[REPORTED] Model Visibility Issues

[gavagai]

DCS needs better rendering quality. Even within the specs of current monitors.

The Model Enlargement is essentially the same thing Falcon did and it doesn't help really.

 

When was the last time you tried smart scaling in Falcon, Sharpe?

 

Here is a demo of the smart scaling algorithm. In DCS you would see chunky transitions between the impostor and the normal 3d model, and at long range the impostors just appear to be sprites.

 

 

Smart scaling in Falcon works exactly as you see in the demo above. No weird, faded sprites at long distance. No chunky transitions.

 

Which is why nobody is suggesting a simple dot. There are realistic ways to do this, but nobody wants to even try it because somebody once did it poorly in the past.

 

I just realized that the above is your video!

 

I wouldn't say that Falcon "did it poorly." There are far worse solutions around...:music_whistling: Ultimately I think Sharpe is correct that the first order of business is just a change in the way DCS renders 3d models in the first place.

Edited September 17, 2016 by gavagai

[Why485]

When I say did it poorly, I was referring to people's issues with the current impostor system, and the ancient IL-2 solution of drawing 1-4 pixels when a model gets sufficiently small. Those systems are too basic, but with minor tweaks would work fairly well for solving the "long range" (i.e. ~10km, which is roughly the max distance for eyes on target) spotting.

 

I think DCS's biggest shortcoming is in the 1km - 4km range where planes are unusually difficult to spot. This is where the contrast comes in to play, and is something that DCS struggles with for whatever reason. DCS is getting a pretty big overhaul of its rendering in 2.5, so maybe, just maybe something about that renderer will help with spotting and identifying planes at short range, but as with everything ED/DCS, we won't know until it's out. ED doesn't talk about things.

 

Of course, one way to help mitigate these issues is to buy all DCS players 4k HDR monitors, but that's a ridiculous proposal, and it's only treating the symptom, not the problem.

 

As for Falcon's Smart Scaling, I think it's a simple and effective solution to the problem. Personally I think Falcon's exact formula is a little generous and I would tone it down slightly given the resolutions of that paper's day. However, the concept is sound and demonstrably effective.

Edited September 17, 2016 by Why485

[Schmidtfire]

+1

 

I agree that the biggest issue in DCS right now is the spotting between 1km-3km range. It doesn't take much to loose and opponent mid-dogfight, and once you've lost him, It's very difficult to re-acquire.

 

There is still a large issue with "invisibility" and that planes suddenly vanish depending on viewing angle and distance. Another example of this is engine smoke that sometimes are visible, sometimes not depending on zoom and angle.

 

In my opinion. With more WWII stuff coming up, I think this issue should be prioritized. This issue is not game-breaking but very serious and impacts gameplay. I would rank it more important than new grass or rain effects...

[Solty]

+1

 

I agree that the biggest issue in DCS right now is the spotting between 1km-3km range. It doesn't take much to loose and opponent mid-dogfight, and once you've lost him, It's very difficult to re-acquire.

 

There is still a large issue with "invisibility" and that planes suddenly vanish depending on viewing angle and distance. Another example of this is engine smoke that sometimes are visible, sometimes not depending on zoom and angle.

 

In my opinion. With more WWII stuff coming up, I think this issue should be prioritized. This issue is not game-breaking but very serious and impacts gameplay. I would rank it more important than new grass or rain effects...

 

I agree. I think imposter models should get a more pronounced contrast so that they are not easily lost over the ground. And the fact that airplanes just vanish when looked at from some angles is the main issue.

[Why485]

I agree. I think imposter models should get a more pronounced contrast so that they are not easily lost over the ground. And the fact that airplanes just vanish when looked at from some angles is the main issue.

 

There was another version of the proof of concept mod where I focused specifically on contrast for impostors. The problem with this is that it works great only when the game is using impostors. Once they get sufficiently close that the impostor no longer draws, you're back to something that's much harder to see even though it closer.

 

I think an ideal system would use something like impostors at long range (> ~5km), and then some kind of subtle (subtle!) visual scaling system at ranges closer than it, combined with a boost in contrast for the object.

 

The impostor system works very well when the target is far and its size can be measured in pixels. Not only is it efficient, but it's also very controllable, so you can tweak exactly when you want things to start being visible and by how much. This system doesn't work well at close range though because the closer it gets the more apparent that it's just a sprite becomes. This is why it goes back to rendering the original model.

 

As I've stated numerous times in the past, the impostor system as currently implemented in vanilla DCS is woefully underused and a potentially very powerful tool. It's a shame that its current implementation is so simplistic, because that's where the biggest problem of ridiculously long range spotting (the problem that GGTharos refers to often) comes from. That specific aspect is trivially easy to solve. The hard part in DCS is what to do with targets when they get closer.

 

Scaling based systems on the other hand, excel at close range where the problem you're dealing with becomes more about how far can you ID and track a target versus just knowing it's there. Scaling systems suffer at longer ranges though because it's difficult to tweak them to get something that works exactly as it needs to at long range.

 

Meanwhile, I would also add some kind of contrast when in the smart scaling ranges that gets slightly more pronounced the further it gets, but in 2.5 the raw contrast situation might be fixed for all we know.

 

I put together this little thing in paint to help visualize what I'm trying to say. Each system has strengths and weaknesses, and the best ones combine them.

 

Edited September 18, 2016 by Why485

[Ktulu2]

You haven't said what's on the y axis so i'm not too sure what this is. Is y the enlargement? The net pixel size?

[Why485]

You haven't said what's on the y axis so i'm not too sure what this is. Is y the enlargement? The net pixel size?

 

It's really three separate graphs combined together.

 

For contrast it's how much added contrast.

For scaling it's scale factor.

For impostors it's size and alpha.

 

I didn't put exact numbers because that's stuff that would be determined experimentally. What's important is the distances and the general shapes of the graphs.

[SharpeXB]

When was the last time you tried smart scaling in Falcon, Sharpe?

 

Here is a demo of the smart scaling algorithm. In DCS you would see chunky transitions between the impostor and the normal 3d model, and at long range the impostors just appear to be sprites.

 

 

Smart scaling in Falcon works exactly as you see in the demo above. No weird, faded sprites at long distance. No chunky transitions.

 

 

 

I just realized that the above is your video!

 

I wouldn't say that Falcon "did it poorly." There are far worse solutions around...:music_whistling: Ultimately I think Sharpe is correct that the first order of business is just a change in the way DCS renders 3d models in the first place.

The trouble with scaling is that size alone really isn't the most crucial factor here. And in focusing on that aspect ED has missed the more important point. Rendering quality. Many of the targets you can have difficulty spotting in this game are close and large enough that the scaling doesn't come into play. The difficulty comes from the muted color palette or shaders or something.

 

We are all set to get HDR or Wide Color Gamut inevitably in all video media so it will be interesting to see how it helps this situation. My guess is that and higher res displays and VR headsets will make scaling unnecessary.

[Why485]

Throwing money at the problem and buying new monitors for all DCS players in 3-5 years is not a solution.

[aairon]

Square wheels don't work, so we shouldn't bother with square wheels.

I want square wheels on my KA-50 so it stays still when I land no need for wheel brake. :smilewink:

[gavagai]

Yes sharpe, I agree with you. I just don't agree that smart scaling looks anything like the DCS impostors. ;)

[GrmlZ]

Hi

Wanted to ask if there are any updates on this?

Anyone know if 2.5 will address this?

[Why485]

Hi

Wanted to ask if there are any updates on this?

Anyone know if 2.5 will address this?

 

No to the first, and unknown to the second. If it helps, we won't know until it gets released.

 

My unrealistic (but not unreasonable) hope is the revised lighting system will just happen to provide much more contrast between planes and their background when it comes to WVR dogfighting.

Edited September 21, 2016 by Why485

[NineLine]

Its still a WIP, they are still looking at all aspects if it.

[Dugong]

Model Enlargement not working??

 

Hi all, i've been away from DCS for a couple of months, just got back into it and after installing all the new patches it seems that the model enlargement feature is no longer working. Anyone know if ED has adjusted this, or it is bugged, or maybe my installation is knackered?

 

Also there is a new check box for 'enable VR headset' does this have any effect on the way my track IR is handled? It was enabled by default and not sure if i should have it on or off.

 

Many Thanks

[Frederf]

Model enlargement is a server/mission difficulty option. The mission can force on, force off, or respect your preference unless you have "use these for all missions."

 

In multiplayer the server can force any state or allow the user to decide.

[Dugong]

Yes thanks, but it doesn't work in single player either for me though.

[Art-J]

Model enlargement doesn't seem to be working indeed, but I haven't tinkered with it all that much lately to comment on it.

 

You can disable all VR features safely. I do it by default to avoid problems with TrackIR we used to have a couple of patches ago.

[TAW_Impalor]

Same thoughts after the latest patches! Flying quick dogfight missions and cannot see sh*t in VR (((

[chihirobelmo]

Has anyone considered about FOV?

We usually see our monitors 30-45deg width in our sight unless you use VR headset, extremely large size monitor, or set your head very very close to the monitor so you can get your eyes worse.

Go here(http://www.projectimmersion.com/fov/) and calculate your monitor FOV,

In my case, I need to set 36 hFOV in the sim to represent 1:1 view against real life.

I usually play DCS or Falcon in 60-80 hFOV. So target aircraft or allies aircraft I see on the screen looks much smaller than what I might see in real life same distance situation.

 

Only increasing resolution wont help target detection, its more helpful for target orientation discrimination. Think there are same size of square box and F-16 miniature on the desk, and you see them 100m,200, and 300m distance from that desk. Is square box harder to detect than F-16 miniature and vice versa? I guess they are not.

 

Representing realistic color change might help representing RL target detection, but its not enough unless we Scale Up the model on the screen when we consider ingame FOV against our monitor FOV in our sight.

 

this academic research I showed Why485 at the another thread would be one of the essential feature to represent both real target detection and orientation discrimination if I set my head position correctly so that magnification matches my FOV setting in game, Monitor size, and My Head Distance from the monitor.

Edited October 7, 2016 by chihirobelmo

[chihirobelmo]

Scaling up models would not be effective when the opposed aircraft is in close distance as its even detectable enough without magnification, but will be essential as they become so small dots on the screen. And this Magnification value can be calculated by tan(Current ingame hFOV)/tan(monitor size in FOV in our sight).

[Why485]

Has anyone considered about FOV?

We usually see our monitors 30-45deg width in our sight unless you use VR headset, extremely large size monitor, or set your head very very close to the monitor so you can get your eyes worse.

Go here(http://www.projectimmersion.com/fov/) and calculate your monitor FOV,

In my case, I need to set 36 hFOV in the sim to represent 1:1 view against real life.

I usually play DCS or Falcon in 60-80 hFOV. So target aircraft or allies aircraft I see on the screen looks much smaller than what I might see in real life same distance situation.

 

Only increasing resolution wont help target detection, its more helpful for target orientation discrimination. Think there are same size of square box and F-16 miniature on the desk, and you see them 100m,200, and 300m distance from that desk. Is square box harder to detect than F-16 miniature and vice versa? I guess they are not.

 

Representing realistic color change might help representing RL target detection, but its not enough unless we Scale Up the model on the screen when we consider ingame FOV against our monitor FOV in our sight.

 

this academic research I showed Why485 at the another thread would be one of the essential feature to represent both real target detection and orientation discrimination if I set my head position correctly so that magnification matches my FOV setting in game, Monitor size, and My Head Distance from the monitor.

 

Abusing the FOV slider to see with a far narrower FOV than you would have in real life is 100% required to get a bead on anything in the game. There's a reason the zoom axis is bound to my HOTAS. Setting a realistic FOV would help when compared to running a traditional 60-80 degree FOV, but in keeping with a realistic FOV you still wouldn't see as much as you should because of pixel densities and it being very difficult to separate planes from the background. To say nothing of the dramatic loss in SA as you will always be "zoomed in" compared to what you would normally fly with.

 

The whole FOV thing is something I know a lot about from playing racing sims. In racing sims, it's a very inelegant solution to a problem that can only really be solved with bigger and higher resolution monitors. Or super high resolution VR headset.

 

I've referenced that paper many times since you posted it, and even implemented my own version of it just for experimentation. While I think it's slightly too generous in how much it magnifies, that's trivially easy to tweak to get something that feels like a good balance.

[Talisman_VR]

Noticed something I had not seen before on a WW2 server with large imposters last night.

Strange green blobs, like glow worms, shining for 3 to 4 seconds, first below me over land (thought it was a ground object at first) and then in the sky above. Later, I saw that it appeared to be LW aircraft types with green tracer. From a distance with the large imposters this tracer was a green glow worm like condensed fluorescent blob of light (at least that is what I think it was). Another reason that I hope the imposter system will soon be consigned to the bin and history. I sincerely hope that the DCS WIP on the air-to-air visibility issue comes up trumps soon. Odd thing is that I did not see a similar red blob effect with the P51D tracer, just the green for the LW. Landing lights look strange with the imposters too.

 

Happy landings,

 

Talisman

[SharpeXB]

Has anyone considered about FOV?

We usually see our monitors 30-45deg width in our sight unless you use VR headset, extremely large size monitor, or set your head very very close to the monitor so you can get your eyes worse.

Go here(http://www.projectimmersion.com/fov/) and calculate your monitor FOV,

In my case, I need to set 36 hFOV in the sim to represent 1:1 view against real life.

I usually play DCS or Falcon in 60-80 hFOV. So target aircraft or allies aircraft I see on the screen looks much smaller than what I might see in real life same distance situation.

 

Only increasing resolution wont help target detection, its more helpful for target orientation discrimination. Think there are same size of square box and F-16 miniature on the desk, and you see them 100m,200, and 300m distance from that desk. Is square box harder to detect than F-16 miniature and vice versa? I guess they are not.

 

Representing realistic color change might help representing RL target detection, but its not enough unless we Scale Up the model on the screen when we consider ingame FOV against our monitor FOV in our sight.

 

this academic research I showed Why485 at the another thread would be one of the essential feature to represent both real target detection and orientation discrimination if I set my head position correctly so that magnification matches my FOV setting in game, Monitor size, and My Head Distance from the monitor.

This is why DCS and every other flight sim has a "zoom view" feature

Explained here.

https://forums.eagle.ru/showpost.php?p=2561114&postcount=220

[chihirobelmo]

This is why DCS and every other flight sim has a "zoom view" feature

Explained here.

https://forums.eagle.ru/showpost.php?p=2561114&postcount=220

 

Yes and that is a system needs to be improved. Always be "zoomed in" and shaking your head with TrackIR to look for visual contact like Radar Antenna does in 35 hFOV would easily lose your SA and Attitude. Doing visual search on DCS should be much more harder than RL. You have to choose whether you look around in Realistic fov and search for unrealistic size drawn aircraft, or look around in Unrealistic fov and search for realistic size drawn aircraft as "Eyeball radar antenna" if the sim doesn't have model enlargement. Smart Scaling solves both.

 

Flying in VFW and participate in 2vs1/2vs2 ACM training based on real operational manuals, I always need to keep eyes on 2-3 aircraft dogfighting while I'm taking 1-2nm spacing from them to maintain mutual support. Visibility system which allows us to notice an aircraft in the same possibility to RL while overlooking around the sky in realistic fov is needed for such a situation.

Edited October 10, 2016 by chihirobelmo