Armoured glass refraction must see video

[zaelu]

But IMHO simulating it is overkill as I doubt it's noticeable from the pilots seat.

 

 

I doubt refraction is suspended from pilot seat. :D

[bongodriver]

I doubt refraction is suspended from pilot seat. :D

 

It's not obviously, it's clear from the pictures that without refraction the revi sight would be half obscured pretty much in line with the centre of the aiming reticule, refraction is what brings the 'bar' down to just cut the bottom of the sight aperture and that is pretty much how it looks in all current simulators, so nothing needs changing IMO.

[VincentLaw]

I dont think that is the same effect you want here...... you want an actual live, dynamic refraction effect.

Basically, trace some reference rays from the viewpoint through the glass with Snell's law to get the angular distribution, then use backwards mapping to render to texture on the armored glass in real time.

 

Games like Portal use similar techniques, and I have no doubt the developers could implement it in DCS, but it would cost money and development time, so things like the F/A-18C might get pushed back further.

[NineLine]

Well as stated in the other thread, its reported, its in ED's hand if they feel it needs addressing or not.

[zaelu]

It's not obviously, it's clear from the pictures that without refraction the revi sight would be half obscured pretty much in line with the centre of the aiming reticule, refraction is what brings the 'bar' down to just cut the bottom of the sight aperture and that is pretty much how it looks in all current simulators, so nothing needs changing IMO.

 

I think in the picture above it seen a second refraction from the Plexiglas of the bubble canopy. The right top frame of the front glass shows a visible distortion that is seen also in the lower frame where the revi is.

 

IMHO anyone can replicate this effect with some sheets normal glass. The armored one might be a bit dense I think that will amplify the effect a bit more. here is a table of reference:

 

http://hyperphysics.phy-astr.gsu.edu/hbase/tables/indrf.html

 

 

It could go from 1.5 normal grass to 1.7 the armored one it seems... the effect would be significant but still one can make an idea of how much with normal glass.

 

Sure Nemessis used 1.5 thicker armored glass but the effect was also "colossal". The image looks like the periscope of Mig's back seats :) .

 

About simulating the refraction, I think the problem complicates when subtle imperfections in glass are to be simulated but a flat sheet of glass would be simpler.

 

At least this is what I understand from here:

 

http://http.developer.nvidia.com/GPUGems2/gpugems2_chapter19.html

 

But I have no programming knowledge basically.

[Bushmanni]

I doubt refraction is suspended from pilot seat. :D

 

I was referring to how the apparent compression of the insides of the glass changes with view angle.

[Wolf Rider]

simples... just swop the "bar" for the cowling :) I notice the video'er didn't spend too much time there

Edited August 31, 2014 by Wolf Rider
add Lightshot snip

[bongodriver]

simples... just swop the "bar" for the cowling :) I notice the video'er didn't spend too much time there

 

Surely you aren't suggesting the advocates of 'the bar is a lie' are a little 'short sighted'? :music_whistling:

[vicx]

I dont think that is the same effect you want here...... you want an actual live, dynamic refraction effect.

 

Now you are thinking with portals.

 

well you should be thinking with portals.

 

 

Good-bye

Edited October 7, 2014 by vicx
:)

[Wolf Rider]

Now you are thinking with portals.

 

well you should be thinking with portals.

 

 

Good-bye

 

 

a crooked house would be much more fun :)

[NineLine]

This thread is like a never ending looping portal...

[Wolf Rider]

Have you read "And He Built a Crooked House"? - Heinlein ;)

 

 

Plot summary:

 

Quintus Teal, "Graduate Architect", while drinking with his friend Homer Bailey, bemoans the conservatism of American architecture. He wants architects to be inspired by topology and the Picard–Vessiot theory. The conversation turns to four-dimensional objects and he shows Bailey three-dimensional models made of toothpicks and clay, representing projections of a four-dimensionaltesseract, the equivalent of a cube. Bailey is baffled, but when Teal constructs an "unfolded tesseract", a three-dimensional object, Bailey suggests building a house to that pattern. Teal, ever hungry for a commission, plies Bailey with more liquor until the contract is signed, as the story has it, "halfway down the second bottle".

The house is quickly constructed, in its peculiar "inverted double cross" shape (having eight cubical rooms, arranged as a stack of four cubes with a further four cubes surrounding the second cube up on the stack). However, the night before Teal is to show Bailey and his wife around the house, an earthquake occurs. The three of them arrive the next morning to find what appears to be just a single cubical room. Believing the top seven rooms to have been stolen during the night, they go inside to look for clues.

What they find is quite unbelievable: Not only are the upper floors completely intact, but the stairs seem to form a closed loop, in that the stairs from the top room lead back into the bottom room and not to the roof. What is more, there appears to be no way to get back out, because all the doors and even the windows lead directly into other rooms. At one point, they look down a hallway and are shocked to see their own backs. Teal eventually realizes that the earthquake caused the house to fold into an actual tesseract.

Teal tries to play up the benefits of the situation, but in attempting to move from one room to another by way of a French window, he falls outside and lands in shrubbery. Ever the optimist, he notes as he re-enters the house that they do have a way of leaving the structure after all. It seems to have something to do with their state of mind while passing through a window.

Exploring further, they find that the windows of the original top room do not connect where they mathematically "should". One gives a dizzying view from above a skyscraper, another an upside-down view of a seascape. A third window looks out on nothing, that is, a place of no-space, with no color, not even black. The fourth window looks out on an unearthly desert scene. Opening the window they find the air on the other side breathable. Just then another earthquake hits, and so they exit in a panic, through the open window. They find themselves in a desert with unearthly, twisted, treelike vegetation around them, with no sign of the house or the window they just jumped through. They are only slightly relieved when they discover, from a passing truck driver, that they are in Joshua Tree National Monument, and not stranded on another planet.

Returning to the house, they find it has vanished. "It must be that on that last shock it simply fell through into another section of space", Teal remarks. "I can see now that I should have anchored it at the foundations."

 

- wiki

Edited October 7, 2014 by Wolf Rider

[IJG27_Nemesis]

 

 

 

 

 

Update 3: I've established contact to noone less than the legendary id Software developer Legend John Carmack (creator of Doom, Quake, ect...) and asked him how to model such a refraction effect without raytracing. It's a honor for me to speak to him! Here's what I've learned:

 

 

 

Nemesis: "Hi John! Is it possible to calculate refraction of light through armored glass in real time? check:Fw190 Windscreen /youtube"

John Carmack: "Pretty easy to get quite close, but you would need ray tracing to get it exactly right with glass that isn't completely planar."

 

 

fortunately, the Fw190 Windscreen is completely planar, and shaped like a simple cuboid box!

 

 

 

Nemesis: "John, how to model refraction of a rectangular cuboid armor glass without raytracing? I need help on that one...any how-to?"

John Carmack: "The refraction can be baked into a projection matrix. Draw outside the window with this, clear depth, draw inside normally."

 

John Carmack: "Actually calculating the projection matrix will be a math chore."

 

 

 

Now thats good news for us flight simmers who have been waiting for more than a decade for such a great effect in real time WITHOUT large strain on the CPU/GPU. Devs? It's your turn!! Show us of what you're made of, and create the best there is on the planet for us pilots! Heck, I would even pay extra money just to have that windscreen fixed :D

[Random]

Hopefully it'll get implemented some day but I'm not holding my breath!

[AndyHill]

That Zaelu's image shows a far more massive difference than I imagined. Whether or not something is done about this issue is another matter altogether, but anyone with contacts to ED please ask them to do that refraction rendering in 3D max and then do another shot from the exact same position with regular rendering. That would be a fantastic and unique opportunity to demonstrate to the community just what they've been talking about for years. I mean something like 35% of the Internet is now filled with FW bars. Just PLEASE do it. It's totally worth it. The video is wonderful, but it's not an exact replica of a 190 cockpit - but luckily ED now have one.

 

And then try the same thing with the P-51, it should also be affected to a degree as are most other WW2 planes, especially the La series. And then we can move on to discussing the effect of stereo view on vertical beams...

[Bourrinopathe]

Very interesting thread.

I assume a refraction rendering should modify the MiG-15bis (and similar windscreen designs) view accordingly?

 

Edited May 5, 2015 by Bourrinopathe

[AndyHill]

Basically yes. That's a pretty neat picture by the way, how did you approximate the effect?

 

Additionally stereoscopic rendering (which is becoming a thing with Oculus & co.) will make vertical cockpit supports look much thinner. Everyone can try sitting in a (stationary!) car, cover one eye and observe the apparent width of the vertical supports change depending on the number of eyes used.

 

That's of course an entirely different issue than the refraction stuff in this thread, but I think it's important to realize that simply creating a model that is accurate to the drawings may result in a view that's not even very close to what you would get in reality.

[Bourrinopathe]

Thanks for the confirmation, AndyHill.

 

I have no idea how difficult it could be to implement refraction rendering in the upcoming engine, but with the current WWII aircraft popularity, it would certainly provide a valuable realism/visibility improvement.

 

Apparently, the Unreal Engine can render different indices of refraction:

 

 

(the MiG windscreen above is just a wild guess based on what I've seen in the OP's video)

[thinkr]

First off let me state I know nothing about refraction or 3d modeling or game engines, but is calculating the refractrion of the glass really nessasary to achieve the view of the gunsite you guys want to achieve?

 

I do not know but I imagine that adding another view port camera as seen with the A10 infared cameras drops frame rates dramatically by adding calculations that will result in another heavy hit on our in FPS within the simulator as a whole?

 

I mean taking more processing power away from the whole software that becomes only bottlenecked by firing multiple rockets on most processors, doesnt seem like a good trade, at least not now until we see what DCS v2.0 is like with multiple graphics cards and a top of the line processor. Its just an opinion but If it were me I would choose more accurate projectiles over a better view port, but I can see why its such a big deal.

 

Hydrolics modeling, flight modeling, terrain weather and airflow models and not to mention AI the already cpu intensive and the sim shows us better frame rates with a faster processor but maybe this would be wiser to wait till DCS 2.0 comes out an then reassess this?

 

I dont know about making game engines, but I think this is a case where I would like to mention there is always a balance of factors required in making a simulation realistic and yet playable at acceptable frame rates, which I am sure you already knew and considered.

 

Taking this into account here is my real question?

 

Is it possible to achieve a simulated effect of refraction of the glass view by changing the thinkness of the 3d cockpit model to create a thinner edge without adding more a complex realtime calcuated viewport to the engine?

Then again I dont know how intensive the calculation is so maybe someone can tell us.

Edited September 17, 2015 by thinkr

[IronJockel]

Is it possible for edge to display refraction?

 

Sorry if this is a report.

[karambiatos]

Is it possible for edge to display refraction?

 

Sorry if this is a report.

 

No computer can handle real time refraction well, ED could make a render to texture kind of thing, but that has it's own set of problems, mainly that you could get half as less FPS when looking through the window, since the scene needs to be rendered twice for the computer and depending on the complexity of a scene, it could be unfeasible to do it.

[Oydoron]

To get this going with the current engine, yes, you'd need to do another render just like with the mirrors. In this case though, it'd need to be full quality full frame rate.

 

You just move the point at which the video is rendered closer to the window in a direction tangent to the window plane, let's say 4 cm for sake of this example. In this way if you are looking tangent to the surface of the window, the only distortion is everything is 4cm closer. If you are looking through the window at 45°, everything is 2.8cm closer, but also 2.8cm lower. If something is very close to the window, this makes a big visual difference, the further away it is the less difference it makes. Once you are past the nose, it doesn't really matter. You'd also of course need a different clipping frustum for this render so that things within the cockpit don't appear in the window render. If you end the frustum at a point within the window, it'd only render a portion of the frame giving the effect of the frame being shrunk by refraction.

 

There may also be a way to cheat. Since the primary issue is with things close to the window, you could have a second cowling model that is shifted slightly from the first. Perhaps a shader could only render the original one if the z buffer does not indicate a window intersection, but render the second one if it does. I'm not sure if this is possible with shaders and would just be a cheap trick. It would look exactly the same in 99% of cases. You'd run into issues if other things got close to the nose of your aircraft as they would not receive the same treatment.

[KLR Rico]

Just couldn't let the thread die in peace, eh?

[Starfire]

The simplest and most cost effective way to deal with this would be to remove the bar, minimize the side and top-frames and treat the front view like an ordinary window.

I know that this would offend some of the hardcore sim aficionados but I could live with it.

 

As an alternate they could heavy frames and bars on all the other airkrafts to level out the playing field ***just kidding***