[REPORTED] Model Visibility Issues

[probad]

Even with that mod it is very boring to fly WW2 in multiplayer simply because you might get 5 minutes of combat for an hour of seat time.

nothing wrong with this

 

however the system definitely requires some more fundamental changes than what we have going. im just waiting to see what 2.5 brings before trying to figure out what would constitute valid suggestions.

[chihirobelmo]

Hi,

 

I finally set up an equation for distant plane magnification which will provide RL 20/20 vision while using 60 or more hFOV regardless of using any resolution of the monitor.

 

Instead of thinking about the angular mil and monitor size, head position relativity like I did before in former discussions, this time I simply consider the pixel number of aircraft in DCS and much it to RL 20/20 vision.

 

First, I have to think about Vision Test. When I look at 7.272mm of Landolt C from a 5m distance and if then determine which direction does ring have 1.454mm of the hole, It means I have 20/20 eyesight. 1.454mm against 5m means 1 minutes of degree = 1/60 degree.

 

 

Calculating 3.0nm distance of line abreast F-15 its seen

 

19.43m(Length of F-15) / 5.56km(3.0nm) = 3.50mil(2π/6400 rad)

3.50mil * 3.375 = 11.8 minutes of degree (as 1 degree is π/180 rad)

 

Let's assume that 1 minutes of degree for 20/20 vision as 1 pixel of a monitor display.

 

Then I calculated How much pixel size would 3.0nm distance F-15 line abreast be seen in DCS with 60hFOV/FHD resolution.

 

X = expected pixel size in DCS;

 

60(deg)*60=3600(minutes)

1920:X=3600:11.8

X=6.29(px)

 

This way I can calculate expected pixel length of aircraft in each FOV/Resolution at DCS.

 

 

The pink curve shows aircraft length size in pixels in DCS, dark blue curve shows expected RL aircraft size at each distance in minutes of the degree.

 

I also took some screenshots and compared the calculation and the result.

 

 

I started DCS1.5 FC3 instant action F-15C intercept, press F2 then backward the camera while seeing F-15 from its side.

 

 

1.0nm-17px

2.0nm-9px

3.0nm-6px

 

It almost matches my calculation!!!!!!!!!!!!!!!!!!!

 

Now then, I want DCS to draw close to 12 pixels of line abreast F-15 3.0nm away regardless of any wider FOV/Resolution setting. So I will write here magnification equation which adjusts aircraft length size in pixel size related with Distance/FOV/Resolution.

 

M=Magnification Factor;

D=Distance(ft)/1000;

R=Resolution Length(ex:FHD will have 1920);

F=Horizontal FOV;

 

M=1+(0.09226*D-0.00148*D^2)*{1.22e-7*R^2-1.17e-3*R+2.8}*(0.02*F-0.4)

 

Original equation comes from Serforss 2003

http://www.dtic.mil/docs/citations/ADA414893

 

M=1+0.09226*D-0.00148*D^2

 

I added Resolution/FOV factor here so that magnification factor matches any gaming environment nowadays, from Half-HD to 4k monitor users will see realistic pixel size aircraft when in 60 or more hFOV.

 

 

The light blue curve shows magnification factor and Orange curve shows aircraft size in pixel after applying magnification.

 

Between 2.5nm-8.0nm magnified plane seems larger than RL but it is only 1-2 pixels of difference.

Now plane in the simulator can be seen almost as same Fineness as what can be seen in RL 20/20 vision.

 

As I also added FOV/Resolution relative equation upper than 60FOV, between Half-HD to 5k resolution monitor users will have same visual aid!

This is what even FalconBMS didn't achieve with their smart scaling!

 

 

 

Here is result for 1920*1080 FOV140

 

 

I also adjusted the equation not to enlarge model while zooming out, there would be no graphical anomaly.

 

 

 

I made magnification factor comes close to zero when zooming in and at FOV20 it will be almost zero magnification.

 

 

 

 

Result for 1280*720 FOV60

 

 

 

 

WQHD users will have bit finer vision but it's still no bigger than 1-2 pixels compared to RL vision.

 

 

 

 

I recommend using FOV70 or 80.

 

 

 

 

4K users will have small magnification. You guys don't need match scaling...

4K with FOV60

 

 

 

And 5K users will have almost zero magnification.

Above than 5k factor will increase again so the magnification must be off at such a resolution setting.

 

 

I will share my excel file so download it and try inputting any settings and you will see the result.

https://drive.google.com/file/d/0B9l8t2itS4vkQWNpT0xqbGNzWkk/view?usp=sharing

 

 

 

With this method

 

1.Plane size continues to decrease by distance so pilots will not lose their sense of closing/leaving of opponent aircraft and distance relation to visual plane size.

 

2.As an original 3D model is to be used, visibility will have relation to each plane wingspan/length/height so the increase/decrease visibility made by opponent orientation will be more realistic.

 

3.As an original 3D model is to be used, each planes coloring will affect its visibility. low vis coloring will work like in RL.

 

4.As it is effective in WVR range (1.0nm-3.0nm) BFM/Formation experience will be a much more realistic thing.

 

5.There will be no visual anomaly while model transiting to imposter/dot (becomes visible-unvisible-visible while zooming in/out)

 

6.Helps VR users who can't zoom in.

 

 

I really hope ED to implement something effective and reliable visibility aid to DCS2.5

This is one of suggestion done by me who is only an amateur...I am waiting for future professional work done by ED.

Edited January 24, 2017 by chihirobelmo

[Gliptal]

You must spread some Reputation around before giving it to chihirobelmo again.

:cry:

[TwojaStara]

Can ED comment on the visibility problems ? It appears to a lot more users that game is hardly playable:

[exray]

When I recently started playing this game, I thought the visibility problems were just because I was new to the game and was having trouble flying and panning the view with kb or hat switch.

 

So I went and got a TrackIR and set-up some zoomed in snap view for strafing runs and what have you.

 

It is still really really really hard to eyeball anything consistently at 1440p.

 

Even when knowing exactly where they are supposed to be, unless anything in the air is throwing a contrail or engine smoke, they might as well be invisible.

 

Yesterday I missed a guns pass in the A10C, on a jeep in an open green field with the view zoomed all the way in, with SPI set on target, because I just couldn't see the thing at less than 2nm.

 

This is compounded by the promised-but-yet-to-be-delivered improvements to the FLIR system.

 

Please fix the visibility problems in DCS.

[tintifaxl]

An easy fix in single player: use customized labels.

[shagrat]

Unfortunately, we still have some weird LOD modeling in place with some ground units.

That means at a certain distance the previously perfectly visible small dots for infantry, simply disappear and reappear when getting a couple meters closer...

This is very obvious at a 1440p or 4k resolution. :cry:

 

I started to mod the LOD files for some infantry and small vehicles a while ago.

 

It works, but does break integrity check in Multiplayer, plus it needs to be done for virtually all models.

As we have the DCS 2.5 merge getting closer, I hope we see this fixed by ED, sooner or later. :worthy:

 

When I recently started playing this game, I thought the visibility problems were just because I was new to the game and was having trouble flying and panning the view with kb or hat switch.

 

So I went and got a TrackIR and set-up some zoomed in snap view for strafing runs and what have you.

 

It is still really really really hard to eyeball anything consistently at 1440p.

 

Even when knowing exactly where they are supposed to be, unless anything in the air is throwing a contrail or engine smoke, they might as well be invisible.

 

Yesterday I missed a guns pass in the A10C, on a jeep in an open green field with the view zoomed all the way in, with SPI set on target, because I just couldn't see the thing at less than 2nm.

 

This is compounded by the promised-but-yet-to-be-delivered improvements to the FLIR system.

 

Please fix the visibility problems in DCS.

[Reflected]

An easy fix in single player: use customized labels.

 

Yepp, the one used on the Burning Skies server is actually very realistic.

 

I realized that if you zoom all the way in, there is no issue with visibility. The problem is that you lose all peripheral vision and SA. If you zoom out, you lose the dot. The ideal solution would be to make sure those dots are just as visible zoomed out as when you zoom in, that's all. (which is what the label mod does)

[SageOT]

Hi,

 

I finally set up an equation for distant plane magnification... (snip).

 

Umm... maybe I missed it... but what do I change and where do I change it to try this? Rift user...

[klem]

Hi,

 

I finally set up an equation for distant plane magnification which will provide RL 20/20 vision while using 60 or more hFOV regardless of using any resolution of the monitor....... snip

 

An interesting piece. If anyone isn't aware, Serfoss proposes this magnification method for professional military flight simulators and proposes magnification factors. Back in the late 80's I was involved in a military simulator programme that used a high definition "area of interest" projector to give high definition rendering in the area of the pilot's head-tracked view (think TrackIR but done differently), over about 15 degrees FOV, the rest of the background being low res. Writing in 2003 Serfoss was still saying that technique wasn't really practical and says why. However it gives me an idea.

 

Against 'increasing the target size' is the argument that the aircraft will be seen 'too soon'. Well, supposing there was a key that would magnify the centre 10 degree FOV of the monitor by the factors chihirobelmo proposes. So, first you must be aware that there is a target, then look at (centre) your target and press the key for more (magnified) detail without losing overall SA as we do with a full screen zoom-in. I suspect this idea may briefly affect frame rates (a bit like the mirror does) but it might be worth considering.

[chihirobelmo]

Umm... maybe I missed it... but what do I change and where do I change it to try this? Rift user...

 

Sorry, I only wrote an idea. There is no mod to implement this magnification to DCS.

 

 

 

An interesting piece. If anyone isn't aware, Serfoss proposes this magnification method for professional military flight simulators and proposes magnification factors. Back in the late 80's I was involved in a military simulator programme that used a high definition "area of interest" projector to give high definition rendering in the area of the pilot's head-tracked view (think TrackIR but done differently), over about 15 degrees FOV, the rest of the background being low res. Writing in 2003 Serfoss was still saying that technique wasn't really practical and says why. However it gives me an idea.

 

Against 'increasing the target size' is the argument that the aircraft will be seen 'too soon'. Well, supposing there was a key that would magnify the centre 10 degree FOV of the monitor by the factors chihirobelmo proposes. So, first you must be aware that there is a target, then look at (centre) your target and press the key for more (magnified) detail without losing overall SA as we do with a full screen zoom-in. I suspect this idea may briefly affect frame rates (a bit like the mirror does) but it might be worth considering.

 

 

Thanks for posting your interesting experience for military simulator and relation to the paper I referred:)

 

I made my calculation because I thought "finding a target at first" is already the problem for DCS user who plays the sim with a single desktop monitor.

Therefore suggesting to be aware that there is a target, then use "magnification toggle" doesn't help for such a people.

 

I think the aircraft was to be seen 'too soon' because head-tracked view system you mentioned gives pilots 1:1 view against RL(In other words the world is about the 'correct size'). As Serfoss 2003 magnifies aircraft size to fit the plane fineness(pixel size) to RL, it also magnified plane size at the same time, so it became much larger than in RL in terms of angular mil.

 

However, for most of the desktop simulators, we display 70FOV view to the monitor which fills only 35FOV(numbers are just for instance). Which Displays about 1/2 view against RL. I did some calculation/measurement and found when we see the monitor to be viewed 35/42/50 degree angle in our sight, setting in-game FOV to 60/70/80 helps to fit the plane mil size to RL.

https://www.bmsforum.org/forum/showthread.php?28660-About-SmartScaling-and-future-high-res-monitors&p=400589&viewfull=1#post400589

So they will get both correct pixel size and mil size at the same time with Serfoss 2003 Magnification.

 

Current VR headgear users who don't have enough resolution to much 20/20 vision, will need magnification to get enough fineness(pixel size) but at the same time, they face the same problem(plane becomes larger in terms of mil) because they see 1:1 view.

 

Triple monitor users who setup with zoom to set the world is about the 'correct size', who sees 1:1 view, might also have enough resolution to get 20/20 vision in some cases. For instance, if you set each 1920*1080 monitor to 35FOV (105FOV in total) it will show you almost enough pixel size of F-15 compared to RL. Users with such a monitor setting already have a correct view size and correct pixel size at the same time so do not need any magnification factor.

 

[SageOT]

Sorry, I only wrote an idea. There is no mod to implement this magnification to DCS

 

You're killing me Smalls!

 

Amazing work man. Hopefully someone with the smarts can figure out what to mod to get your figures into the sim.

Sage

[shagrat]

You're killing me Smalls!

 

Amazing work man. Hopefully someone with the smarts can figure out what to mod to get your figures into the sim.

Sage

That would be ED Devs who need to adapt the rendering according to this calculations, I guess.

 

The concept sounds very sound and reasonable.

[SageOT]

So it's more of a core fix then. Got it. Well here's hoping his work can help them out.

[BadHabit]

For now servers should use the Label Mod that

Burning Skies Server is using, makes life easier and it is the most realistic setting out there.

[Why485]

Now then, I want DCS to draw close to 12 pixels of line abreast F-15 3.0nm away regardless of any wider FOV/Resolution setting. So I will write here magnification equation which adjusts aircraft length size in pixel size related with Distance/FOV/Resolution.

 

M=Magnification Factor;

D=Distance(ft)/1000;

R=Resolution Length(ex:FHD will have 1920);

F=Horizontal FOV;

 

M=1+(0.09226*D-0.00148*D^2)*{1.22e-7*R^2-1.17e-3*R+2.8}*(0.02*F-0.4)

 

I thought this was pretty interesting, so I decided to implemented it into a little demo scene. In the demo there's a few squadrons of planes flying around, and you can move the camera around to see what the scaling looks like. Labels and scaling can both be toggled. Field of view can be changed with the middle mouse.

 

 

If you want to mess with this yourself, here's a download link to the demo. Extract the contained folder anywhere and then run the executable inside.

 

Overall, I like it a lot. I especially like that it adjusts based on field of view so that when you zoom in on far things it doesn't look weird. I haven't thoroughly tested the resolution compensation, but I think that's a great idea as well.

 

Personally, I think the magnification is a little too strong (I feel the same way about a certain F-16 flight sim), but that's trivial to tweak. What's important is that the foundation is solid, and it most certainly is.

Edited January 26, 2017 by Why485

[chihirobelmo]

Thank you for this demonstration!

Your work specifically shows how my theory works in an actual 3D world.

 

Overall, I like it a lot. I especially like that it adjusts based on field of view so that when you zoom in on far things it doesn't look weird. I haven't thoroughly tested the resolution compensation, but I think that's a great idea as well.

 

Actually, 4k monitor users might not have enough FOV related magnification.

This is because my equation goes this way.

 

 

As F(Resolution) kills overall magnification factor at 4k res it also kills F(FOV).

I still could not find out the way to give 4k users enough FOV related magnification.

If I allow F(Resolution) curve more gentle to allow 4k users get enough magnification at 90-140 FOV, then WQHD users will have too large magnification, and 60-80 FOV at a 4k monitor would get the same problem. Perhaps I can include resolution factor to F(FOV) so that higher res monitor will get a more steep angle for F(FOV).

 

At least 4k users will have realistic pixel size around 70-80 FOV with the current equation. HD/Full HD/WQHD user will have proper magnification adjustment in any FOV settings.

Edited January 26, 2017 by chihirobelmo

[klem]

........snip

 

Thanks for posting your interesting experience for military simulator and relation to the paper I referred:)

 

I made my calculation because I thought "finding a target at first" is already the problem for DCS user who plays the sim with a single desktop monitor.

Therefore suggesting to be aware that there is a target, then use "magnification toggle" doesn't help for such a people.

 

I think the aircraft was to be seen 'too soon' because head-tracked view system you mentioned gives pilots 1:1 view against RL(In other words the world is about the 'correct size'). As Serfoss 2003 magnifies aircraft size to fit the plane fineness(pixel size) to RL, it also magnified plane size at the same time, so it became much larger than in RL in terms of angular mil.

 

However, for most of the desktop simulators, we display 70FOV view to the monitor which fills only 35FOV(numbers are just for instance). Which Displays about 1/2 view against RL. I did some calculation/measurement and found when we see the monitor to be viewed 35/42/50 degree angle in our sight, setting in-game FOV to 60/70/80 helps to fit the plane mil size to RL.

https://www.bmsforum.org/forum/showthread.php?28660-About-SmartScaling-and-future-high-res-monitors&p=400589&viewfull=1#post400589

So they will get both correct pixel size and mil size at the same time with Serfoss 2003 Magnification.

 

Current VR headgear users who don't have enough resolution to much 20/20 vision, will need magnification to get enough fineness(pixel size) but at the same time, they face the same problem(plane becomes larger in terms of mil) because they see 1:1 view.

 

Triple monitor users who setup with zoom to set the world is about the 'correct size', who sees 1:1 view, might also have enough resolution to get 20/20 vision in some cases. For instance, if you set each 1920*1080 monitor to 35FOV (105FOV in total) it will show you almost enough pixel size of F-15 compared to RL. Users with such a monitor setting already have a correct view size and correct pixel size at the same time so do not need any magnification factor.

 

 

Hi chihirobelmo

 

I don't know if the 'area of interest' projector I mentioned used 'Serfoss' magnification. It was certainly about reducing the load on the graphics system and provide hi-res detail only in the area being looked at.

 

Regarding 'seeing the target too soon', that is a separate consideration from 'magnifying to see the aircraft attitude/detail in more RL form' and you probably know there are arguments for making the dots easier or harder to see at various distances regardless of shape/aspect. In DCS there are the two issues of dots and models. Why485's demo shows your model theory very nicely but there is that point where the magnified distant model suddenly disappears altogether instead of phasing out to a faint dot. There's a 'gap'. Fix that problem and we're well on our way.

 

I couldn't open that link because I'm not a member but no problem.

 

I use triple 27" monitors, my eyes are about 70cm from the centre monitor and the RL FOV to the monitor edges is about 70 degrees. The three monitors reach out to the edge of my peripheral vision and with the a/c FOV set to 70-80 degrees (in SnapViews.lua) the cockpit view is very close to RL size. The only reason they differ between a/c is for fine tuning of some instrument positions. Even so, I find it hard to see distant dots and the models are OK within about half a mile (estimate) but shrink too quickly so even triple monitor setups could do with some magnification versus distance.

 

I have had second thoughts about an 'area of interest' magnification key because it would have to be where I am looking, i.e. using TrackIR co-ordinates, and I expect that would be too difficult to implement.

[Why485]

Regarding 'seeing the target too soon', that is a separate consideration from 'magnifying to see the aircraft attitude/detail in more RL form' and you probably know there are arguments for making the dots easier or harder to see at various distances regardless of shape/aspect. In DCS there are the two issues of dots and models. Why485's demo shows your model theory very nicely but there is that point where the magnified distant model suddenly disappears altogether instead of phasing out to a faint dot. There's a 'gap'. Fix that problem and we're well on our way.

 

Where do you see a gap?

 

Because the model gets magnified over a distance, by the very nature of that it smoothly disappears into the distance. It doesn't ever pop out of existence.

[Ramsay]

Why485's demo shows your model theory very nicely but there is that point where the magnified distant model suddenly disappears altogether instead of phasing out to a faint dot. There's a 'gap'. Fix that problem and we're well on our way.

 

Where do you see a gap?

 

Perhaps klem has mistaken the points in the video, where you turn the scaling on and off, for LOD changes and not tested the demo ? i.e. @ 0:12

 

 

Note: there's text at the bottom left - "Scaling: Off" when the distant (3 nm) planes become hard to see.

 

Being able to easily see/find planes would be a game changer (so subject to the usual criticism) and is something DCS needs if ED want their WWII modules to have even greater success.

 

Being able to flick switches, while unable spot targets doesn't keep servers full or encourage realistic tactics.

 

As ED's impostors didn't take resolution into account, I'm not particularly optimistic but can only hope ED are taking note of these suggestions.

Edited January 27, 2017 by Ramsay

[Why485]

Perhaps klem has mistaken the points in the video, where you turn the scaling on and off, for LOD changes and not tested the demo ? i.e. @ 0:12

 

 

Note: there's text at the bottom left - "Scaling: Off" when the distant (3 nm) planes become hard to see.

 

Being able to easily see/find planes would be a game changer (so subject to the usual criticism) and is something DCS needs if ED want their WWII modules to have even greater success.

 

Being able to flick switches, while unable spot targets doesn't keep servers full or encourage realistic tactics.

 

As ED's impostors didn't take resolution into account, I'm not particularly optimistic but can only hope ED are taking note of these suggestions.

 

We will see. If 1.5.6 is supposed to come out tomorrow, then there may be more changes to visibility. They've tweaked it in some way on every major 1.5.X release, so I'll definitely be running my usual battery of tests.

[klem]

Where do you see a gap?

 

Because the model gets magnified over a distance, by the very nature of that it smoothly disappears into the distance. It doesn't ever pop out of existence.

 

Perhaps klem has mistaken the points in the video, where you turn the scaling on and off, for LOD changes and not tested the demo ? i.e. @ 0:12

 

 

Note: there's text at the bottom left - "Scaling: Off" when the distant (3 nm) planes become hard to see.

 

Being able to easily see/find planes would be a game changer (so subject to the usual criticism) and is something DCS needs if ED want their WWII modules to have even greater success.

 

Being able to flick switches, while unable spot targets doesn't keep servers full or encourage realistic tactics.

 

As ED's impostors didn't take resolution into account, I'm not particularly optimistic but can only hope ED are taking note of these suggestions.

 

I did watch for that, perhaps I missed it, I'll look again. But that is what I meant by the gap, the models suddenly disappeared. I'll watch it again later (I'm on my tablet atm, another $£&*@ sleepless night!)

[chihirobelmo]

Hi chihirobelmo

 

I don't know if the 'area of interest' projector I mentioned used 'Serfoss' magnification. It was certainly about reducing the load on the graphics system and provide hi-res detail only in the area being looked at.

 

Regarding 'seeing the target too soon', that is a separate consideration from 'magnifying to see the aircraft attitude/detail in more RL form'

 

Hi klem

Sorry, it seems I misread the meaning of your previous post.

 

and you probably know there are arguments for making the dots easier or harder to see at various distances regardless of shape/aspect. In DCS there are the two issues of dots and models. Why485's demo shows your model theory very nicely but there is that point where the magnified distant model suddenly disappears altogether instead of phasing out to a faint dot. There's a 'gap'. Fix that problem and we're well on our way.

 

On my rig magnification at why485's DEMO worked until model phasing out to a faint dot. If you could upload the video and show me what is happening on your rig I may can able to know how the "gap" is happening.

 

Or if you were referring to the Youtube video of the demo, it seems the video is showing older version of the demo which miss implemented my equation. It is calculating {1+F(Distant)}*F(Resolution)*F(FOV) instead of 1+F(Distant)*F(Resolution)*F(FOV)

 

His latest demo(v1.2) fixed the issue and has corrected the code, so try downloading the latest demo again.

 

I use triple 27" monitors, my eyes are about 70cm from the centre monitor and the RL FOV to the monitor edges is about 70 degrees. The three monitors reach out to the edge of my peripheral vision and with the a/c FOV set to 70-80 degrees (in SnapViews.lua) the cockpit view is very close to RL size. The only reason they differ between a/c is for fine tuning of some instrument positions. Even so, I find it hard to see distant dots and the models are OK within about half a mile (estimate) but shrink too quickly so even triple monitor setups could do with some magnification versus distance.

 

I have had second thoughts about an 'area of interest' magnification key because it would have to be where I am looking, i.e. using TrackIR co-ordinates, and I expect that would be too difficult to implement.

 

Does each of your monitors have 1920*1080 resolution? If so set to 32 horizontal FOV for a single monitor, in my calculation that would be a FOV setting to represent realistic pixel size to Full-HD monitors without magnification. (And then set your head to about 103cm if you would like to have 1:1 view size)

 

And yes even triple monitor setups could do with some magnification versus distance if users feel this "32hFOV for a single monitor" setting too narrow view and prefer to fly with "70hFOV for a single monitor". Then he needs magnification.

 

BTW viewing 27" inch monitor from 70cm only has RL 46 hFOV according to http://www.projectimmersion.com/fov/

 

Width of 27" display would be 60cm and 2*atan(30/70)=46.

Edited January 27, 2017 by chihirobelmo

[klem]

I did watch for that, perhaps I missed it, I'll look again. But that is what I meant by the gap, the models suddenly disappeared. I'll watch it again later (I'm on my tablet atm, another $£&*@ sleepless night!)

 

My apologies, the Off periods are quite short and I missed the on/off label changes. Looks very good after all.

[klem]

OK I have downloaded Why485's demo assuming it will allow me to switch the settings but I don't know how to use it. Is this run as a Mod under JSGME? Or if I run it does it permanently change my DCS files (integrity??).

 

Incidentally, regardless of the FOV theory my triple screen only looks right on 80FOV (I am using 3Screen). Anything else and I'm too far forward or back in the cockpit. I'm content with that and looking forward to trying out the demo.