Jump to content
Sign in to follow this  
hailstorm

Parallax effect on the HUD

Recommended Posts

Hailstorm, I'm sorry I wasn't particularly clear. But as Myke says "don't confuse the weapons crosshair with the HUD". You are correct re the parallax, but you are ignoring the HUD.

Imagine a rifle with iron sights fixed to a bench. 300m away you have a target. The rifle is perfectly aligned on it. Sight down the rifle. Move your head around from behind the rear sight. What happens to the placement of the front and rear sight in relation to the target? Think of these sights as the HUD elements.

What I am saying is the function of the existing HUD is wrong. The way the static gun aim point in the HUD works is wrong . I'm not arguing that the in game parallax is wrong just the function of the display. And allowing the display to move around doesn't fix that. What i'm asking for is a change in the way the HUD command works.

Don't get me wrong, I'm pretty sure a real HUD has never, and will never work in how it's depicted in this beta, and also I'm well aware that the game crosshair had no physical link to the HUD.

My point is that due to number of factors (like head movement, possibly badly configured HUDs, etc) the HUD in an aircraft pre-73116 OA was totally useless as an aiming device, and as a player the pilot had to rely on a completely external, unrealistic, game-generated object (i.e the crosshair) for any degree of accuracy.

I too don't like that the idea of a solution is to make the HUD image a moving object, like you said Rock it's unrealistic, it currently causes more problems than it solves, and I personally feel there's lots of other, better ways to fix the problem. BUT, it does solve the parallax problem. I don't like this kind of roundabout fix to a different problem as opposed to fixing the true problem head-on, but at the end of the day, if BI think that this is going to be the solution they're going to run with, and don't have the time/resources for a better solution, I'll have to accept what I'm given. I'm taking solace in the fact the HUD at least now does the job of aiming weapons like it's supposed to.

Personally, I'd prefer that the pilot's viewpoint was completely static in the cockpit when the pilot looks around, and not fixed a foot in front of the rotational axis like it currently is.

With a static viewpoint, It would eliminate the moving HUD problem and still be accurate while looking in different directions. There is a reason most flight simulators do not simulate head movement, after all.

Following Rock's example of the rifle iron sights, the error for the HUD in pre-beta was basically the equivalent of the entire rifle skewing whenever the player aimed to the left/right/up/down of dead centre while looking down the sights - if this happened, there's be a huge outrage from all the infantry players (and we know how huge a majority those guys are) who now couldn't aim and probably instead rely on the aim-from-the-hip crosshair for the majority of aiming.

Share this post


Link to post
Share on other sites

Personally, I'd prefer that the pilot's viewpoint was completely static in the cockpit when the pilot looks around, and not fixed a foot in front of the rotational axis like it currently is.

Does anyone know how is this solved in real fighter planes? The pilot viewpoint is not "a foot in front of the rotational axis", it is just a point close to the eyes in the game. As for real life, anyone can check easily that when you are turning your head left or right, your eyes move left or right a little bit. Those few centimetres to cause a parallax effect strong enough to destroy your aim.

Just count with me:

- HUD distance from head is ~ 1 m

- Eye offset when turning 45 degree left is about 3 cm

- With target distance 1000 m this gives 0.03m/1m*1000m = 30 m error.

Share this post


Link to post
Share on other sites

A while ago i was looking at a holo site scope for my lever action 44 mag rifle and the model i was looking at was parallex free. If i moved my head to side a little the illuminated dot was always on target. It was pretty much parallex free. I also have alot of standard scopes on hunting rifles where parallex can be set or is set at certain distances eg 100, 200, 300m. Basically the scope is parallex free at those distances set but even with head movememt at distances where parallex occurs the aiming reticle will not move much off target and for FFAR it shouldn't matter. i'm guessing the aircraft HUDS work like that in real life but not sure as i havn't researched it yet.

Update:

check this out Suma, just found it in my links

might help http://www.helitavia.com/avionics/TheAvionicsHandbook_Cap_4.pdf

search parallax and there is some really handy info on how the HUD works

Edited by vasmkd
added pdf link on hud info

Share this post


Link to post
Share on other sites

check this out Suma, just found it in my links

might help http://www.helitavia.com/avionics/TheAvionicsHandbook_Cap_4.pdf

search parallax and there is some really handy info on how the HUD works

Great. Reading this I think the real HUDs (at least those with a "wellcorrected optical system") work very closely to what we have. If they are parallax free, this basically means as you are moving your head around, the image projected on the collimator area is "moving", so that it still matches the desired directions in the outside world:

Without a wellcorrected optical system, conformal symbology will not properly overlay the outside world view and symbology will not remain fixed with respect to the real-world view as the head is moved around within

the HUD eyebox.

Share this post


Link to post
Share on other sites
With a static viewpoint, It would eliminate the moving HUD problem and still be accurate while looking in different directions. There is a reason most flight simulators do not simulate head movement, after all.

that is simple not true. FSX , Lockon and IL-2 for example have 6-dof with tools like track-ir and freetrack.

so it would be nice to have 6-dof in arma2-oa in vehicles and planes. so you have a better view , for example landing a chopper looking out of a hmmv as driver :) a.s.o.

suma, if you have any of these games: il-2 and blackshark for example you can

check how they have done that part.

btw. thanks for putting so much work into that game. hoping arma3 is ready soon ;)

greetz

Edited by themaster303

Share this post


Link to post
Share on other sites

@Suma

you should search contact with RKSL-Rock. I guess in this whole Forum no one has better and more accurate informations about this.

Glad to see you put so much effort in interest in this matter. Much appreciated.

From my point only one request: if HUD remains movable, please keep the config setting to disable the parallax effect on purpose. This way, HUD tech can be used for MFD's aswell.

Finally the class is called MFD and not HUD. ;)

Share this post


Link to post
Share on other sites
that is simple not true. FSX , Lockon and IL-2 for example have 6-dof with tools like track-ir and freetrack.

so it would be nice to have 6-dof in arma2-oa in vehicles and planes. so you have a better view , for example landing a chopper looking out of a hmmv as driver :) a.s.o.

suma, if you have any of these games: il-2 and blackshark for example you can

check how they have done that part.

btw. thanks for putting so much work into that game. hoping arma3 is ready soon ;)

greetz

What he means is the effect of motion, for example when you're driving and you turn you're body moves, if you are piloting a helicopter and you turn your body moves, basicly since OA your body resopnds to G-forces..if you brake your body lurches forward, a burst of speed will momentarily push you back and so on.

As for removing it..fixed wing maybe, helicopter NO.

Share this post


Link to post
Share on other sites

Maybe I misunderstand something, but I think what RKSL-Rock has described in his post as a 3D holographic HUD is exactly the moving HUD we have in the game now.

The option which you can you to turn this on/off in the MFD/HUD config will definitely stay.

Share this post


Link to post
Share on other sites

I cross checked this wit the HUD of the Ka-50 in DCS...the behaviour and depiction when moving heads is close to what we have now in OA.

The only thing wrong is the HUD moving with G-loads...only the gun pipper should move that way not the data frame.

Share this post


Link to post
Share on other sites

I made a personal edit that splits the HUD into two, one with the non-moving static components and another with the correcting gunsights. No idea how this matches RL but it's working for me.

Share this post


Link to post
Share on other sites
I made a personal edit that splits the HUD into two, one with the non-moving static components and another with the correcting gunsights. No idea how this matches RL but it's working for me.

Indeed a interesting workaround i could/can live with.

@Suma

i still suggest you contact RKSL-Rock, i guess the most of us can only reference to other games and IMHO, no matter how accurate they are, main reference should always be the real thing.

And while we're talking about HUD's, i hope you don't mind if i point to some things i've remarked or wish to see realized...with all due respect of course. ;)

- since i'm using HUD also for MFD's i've noticed that text and numbers (which technically are also text) is not occluded by the model. The F-16 i work has a center console which occludes the MFD's partially depending on head movement. Line drawings were occluded correctly but text was just shining through.

- data sources for HUD/MFD are pretty limited. Would be nice if at least user data could be projected, maybe reading out a variable which could be defined.

- data source "targets" has a fixed FOV which fits perfectly for HUD use but doesn't work well for MFD's. Maybe introducing a definable FOV is possible?

Share this post


Link to post
Share on other sites

I have found a real life video which I think shows again what we have now is very close to the real thing:

Note how even if the camera is shaking (and as a result the cockpit and canopy are shaking), the HUD sticks to the outside world, it is not moving with the aircraft body.

Share this post


Link to post
Share on other sites

Though I was an S-3 guy and not a Hornet guy (we didn't have HUDs in the Viking) and went through Advanced NFO training before we got the T-45s and T-6s, I did once get a chance to look through the HUD of an F/A-18F powered up on the ground, and it looked pretty much as I expected and the same as the videos that Suma and themaster303 posted. (If anyone has more direct experience with this in real life, please pipe up.)

The symbology is projected at infinity (there might be some way to adjust the focus on some types of HUDs, but I haven't really tried to research this), and the collimation means that the symbology stays the same angular size relative to the background and that the pitch ladder is "fixed" to the ground and sky. In other words, if you were looking at the night sky through the HUD (the FOV isn't very big, though), it would look like the pitch ladder was part of the sky like the stars were. Sort of like in the old collimator gunsights that seem to be represented in the "Wings of Prey" video clip that themaster303 linked to, the symbols would not get bigger or smaller if you moved your head closer to or farther from the HUD, respectively, and they would disappear from view if you moved your head too far left/right/up/down, since they appear to exist on the world outside as seen through the HUD combiner glass and not on the surface of the glass itself.

I don't have TrackIR or anything (yet), so I haven't had a chance to check this out in ArmA2/OA in any meaningful way so far, so i might be able to give more feedback later.

Share this post


Link to post
Share on other sites

Ok, since i perhaps wasn't as clear as i should have been in previous posts, let me explain my own experience has been with 3 aircraft types and HUDs.

  • Harrier GR7 - I have 8+ hours in the Simulator at RAF Wittering.
  • F-16A/B - 4-5 hours on the Belgian AF Simulator when working for Thales.
  • EuroFighter Typhoon F2 - ~7 hours in both static and procedural simulators at BAE Systems

It been my experience that HUDs do not behave how BIS has portrayed them in the earlier patch. They do not move with your view point in the same way as the patch. Specifically, they move in the opposite direction than shown. Nor is the movement as extreme.

Having said that I've never really liked the way the HUDs were represented. So im delighted that you are looking at this so seriously. :D

I have found a real life video which I think shows again what we have now is very close to the real thing:
Note how even if the camera is shaking (and as a result the cockpit and canopy are shaking), the HUD sticks to the outside world, it is not moving with the aircraft body.

That appears to be a "direct projection" reflected HUD. The cheaper end of the scale where perfect alignment ie weapons delivery is not important. The relative "movement" is an illusion sometimes called "bowling". Its only ever seen on single lens arrays and is caused by an uncorrected projector lens. You get the same effect on pretty much any rounded glass surface. A fish bowl is as perfect example. Its corrected by adding a second lens array.

You can see a typical dual array lens arrangement on page 5

http://www.helitavia.com/avionics/TheAvionicsHandbook_Cap_4.pdf

As for the lack of movement even when shaking most modern HUD's are stabilised anyway. (The technology is now very cheap. Its included in most video cameras these days)

Cheaper setups usually only have 1 lens array. Its similar to an overhead projector. The lens magnifies the HUD output, projecting it onto the combiner plate. But this setup is prone to distortion and errors. More complex setups, like those used on most military HUDs where alignment and accuracy are key for weapons delivery use a far more complex setup. Usually a dual lens array in conjunction with a combiner plate with very specific refractive properties so that light is only reflected at certain angles. This reduces glare and prevents "bowling" or "fisheye" effects. The key effect here with the dual lens setup is that it corrects the displacement effect of single lens systems. eg:

  • Single lens - move to right, the HUD display "appears" to move to the left, offsetting the display around the curve of the lens. Making the display appear to move with your view point.
  • Dual or better lens - Move to the right and the display "appears" to move to your right maintaining the alignment on the weapons delivery axis. This corrects for the deflection you get when not properly aligned with a single lens display. It behaves in exactly the same way as the rifle iron sights i mentioned in an earlier post. With the rear sights as the HUD frame and the fore sight as the HUD display itself.

You can see the DUAL Lens effect on the Harrier GR7 HUD here:

Look at 2:34, then 3:40 while manoeuvring under G, shows the described effect perfectly. You should also notice the apparent depth of the projection. It can be clearly seen several times during the clip. This depth is commonly used to help pilots gauge their angle and ensure they maintain alignment to correct for parallax distortion.

Holographic HUDs use the same principle but goes a bit further to use multiple projectors (min 3) and multiple lenses to "layer" the display creating a true 3D display which ensures alignment using the same principle are the rifle iron sights again. With background and foreground ques so the pilot can be sure he's aligned.

Neither system behaves how its portrayed currently in game with the patch.

Edited by RKSL-Rock
typos and gramar

Share this post


Link to post
Share on other sites

I'm sorry Rock, but I can't see what you're trying to describe here :(

In the vid posted by Suma, the dispay also moves to the same direction the head is moving to, I can't catch what you describe :( ("Single lens - move to right, the HUD display moves to the left, offsetting the display around the curve of the lens. Making the display appear to move with your view point.")

It's probably this subtlety which makes the proper implementation difficult for people not in the knowledge.

Share this post


Link to post
Share on other sites
I'm sorry Rock, but I can't see what you're trying to describe here :(

In the vid posted by Suma, the dispay also moves to the same direction the head is moving to, I can't catch what you describe :( ("Single lens - move to right, the HUD display moves to the left, offsetting the display around the curve of the lens. Making the display appear to move with your view point.")

It's probably this subtlety which makes the proper implementation difficult for people not in the knowledge.

OK, it is a bit confusing and perhaps quite a subtle distinction.

With the patch the entire HUD moves so its always aligned on the impact point. Try it for yourself in OA beta build 73116 (its disabled in later builds)

In the video Suma links to the HUD follows your viewpoint. So if you move to the right and look left the HUD will still be aligned to you viewpoint. This behaviour is wrong for a military grade HUD because it gives a false impression of where the weapons/rounds will land.

Imagine that the HUD plate is a goldfish bowl. With the right light on it if you move around you will see that your reflection always aligns to you. This is what the patch does. Its not realistic.

Now in most military HUDs, not all but certainly the majority, this effect is designed out to make sure the pilot can either shoot or drop weapons perfectly on target. This means that our hypothetical fish bowl would have a specially shaped lens sat in front of it, fixed in place. And unless you are aligned with it you wont see your reflection. Align yourself with it and you will see your reflection without distortion. Now as before move to the right. Your reflection would appear to move to the right from your point of view due to the lens correcting the angle of reflection.

Suma's link shows the "Fishbowl" while my Harrier link shows the "corrected lens and fishbowl". We should have the Corrected Fishbowl scenerio if we want realistic HUDs.

Any clearer?

Edited by RKSL-Rock

Share this post


Link to post
Share on other sites

Not quite sure, what I get is this :

* On single lens systems, HUD symbology is always aligned with a fixed point somewhere behind the HUD plate

* On military systems, HUD symbology is aligned with viewpoint (which can be roughly considered as "infinity"? )

Is this somewhat right? :)

Now, question is :

With the patch the entire HUD moves so its always aligned on the impact point.

then

Now in most military HUDs, not all but certainly the majority, this effect is designed out to make sure the pilot can either shoot or drop weapons perfectly on target

Why is BI wrong in showing the impact point on the HUD, thus making sure the pilot can shoot perfectly on target? Is it because the effect should show the HUD symbology going out of the HUD faster than simply following the head movement? (ie move your head to the right 3 cm means the symbology moves 15cm to the right on the HUD)

Share this post


Link to post
Share on other sites
Not quite sure, what I get is this :

* On single lens systems, HUD symbology is always aligned with a fixed point somewhere behind the HUD plate

* On military systems, HUD symbology is aligned with viewpoint (which can be roughly considered as "infinity"? )

No, its actually the reverse.

You have the releationship of 3 points to consider:

  1. Player's viewpoint
  2. HUD location
  3. Target/Impact point

Due to parallax error when you turn your head the impact point will appear to drift. You can see it for your self. Pick a point out of your window at least 50m away. Turn your head side to side slowly, because you have binocular vision the "aim point" will appear to move slightly. Now find a small 3D object (lighter, box of matches etc), this will be our "HUD". Hold that up directly in your line of sight between you and the aimpoint. Now, keeping the object perfectly still, move your head the same way. Does the object always remain over the aimpoint?

No it doesn't.

However with BIS' patch it does. The patch always centres the HUD on the aimpoint as though its perfectly flat. This is wrong. With simple HUDs something like this does occur due to the lack of lens correction, but not to the extent BIS are representing with this patch. However 90%+ of Military aircraft do not use simple HUD projectors. They use a correcting multiple Lens assembly that corrects for this.

Repeat the earlier exercise with your aimpoint and object. Now imagine that object is actually a projected display as in our simple HUD. the furthest face of that object is where the projected information would appear to be displayed.

Turn your head slightly to the right, the "display" will appear to move to the left. If we add a correcting lens to the projector (as in a military spec HUD) and repeat the exercise the "display" will appear to move to the right instead.

The logical/apparent behaviour of the displayed info is reversed. This is due to the way light travels through the lens and is reflected on the combiner plate in the HUD. This correction is necesary because of parallax error. This allows pilots to reliably align the aircraft on the impact point every time.

This is where you have to know a bit about optics and lenses to understand properly. I appreciate that this is counter intuitive but this is really how it works. Its pretty simple physics (even if it does take a bit of mind bending). But you can see it working in the Harrier video very clearly.

Is this somewhat right?

Now, question is :

then

Why is BI wrong in showing the impact point on the HUD, thus making sure the pilot can shoot perfectly on target? Is it because the effect should show the HUD symbology going out of the HUD faster than simply following the head movement? (ie move your head to the right 3 cm means the symbology moves 15cm to the right on the HUD)

Have you actually tried the patch?

Why is it wrong? Well they just centred the entire HUD on the impact point and move it with your point of view. To use our 'aimpoint and object' experiment, it would be the same as you moving the object each time you move your head so that you only ever see the front face.

Now considering that you cant actually move the barrel of the gun, but you need to aim down the centre line of the aircraft. And due to parallax errors, when you move your head around the cockpit this uncorrected view point can give you a false impression of where the rounds would land.

Using a "corrected" sight gives the pilot a more accurate visual queue about his alignment.

Purely in terms of the game what needs to happen is a change in the way HUD moves. If we're going to keep the moving HUD it needs to have the directions reversed and reduce the scale of movement.

Edited by RKSL-Rock
typos, formatting and a bit of clarification.

Share this post


Link to post
Share on other sites

Got it.

I didn't tried the patch, btw, will do. In the patch, does the aiming reticle have the exact same behavior as single lens HUD (ie, reticle moving left of aiming point when head moving right), or does it keep exactly on target?

Share this post


Link to post
Share on other sites
Got it.

I didn't tried the patch, btw, will do. In the patch, does the aiming reticle have the exact same behavior as single lens HUD (ie, reticle moving left of aiming point when head moving right), or does it keep exactly on target?

No, it stays exactly on target which is also pretty unrealistic. Even on a single lens uncorrected HUD there will be some error.

In my opinion the camera movement under G is too dramatic which is making it worse. But try it for yourself and see what you think.

Edited by RKSL-Rock

Share this post


Link to post
Share on other sites

If I had 6DOF in the cockpit , I could really test it out... ;)

Share this post


Link to post
Share on other sites
If I had 6DOF in the cockpit , I could really test it out... ;)

That's actually an interesting idea. Since we now have simulated g-forces in cockpits, moving the viewpoint around through what I presume to be inverse kinematics, it would be cool if we could produce that effect voluntarily using TrackIR. This could then be used to counteract the shifting viewpoint, for example if the g-forces are pushing you to the right, you lean left "into the turn" to counteract it and keep the view centered...

Sorry for the OT, I just felt a wave of inspiration and needed to vent it. :)

Share this post


Link to post
Share on other sites
That's actually an interesting idea. Since we now have simulated g-forces in cockpits, moving the viewpoint around through what I presume to be inverse kinematics, it would be cool if we could produce that effect voluntarily using TrackIR. This could then be used to counteract the shifting viewpoint, for example if the g-forces are pushing you to the right, you lean left "into the turn" to counteract it and keep the view centered...

Sorry for the OT, I just felt a wave of inspiration and needed to vent it. :)

More importantly (IMO ;)) it would allow heli pilots to lean around their pits looking down & around various obstacles in the cockpit when landing at difficult areas. Currently to simulate being able to do this I have to switch to external view on the last seconds of landing, when really I should be able to lean around to see around struts & panels etc.

Share this post


Link to post
Share on other sites

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
Sign in to follow this  

×