. Similar to object clipping, while the object's visual model can appear to be there, it's collision or physics model may not be, meaning objects around it don't interact with it. I do, but to respond to airflow we need the air in first place. Those white balls in the promo video could be the air molecules for example. But how many of them do we see, a couple of hundres, maybe a couple of thousands. How many of them would we need if they had to be much smaller and we want to fill the complete stalingrad map with them? Well try fly behind and little higher of pusher plane like ROF D.H 2 - you will notice prop wash or else said distrubted airflow behind pusher. I believe what you say, but to be pushed aside from another instance you need no air at all, just a good 3D engine. And just to clarify my statement, I don't blame DN engine or 777! What I want say is that no PC can compute the behaviour of real air or its flow in a large amount, and therefore we don't have this in the Sim, nor in any other.
-TBC-AeroAce Posted March 3, 2015 Posted March 3, 2015 X plane uses a cell structure around the plane that does account for geometry. This is a very basic version of cfd that is the computing standard for aerodynamics but even this simplified version I guess will have a larger computing overhead. If we wanted to directly simulate airflow and all the forces..... We may need to wait for quantum computers
JtD Posted March 3, 2015 Posted March 3, 2015 I don't think you'd need to model every air molecule on the map to model air flow reasonably accurate. You can use simplified models in areas of lower importance (such as far away from aircraft) and still be fairly precise where it matters. The question is also what would you consider as modelled air flow? Adding a map wide wind vector is also an airflow model, but I'd say nothing that really qualifies. So what does qualify?
GOAT-ACEOFACES Posted March 3, 2015 Posted March 3, 2015 (edited) I don't think you'd need to model every air molecule on the map to model air flow reasonably accurate. Agreed 100% Vectors for the win! Edited March 3, 2015 by ACEOFACES
StG2_Manfred Posted March 3, 2015 Author Posted March 3, 2015 I don't think you'd need to model every air molecule on the map to model air flow reasonably accurate. You can use simplified models in areas of lower importance (such as far away from aircraft) and still be fairly precise where it matters. The question is also what would you consider as modelled air flow? Adding a map wide wind vector is also an airflow model, but I'd say nothing that really qualifies. So what does qualify? Even simplified it would be way too much. You would already need, let's say millions of particles just to cover the airframe. And yes you can add a wind vector, which is by the way a nice effect, but this 'wind' doesn't stream around your fuselage, it just push all objects in the environment continously into one direction.
GOAT-ACEOFACES Posted March 3, 2015 Posted March 3, 2015 Yet it has already been done, 20 years ago http://en.wikipedia.org/wiki/Flight_Unlimited
JtD Posted March 3, 2015 Posted March 3, 2015 (edited) Even simplified it would be way too much. You would already need, let's say millions of particles just to cover the airframe. And yes you can add a wind vector, which is by the way a nice effect, but this 'wind' doesn't stream around your fuselage, it just push all objects in the environment continously into one direction. Yes, but you can do a couple of functions for specific objects and have them interact. Such as [behaviour of aircraft1]=[global atmosphere function]+[local weather functions]+[impact of other objects functions]. For instance, if you wanted to model wing vortexes and the interaction with following aircraft, you could take a time/distance limited function, composed of a couple of vectors, check for interference with other aircraft and if so, do the proper maths. So now you have wing vortexes included in the FM, and that's pretty airflowish. Even without a single particle of air modelled. I understand that you mean modelling air flow as in modelling air as such with a fairly good resolution, and if so, no, air flow isn't modelled, certainly not globally. But, if I was asked to model a circle and due to computing restrictions am using a square, I'm still modelling a circle, just very primitively. Same can be true on a more complex scale for air flow, if you catch my drift. Edited March 3, 2015 by JtD
StG2_Manfred Posted March 3, 2015 Author Posted March 3, 2015 (edited) Exactly JtD and that's all I wanted to say. You can simulate vortexes like you explained in your example and also other things which are necessary to get a plausible FM but you have to do that for each effect/behaviour separately (and always consider the CPU power alongside). In FM discussions it is often claimed that a certain behaviour of a plane occurs as a unavoidability of the FM or the air around it. But every behaviour/effect happens on purpose, or more precisely has to be added to the sim (as the wind vortexes in your example). Maybe this helps someone next time a FM discussion emerges to get a better understanding of how it works. As far as I'm concerned I'm good for now. Thank you for your contribution! Edited March 3, 2015 by StG2_Manfred
-TBC-AeroAce Posted March 3, 2015 Posted March 3, 2015 In A vortex model, information depending on resolution would be a quite large matrix of data even if resolution was low and distance limited. Then u have to recalculate every time step based on the induced angle of attack. I have made manytime irrelevant lifting line models and not very simple to get flow field calculations even from simplest of models
GOAT-ACEOFACES Posted March 3, 2015 Posted March 3, 2015 (edited) Even simplified it would be way too much. I would have to say that comment does not hold much weight.. In light of the FACT that it was done on PCs 20 years ago (Flight Unlimited) and apparently 1C/777 is doing it (atmosphere simulation) today, or at least tried it in RoF. Which should not be confused with me saying the RoF FM is a real-time computational fluid dynamics based FM like Flight Unlimited was 20 years ago. AFAIK the RoF FM is still a 6DOF FM, and thus the atmosphere simulation may be just for smoke, clouds, etc. if you catch my drift. Pun intended? Edited March 3, 2015 by ACEOFACES
Rodolphe Posted March 3, 2015 Posted March 3, 2015 ... A good example of Prop Blast on the ramp. https://www.youtube.com/watch?v=5HbwxR2dRlc&feature=youtu.be ...
303_Kwiatek Posted March 3, 2015 Posted March 3, 2015 Airflow is modeled but it still there is no slipstream behind props. We checked it trying to fly close behind other plane ( below, the same level behing or little side) but there was any turbulent air.
GOAT-ACEOFACES Posted March 3, 2015 Posted March 3, 2015 I have not look at this in years, i.e. computational-fluid-dynamics But looks like they can do allot more than they were doing 20 years ago! Go Figure! And note the following is more for analysis! Thus, in theory a simplified versions of what they do here could be implemented on a PC for a 'game' that does not require the level of detail an analysis tool would require. http://www.simspace.es/esoriano/solutions/computational-fluid-dynamics/ Long-term experience with CFDs makes it possible to develop reliable, efficient tools, even applied to Real Time simulation. CFD tools can be used for generating realistic flight data in simulators, as well as for creating innovative solutions running in real time, like our Real Time Wind Server©. Real Time Wind Server© is a simulation software used to inject highly realistic wind profiles into real-time flight simulations (i.e. helicopter/aircraft/UAV, shipboard operations, A-A refueling). Runs on a workstation and can simulate: Wakes behind moving vehicles: ships, airplanes, cars, etc. Wind profiles around static objects: terrains, buildings, fires, etc. Realistic Wind Computation Process:An offline process computes the detail wind characteristics of an specific scenario, such as a ship wake, using Computer Fluid Dynamics (CFD). A post-processing analysis and data reduction of the CFD results generates a real-time database optimized for real-time. Real-Time Wind Server:The real-time server is designed to function as an isolated process which receives the real-time input information via LAN and respond within 100Hz. Wind information is computed at 100 points around the requested point and includes steady and turbulent component. Not so ridiculous now is it!
71st_AH_Mastiff Posted March 3, 2015 Posted March 3, 2015 (edited) Airflow is modeled but it still there is no slipstream behind props. We checked it trying to fly close behind other plane ( below, the same level behing or little side) but there was any turbulent air. I took a 109 and flew behind a HE111 @ .10 and was being tossed all over the place, Single player. I don't think it's in MP... Edited March 3, 2015 by 71st_Mastiff
303_Kwiatek Posted March 4, 2015 Posted March 4, 2015 (edited) Behind fighter i dont feel any turbulent. Checked behind He-111 and when i was very close below 10 m i feel slighty turbulent but very weak and narrow. Suprisly little above engine not below. Edited March 4, 2015 by 303_Kwiatek
StG2_Manfred Posted March 4, 2015 Author Posted March 4, 2015 Ok, I give it one more try to explain. Beforehand I apologize for being a bad artist. I made a drawing which makes it hopefully more clear. It's of course possible to simulate a force like the turbulence of a propellor, the question is how it is achieved. In my pic you can see two cases, in case one I "put the wind force together" and have (in the best case) only one more calculation (e.g. speed of plane 2 minus the air pressure caused by plane 1). It does not behave like real air, but for the most situations it would be plausible enough to get the feeling of wind. In the second case, when we want to have "real air flow" with all it's consequences we would have to calculate tons of air particles and this brings a PC easily to it's knees and therefore does not happen.
216th_Jordan Posted March 4, 2015 Posted March 4, 2015 A quadcore with 4 ghz - 4,000,000,000 clock cycles per second - on each core can calculated that in a reasonable manner if its not a bad code.
GOAT-ACEOFACES Posted March 4, 2015 Posted March 4, 2015 Agreed Heck they were doing simplified version of it 20 years ago in PC real time flight sims, I would expect they could do it today too
StG2_Manfred Posted March 4, 2015 Author Posted March 4, 2015 A quadcore with 4 ghz - 4,000,000,000 clock cycles per second - on each core can calculated that in a reasonable manner if its not a bad code. It can't and it just not work that way.
GOAT-ACEOFACES Posted March 4, 2015 Posted March 4, 2015 It can't? Yet they already did.. http://en.wikipedia.org/wiki/Flight_Unlimited Over 20 years ago
Recommended Posts