Prepp3d Views: Peripheral Distortion Correction

[GSalden 4,700]

Hi guys,

 

I have had some fellow Fs flyers asking me how both Denalis fix and Nikolas LCD Desghner Pro + Immersive Display Pro looked when using it.

 

Denalis fix ( freeware )

It is based on using NVSurround to have the best framerate. One can use 3 seperate views, but that always costs performance

His current beta uses Reshade too and he is still developing it

 

Pros :

- Looks very good and sharp

- Easy setup

Cons :

- Reshade causes some jittering when moving fast ( jetflying ). No issue with GA like a Cessna.

- in my opinion the zoom value might be lower ( but that increaes the fOV and is perhaps less realistic )

 

 

 

 

 

Nikolas fix with LCD Desgner Pro + Immesive Desktop Pro ( payware )

 

Also based un using NVSurround, but it uses 3 seperate views withing the NVSurround config that will be corrected.

As I now use no longer NVSurround, because I use different size of monitors ( 30"- 32"- 32") I contacted Nikola and he immediately offered help thropugh teamviewer.

 

It was impossible to have the larger 40" alligned with the 2x 32"monitors. The zoom value seemd to big.

Finally using the size of a 45"+ 13% for my left monitor in LCD Desginer Pro all views nicely matched up when using Immersive Desktop Pro.

However my side monitors have stretched sides the further to the outside like when using NVSurround before.. And I am not using NVSurround.

 

 

​

 

 

Pros:

-Very professional programs

-The programs keep evolving quickly

Cons :

- At this moment you need to copy camera settings into a saved flight

- Payware ( not that expensive compared to other corretion programs )

 

 

Nikola thought that something in my Windows 10 config or P3D config might cauase these stertched results.

Perhaps that is thue. But I have never had any CTD's or other Windows errors...

 

 

 

But 3 seperate views, as I am using now look good and also Denalis fix is working...

 

 

3 seperate views without any correction software

[denali 74]

Congratulations Nicola!  You've finally made your 10th post here at Avsim!  And you've managed to lecture somebody who has been using this thread of this simulator since the 1980's, and multi-monitored in the mid 90's!  Good Job!

 

I also had my second year of Calculus in the early 90's.  Let me give you a hint.   Account for FOV when you use the shader, by using the software I provided with it.  Then tell me about how nice and round the circles are.  If you are using lots and lots of circles your solution is not very elegant.  You actually should be using triangles Nicola.  

 

 

Regards

You know, I tried and I tried with that cylinder thing you were talking about, and I just couldn't make one.  This is all I could come up with.  

 

 

Meh!   Forgive me.  I'm just gonna go lift weights.

 

My name is John.  It's in the shader file, the one you say you have.

 

 

In the next version of LCD Designer I will add this "spheres" scene as a standard scene in the software to evaluate the distortion.

 I used P3D.  There is this thing called "mouse look" in it.  You can put something right dead smack in the middle, like another airplane nearby, and then look to the side so that the plane is dead smack on the edge of one of your side monitors.  It doesn't change it's shape in any way.  I think there were tires on that plane.  They look kinda like circles, but possibly they are oblique in some way because I may not be straight on.  Is that what you mean by my distortion fix doesn't work?

[Mike_CFII_MEL 790]

Hi,

 

Does the LCD Panel software require the purchase of any additional software in order to run?. If so, total cost?. I'm running three 75" 4ks, will screen size and/or resolution be an issue... Right now I'm running NVSuround at 11520x2160.

 

 

 

 

Hi All,

 

No PhD required to make or use LCD Panel Designer software.

 

Just to let you knwo that we just released an new version of LCD Panel Designer.

Thanks to everyone who used LCD Panel Designer v1.0r1 and provided us with useful feedback.

We integrated some of the feedback and suggestions in this new release.  

Give it a try.....

 

http://fly.elise-ng.net/index.php/news/78-news-immersivedisplaypro10r2

 

 

Regards,

Nikola

[GSalden 4,700]

Mike,

 

LCD Designer Pro needs Immersive Display Pro, which costs € 150.

[denali 74]

lol, was I getting mad.  I better quit doing steroids.

GSalden, go ahead and share everything if you want with Mike.  I don't think I've made that many changes since you last had an update.  His setup is a twin of yours.  Well, a big screened twin.

[dukeav 46]

lol, was I getting mad.  I better quit doing steroids.

GSalden, go ahead and share everything if you want with Mike.  I don't think I've made that many changes since you last had an update.  His setup is a twin of yours.  Well, a big screened twin.

 

Can you check your pm?

[GSalden 4,700]

Mike, if you contact me by pm I will sent you tne fix with my settings so you can start right away...

[ngidalov 0]

Congratulations Nicola!  You've finally made your 10th post here at Avsim!  And you've managed to lecture somebody who has been using this thread of this simulator since the 1980's, and multi-monitored in the mid 90's!  Good Job!

 

I also had my second year of Calculus in the early 90's.  Let me give you a hint.   Account for FOV when you use the shader, by using the software I provided with it.  Then tell me about how nice and round the circles are.  If you are using lots and lots of circles your solution is not very elegant.  You actually should be using triangles Nicola.  

 

 

You know, I tried and I tried with that cylinder thing you were talking about, and I just couldn't make one.  This is all I could come up with.  

 

 

Meh!   Forgive me.  I'm just gonna go lift weights.

 

My name is John.  It's in the shader file, the one you say you have.

 

 

 I used P3D.  There is this thing called "mouse look" in it.  You can put something right dead smack in the middle, like another airplane nearby, and then look to the side so that the plane is dead smack on the edge of one of your side monitors.  It doesn't change it's shape in any way.  I think there were tires on that plane.  They look kinda like circles, but possibly they are oblique in some way because I may not be straight on.  Is that what you mean by my distortion fix doesn't work?

 

 

 

TO: John (denali)

 

Here is how the 100% geometric correction works:

 

- Imagine you have 3 LCD panels in front of you. For this post I just use 3 LCDs put in a configuration to cover 180 deg horizontal field of view from the user eye-point.

Note: the left and right LCD panels are rotated so that when the user turns his head about 90 deg left it sees the left edge of the left LCD (exactly-90 deg left). The same when hie turns 90 right. This is what the pilots do when they fly traffic patterns.

 

- Now imagine the the 3 LCD panels are your "windows" to the real outside world. You can only see your real outside world from the eyepoint through those "windows". This also means that if part of the real world is not visible through the windows,  it can never be shown on the windows. Simple right?

 

Ex: Imagine that you have a 1m x 1m rectangular window at distance 1m in front of you (eyepoint),  you can only see the objects in the outside view which are on 45 deg (left, right up, bottom) from your eye point.

So again, 100% geometrical correction makes sure that when you have such 1m x1m LCD display at distance 1m in front of you, you will not be able to observer the outside world beyond 90 deg field of view.

 

- Now, to prove that LCD Designer PRO does exactly this and ensures 100% geometrical correction here is what I did:

 

Imagine the outside world is a simple scene created in a shape of a cylinder with floating spheres. A "stonehenge" like structure if you like :-).

The eyepoint is in the middle.

 

 

So, now let's put our LCD displays in a 180 deg horizontal FOV configuration. By this out LCD displays are the views to the outside world and we can only observer the outside world through those LCD display "windows".

 

 

 

When you position the person in the eyepoint and let the person look just straight ahead at the outside world without the LCD panels this is whey the person will see. 

 

 

 

So now we position the LCD panels, and on the LCD panels we show the images which are generated with the exported FOV for each LCD panel and additionally warped by Immersive Display PRO. What you see is that the warped image on the LCD panel exactly resembles what the users would see from the outside world when the LCD panel would be his window to the outside world.

 

 

 

Let's do the same for the left and right view. Just rotate the persons head to the left to be able to see up to around 90 deg left. This is what the user would be able to see from the outside world. 

 

 

 

When we turn the LCDs with the warped images person sees the exact outside world on the LCD panels.

 

 

For completeness, turn the user head to the right to see the right 90 deg of the world and this is what he sees:

 

 

Now again with the warped LCD panels.

 

 

Important note: Do not mind the stretched spheres on those screenshots. This is because I'm using computer graphics eyepoint to prove this concept and the computer graphics perspective projection tends to stretch images. That's why we need geometry correction after all.

 

Just to show you how the image will look on the most right LCD panel

 

This is how the image look when it is not corrected (warped):

 

 

 

,and here it is with geometry correction (warping)

 

 

for 100% correct geometry when observer from the eypoint and person turning his head 90 deg right.

 

You can say "But it is stretched.... And I say: yes it is stretched but when you put this image on the LCD panel and observed it from the the eypoint here it how it will look like: No distortion and geometrically correct.

 

 

The images will only look corrected from the eyepint location.

 

So John, please feel free to apply your lens shader and create this 180 deg geometrically correct outside views using just the same or similar scene. I would be very interested to learn from your research. 

 

/////////////////////////////

To ALL: Yes, it does not feel very intuitive, and a lot of people have issues grasping this, but that's simply how it works.

 

This software is for those who really do care about their correct geometrical views. For those who do not care and those who are satisfied with less, fortunately there are alternatives and John can provide you one. 

///////////////////////////

 

 

Nikola

[denali 74]

and additionally warped by Immersive Display PRO. What you see is that the warped image on the LCD panel exactly resembles what the users would see from the outside world

 This is what you are not explaining.  The views are the easy part.  If you are doing any kind of warping to fix the problem, it MUST use the math that I previous published.  

 

Everything else you are writing here is smoke and mirrors.  Perhaps to cover up what I am trying to get you to be honest about.  It either doesn't actually use mathematical precision, or you are using the math I have published.  

 

If you don't understand the significance of this one image, that every triangle you need to measure is represented, you're just blowing smoke balls.

 

 

That is the math you are attempting to obscure, to not discuss, either because of a lack of awareness, or an unwillingness to acknowledge that you are not the first.

Actually, I still think your balls are stretched from what I can see here.

I'm actually being a nice guy Nikola.  You might not be able to see that in your "point of view", but in the sphere of the 3d space we've been discussing I'm being very generous.

[Slayer 190]

I  am very intrigued by this discussion, albeit it's way over my head but I imagine that this is what happens when two astrophysics PHD's go at each other in the backrooms MIT etc. The info you guys are putting out is very enlightening.

 

This stretching has always been a pet peeve of mine in these sims that no one seems able to finally solve until now. I just can't figure why it has to be addressed by a 3rd party when first MS and now LM are at the helm. I'm sure they have a hell of a lot of people that could have made it right in the first place.

 

Edit: Any game or application that uses a first person perspective could benefit from this type of correction, 99% of them are wrong...

[GSalden 4,700]

I simply look at it this way :

Go stand away 100 ft from a car in the way that it is directly in front of you.
Now make a 90 degree turn to the right.
Turn your head 90 degree to the left and look at the car.

The car has the same size...
Because it is still 100 ft away...

If a pilot looks out of the window in front of him and sees an object 20 mls away it won't look zoomed in if he then makes a turn and looks at it from the side windows.
If it would look more zoomed in to the outer sides it would mean that by turning the outside world would come closer to you ( read : it would bent towards you )...

One doesn't need math for understand this.

 

 

Looking at Nikola's screenshots : from the 4th on it is going wrong. The balls maintain the same distance to the eyepoint but the outer balls are getting larger because the screen is flat.

But it is all about the distance from the eyepoint, which remains the same. Not by taking the flat screens as distance measurements.

Therefore  one will end up with a wrong correction.

 

---------------------------

The reason you get the distortion with NVSurround is that the projection being used is a planar one and the FOV of the view doesn't match the FOV of the monitors.

You can see the same type of effect on some games that use 90 degree plus FOV's on a single monitor to make it feel like you're moving faster.
Even with FSX/P3D you can create a fisheye effect on 1 monitor by zooming out till 0.30.

With FSX/P3D a zoom of 1.00 on a 3:4 monitor makes a correct canera view without distortion.
Using a 16:9 monitor you can go as low as about 0.75.

With NVSurround we are trying to have a wider view than the camera can show.
Therefore the sides gets stretched . To fill up the space.

A little can be compensated by using Wideview Aspect is true, but the issue remains the same.
Distortion when the monitors FOV is larger than the cameras FOV.

 

 

 

Solution 1 : use extra views. Three 16:9 views with a zoom of 0.75 or higher will look good.
P3Dv3.2.supports multimonitor setups better than earlier versions. Two extra undocked views now show the same performance than one before.

Solution 2 : try Denali's fix icw NVSurround. Using 1 view has better fps than 3.

[ngidalov 0]

 This is what you are not explaining.  The views are the easy part.  If you are doing any kind of warping to fix the problem, it MUST use the math that I previous published.  

 

Everything else you are writing here is smoke and mirrors.  Perhaps to cover up what I am trying to get you to be honest about.  It either doesn't actually use mathematical precision, or you are using the math I have published.  

 

If you don't understand the significance of this one image, that every triangle you need to measure is represented, you're just blowing smoke balls.

 

 

That is the math you are attempting to obscure, to not discuss, either because of a lack of awareness, or an unwillingness to acknowledge that you are not the first.

Actually, I still think your balls are stretched from what I can see here.

I'm actually being a nice guy Nikola.  You might not be able to see that in your "point of view", but in the sphere of the 3d space we've been discussing I'm being very generous.

 

To: John,

 

I just need to repeat the experiment and the proof I provide in my post, but this time with your lens shader. When you have the LCD panels in one plane you can simply adjust the distortion using your algorithm. But this is not what almost all is using. People are using rotated (turned) LCD panels. 

So, go ahead and provide a proof that with your shader produces correct geometry when using 45 deg left and right panels FOV with 180 deg.  Just use the same or similar line of reasoning about the real outside world. This is the only way you will be convinced if your algorithm can be used for such a setup or not. Just provide some solid evidence as i did in my post.

 

And again: STOP crying like a small child. NOBODY stole your algorithm. Those trigonometry rules have been discovered way way before me and I ware born. 

 

The way I use the trigonometry for geometry correction is different that the way you use in your shader.

So, gain.... Go ahead,  "go lift weights, eat steroids or whatever you need to do", and come back with the solid analysis and visual proof of 180 deg setup.  I would be very interested to look at them and provide you some feedback. 

 

I already asked you to contact me directly on this subject and not to interfere on this forum thread. Unfortunately you continued here. Your choice. 

Now I'm really stopping. I will restrain replying to your comments any more on this thread. I already made my point and provided evidence. 

 

To ALL: Sorry for this interference. A "disagreement" on any subject is actually a very good think and a solid discussion helps very much in improving the insights and eventually improving the end-products. But this guy simply jumped in as a "crying child" and in his first sentence already used the words "I discovered the algorithm", "you steel the idea from me", etc. He is simply WRONG. I had to provide a solid proof to convince him. Now it is his turn....

 

Unfortunately it took some post to reach this state, and I promises no more posts of such a nature from my side any more.

Feel free to contact us at http://fly.elise-ng.net/ or support@elise-ng.net. We are continuously improving our solutions so together with your input and suggestions we will provide even better next version.

 

Happy flying :-), 

Nikola

[GSalden 4,700]

Just made 2 pics from the driver seat of my car with the seat just above a marked point on the pavement.

 

Front window :

 

 

Left front window with the card turned 45 degrees :

 

 

The doors have the same size when lookjng out of the front window then when looking out of the left front window...  

 

 

When turning the aircraft with Nikola's correction the more to the side to more zoomed in... 

When using Calculate All in LCD Designer Pro it looks good on the animated displays, but after copying the eyepoints into the Saved fligh in P3Dt and using the calibration files in Immersive Display Pro it doesn't look the same anymore in P3D.

I have posted pics from my end result which shows stretched sides.

 

And because the view is dsiplayed on a flat monitor , moving your eyepoint will NOT correcdt the streched distortion...

The image simply looks the same.

 

Immersive Display Pro does a great job on curved projection screens. No doubt about it.

LCD Designer Pro icw Immersive Display Pro has a false outcome on flat displays.

 

It would be nice if Nikola would dig into that... and was able to solve that. 

[dukeav 46]

Hi Denali,

 

I had one question about your solution. 

 

You solution does not seem to (does it?) to take into account the angle at which the side screens are placed.

 

1. Imagine side screens at 45deg off the center screen. 

2. Using Prepp3D views the geometry appears perfect. So there are three circles on each screen and they appear as perfect circles to the user in front of screens after Prepp3D correction.

3. Pause the sim. (Just to convey that the sim is rendering the same static 3 circles view)

4. Now the user physically tilts the side screens to a 60deg angle.

 

What would the circles on the side screen look like? They will be horizontally squished, optically to the user.

 

So no single correction can account for all angles? Please comment.

 

Or is your solution for a fixed angle?

 

Thanks

 

[denali 74]

My solution is not completed. This is a hobby for me, not a commercial venture at this time. One of the reasons why I've kept the current work out of view is because I don't wish to have to deal with the support issues. But I realize there is growing demand for the solution, and a group working together around the idea can be helpful.

Right now there is a predefined, but changeable, FOV in my solution. You set the FOV for the FOV your screens span. Typically your screens are tilted in so that the viewpoint for each screen is in the center of that screen. There is no need to account for angle if this is the case. Angles off the center viewpoint can be accounted for, but I have not done this as demand for that, that the lack of adjustment is a noticeable detraction from an effective solution, will probably not exist. Vision is based on many illusions we just ignore.

Because the shader and the program both need to be coordinated somewhat in the width and height, as defined by FOV, I do not have them programatically tied in together at this time. So a user has to define the height and width of the oversize to their purposes. If height or width are off there will be a little compression or stretch in either direction, but across the whole display evenly, and just changing the height or width of the oversize corrects this.

The portion I am working on now is to remove the need for the over-sizing software, which Is the source of a support nightmare because it is confusing. Instead of doing that math, explaining a compiler the correct mathematical model for the final portion of the solution, I find myself re-confirming items with insurance adjusters and contractors (we were affected by Houston's recent flooding) and dealing with other personal issues, one of which is recovering from previous years levels of extreme stress (the sources are gone, but the anxiety it is lingering with me for some reason, the flooding did not help).  So that's the domain of the math involved in finally finishing this.

So yes, the single correction I have created accounts for all angles if the screens are centered around a general viewpoint. It is possible that a screen could be not on center, and the math is available to make that adjustment, but to begin it's a good idea that most people would like to be in the center viewpoint of their screens to see the maximum of the display to be presented. The correction approaches 100%, IS 100% in the sense than an object will maintain it's pixel size across all areas of all displays; I say approaches because we could argue that the obliqueness to the center viewpoint of the very edges of any display needs to be accounted for to be 100% to the viewpoint, and perhaps this is the case for an extremely large display, and possibly what Nikola is arguing. Although others that have tested it do not see Nikola's solution as solving the stretch problem, and his solution is already available, although tricky to implement, with P3D's multiple views.

Nikola said once that it is impossible to be 100%, then he went back on it to claim that his does so. All things are possible with math if you can measure it. But because perception is not perfect perfect math isn't always necessary, or needed.

 

dukeav, on 19 May 2016 - 3:34 PM, said:
Or is your solution for a fixed angle?