Spotting Distance... again

[NineLine]

Look at this image - it's a tanker, 900 meters away. IRL it would be easy to see it, it would be so obvious, only a blind man would not see it.

 

 

And here it's barely visible, no contrast at all, almost blended with the sky. If you look slightly away, it is gone, while in real life you could easily see it with peripheral vision.

 

 

And this is a problem with visibility in DCS, where aircraft in close vicinity are hard to spot, not the ones 10 miles away.

 

 

 

 

 

 

 

What are your settings, and do you have this exact mission, I just set this up real quick, and can see it fine. I can do a comparison if you want, sun position makes a difference, as it does in real life.

 

[grunf]

What are your settings, and do you have this exact mission, I just set this up real quick, and can see it fine. I can do a comparison if you want, sun position makes a difference, as it does in real life.

 

Thanks for looking into it. I don't have the mission, it was just a quick one from the editor, but I recreated the new one with more-less same conditions (position, course, time of day).

 

 

i5 4460 3,2GHz

GTX970

16GB RAM

 

 

edit: added slightly modified mission, more similar to the original conditions with

vis_tst.miz

vis_tst2.miz

Edited March 18, 2019 by grunf
added another mission

[Baco]

What everybody is omiting is that he human EYE is designed to spot moving things "prey" you do not have preifherial vision in a monitor, that is why aircraft shuld be more visible than in "reality" beacuse we are looking in windos thats is 2D. we need artificial methods like track ir, to interact wth what we see in the sim. having planes at real sizes is a handicap, not realism...

[Baco]

The sim that shall not be named works well because planes are bigger that real size, so you can actualy spot them in a realistic manner considering the handicaps the 2D works presents. In Falcon planes are really dar so you get the same efect of being able to spot a dark moving contrasting object...

[SharpeXB]

The ___________________ <—— fixed this for you :music_whistling:

works well because planes are bigger that real size, so you can actualy spot them in a realistic manner considering the handicaps the 2D works presents.

For the umpteenth time... DCS cannot do “smart scaling”

 

having planes at real sizes is a handicap, not realism...

And having artificially larger planes and ground targets makes them TOO visible. It’s been done here already and the results were terrible. And even on the largest setting people still complained that they couldn’t see anything.

Edited March 18, 2019 by SharpeXB

[NineLine]

Thanks for looking into it. I don't have the mission, it was just a quick one from the editor, but I recreated the new one with more-less same conditions (position, course, time of day).

 

 

i5 4460 3,2GHz

GTX970

16GB RAM

 

 

edit: added slightly modified mission, more similar to the original conditions with

 

 

From your second test file, one shot through the front and then off to the side, I use a slightly higher gamma of 2.2 so mine is a little darker, I also use higher quality shadows and some other higher settings.

 

 

[Basco1]

@ Nineline........Would it be at all possible Nineline for us to fully see your full in game settings please,your screenies do look clear.

[chihirobelmo]

Nineline,

 

Considering each user has different settings, DCS cannot just have "everyone would be happy" system easily. I agree.

But why not just consider "works for some" ideas like Serfoss 2003 magnification, as an "option" like current dot only label system is there?

 

A server can allow/force/disable those options so it can just help a group .. like Virtual Fighter Wing ... who play with well-known friends.

Just only small monitor / low resolution setup user can enable the option they want for realism. Others can just uncheck that option.

 

A public server like 104th or GAW might needs "everyone would be happy" system or force nothing for balance, but closed server don't need it, we only need an option.

 

EDIT: and also for single player.

Edited March 19, 2019 by chihirobelmo

[Random]

Question... if smart scaling would make radar crosss section larger as has been implied. Will dropping resolution do the same? Bigger pixel after all.

[SharpeXB]

Question... if smart scaling would make radar crosss section larger as has been implied. Will dropping resolution do the same? Bigger pixel after all.

As I understand it, smart scaling changes the size of the 3D models so players changing their resolution isn’t the same thing.

[Tippis]

Question... if smart scaling would make radar crosss section larger as has been implied. Will dropping resolution do the same? Bigger pixel after all.

Different resolutions make sensors behave differently and since sensor detection is apparently tied into the rendering pipeline, this makes sense. In VR, for instance, EO sensors have a much easier time picking up on and locking targets. Whether the same holds true for radar is hard to tell since you'd only notice a difference at ranges where they're already very tiny targets.

 

Of course, this just further demonstrates the need to actually fix the underlying horrendous problem, and the ability to implement proper and normalised scaling would more be a happy side-effect of such a desperately needed fix.

Edited March 19, 2019 by Tippis

[Schmidtfire]

+1

 

Hundreds of DCS users (posting here or elsewere) are not wrong. There is an issue with spotting. I think that has been well established and documented over the years. Very happy that one of the biggest drawbacks in DCS gets recognition once again. Hopefully the brilliant minds at ED will figure out a way to make it better :)

 

It's great to have NineLine present in this thread. As for the screenshot comparisons. While the tankers are more visible in his screenshots, I think they are still a bit hard to notice. They are flat and un-saturated colorwise. A fighter sized target will easily melt into the background.

 

And to those who say spotting is hard IRL. Fighter pilots have good vision. We players for the most part have average vision and doing all the spotting on a small screen (of different quality and resolutions). Even though the "math" inside DCS engine might be right in how it draws the objects, It does not factor in hardware limitations on the player side. Im sure it is great if we all are running 12K+ VR goggles, capable of HDR P3 or Rec2020 colorspace at 90fps. But that is sadly far far away.

 

Thanks for revisiting this topic. Hope that it'll gain some traction.

[bell_rj]

Thank you for the supportive comments. When I started the thread I had the sole objective of achieving greater visibility of the topic with the ED community manager, and then hopefully designers and developers. Thank you to NineLine for his involvement. The icing on the cake for now would be confirmation from NineLine that this topic has, or will soon have, received some additional focus within ED.

 

I'm sorry if my initial post made some uncomfortable. Flying in DCS can be an emotive experience. I'm a big DCS fan, and that means getting excited about new developments and so it's only natural that the flipside is sometimes disappointment. That's life.

[SharpeXB]

Different resolutions make sensors behave differently and since sensor detection is apparently tied into the rendering pipeline, this makes sense.

Do you have evidence of this? What Wags was talking about was changing the size of the 3D models themselves not screen resolution. Having sensor effectiveness simply tied to screen resolution doesn’t seem to be a really good method and I’m doubting DCS would do something like that.

 

+1

As for the screenshot comparisons. While the tankers are more visible in his screenshots, I think they are still a bit hard to notice. They are flat and un-saturated colorwise. A fighter sized target will easily melt into the background.

Right. And a problem with just using “scaling” as a solution is that it does nothing to fix this. That’s one reason why the previous Model Enlargement feature didn’t help. It didn’t improve on rendering or picture quality.

IMO the best solution for that is HDR video output. Greater contrast and color depth are the key factors here. If you see the brightness and contrast of an HDR picture you’ll know what I’m talking about, especially with games this looks really brilliant. This is not a new technology and is widespread in current displays, HDTVs, game consoles and most AAA games. Sadly the one realm that is falling behind in adoption is PC gaming.

Edited March 19, 2019 by SharpeXB

[Tippis]

Do you have evidence of this?

It is a requirement for Wags' statement to be true. So it comes down to whether or not you believe him or not.

 

Having sensor effectiveness simply tied to screen resolution doesn’t seem to be a really good method and I’m doubting DCS would do something like that.

No, it is not a good method but again, it would be a direct consequence of what Wags is saying and would also explain behaviours such as how EO effectiveness is much improved in VR.

[SharpeXB]

It is a requirement for Wags' statement to be true. So it comes down to whether or not you believe him or not.

What I understand Wags saying was that “smart scaling” affected the radar cross section. Assuming that means smart scaling changes the size of the 3D models in the game and those 3D models are also used to generate the RCS. He’s also mentioned something like “object fusing”

As I follow it that’s why ED implemented a 2D sprite in order not to change the size of the 3D model. Regardless the size of the 3D model in the game is unrelated to screen resolution.

[Ironhand]

Different resolutions make sensors behave differently and since sensor detection is apparently tied into the rendering pipeline, this makes sense...

FWIW, just played back a track involving EOS under two different resolutions. Picked up the target at exactly the same range and time.

[Tippis]

What I understand Wags saying was that “smart scaling” affected the radar cross section. Assuming that means smart scaling changes the size of the 3D models in the game and those 3D models are also used to generate the RCS. He’s also mentioned something like “object fusing”

 

As I follow it that’s why ED implemented a 2D sprite in order not to change the size of the 3D model. Regardless the size of the 3D model in the game is unrelated to screen resolution.

 

There are two options:

 

1) What he says is true — that smart scaling cannot be done because it would affect RCS. This means that RCS calculations must necessarily be leeching off data from the rendering pipeline. Otherwise, the two simply could never ever have any relation whatsoever to each other.

 

In other words, when the game decides what size to render a unit on any given frame, that data is also siphoned off to the sensor calculation and used to determine how large it is to the sensors. If during that rendering process, smart scaling is applied, the size to be rendered is changed, and therefore the “sensor size” is also changed. This also means that resolution (since it directly affects that whole process) affects sensor effectiveness.

 

This would also explain why EO sensors have an easier time picking up targets in VR, which is an observation in favour of this claim being true.

 

or

 

2) What he says is not true. The rendering pipeline determines what size it should render the unit. A completely different and separate process determines how sensors see the unit (which may or may not use the 3D mesh, but that's not relevant).

 

This means that it is utterly trivial to add a scale transform to the rendering process that changes what size a unit will be on-screen, and since that data is not being used anywhere else, the process that determines “sensor size” can go on its merry way because its base data remains untouched (even if it uses the same 3D mesh because the mesh is not scaled after all).

 

In favour of this is that such a solution would make infinitely more sense than having the rendering pipeline contaminate unrelated processes such as sensor simulation.

 

 

Those are the only two options: either rendering parameters — including resolution — affects sensors, which makes unit scaling difficult, or they are separate and unit scaling can be done without affecting sensors. Pick which one you believe in.

[speed-of-heat]

I think there may be third option, which that aircraft RCS is directly derived from aspect and size, rather than the physical render pipeline, so its a single value that is used for both... you may be saying the same thing...

[Tippis]

I think there may be third option, which that aircraft RCS is directly derived from aspect and size, rather than the physical render pipeline, so its a single value that is used for both... you may be saying the same thing...

If the RCS is not directly derived from the rendering pipeline, you could trivially scale the on-screen representation up or down to your heart's desire — an MQ-1 could be drawn the size of VY CMa and still be difficult to pick up on radar.

 

Supposedly, this is not the case.

Edited March 19, 2019 by Tippis

[SharpeXB]

There are two options:

 

1) What he says is true — that smart scaling cannot be done because it would affect RCS. This means that RCS calculations must necessarily be leeching off data from the rendering pipeline. Otherwise, the two simply could never ever have any relation whatsoever to each other.

 

In other words, when the game decides what size to render a unit on any given frame, that data is also siphoned off to the sensor calculation and used to determine how large it is to the sensors. If during that rendering process, smart scaling is applied, the size to be rendered is changed, and therefore the “sensor size” is also changed. This also means that resolution (since it directly affects that whole process) affects sensor effectiveness.

 

This would also explain why EO sensors have an easier time picking up targets in VR, which is an observation in favour of this claim being true.

 

or

 

2) What he says is not true. The rendering pipeline determines what size it should render the unit. A completely different and separate process determines how sensors see the unit (which may or may not use the 3D mesh, but that's not relevant).

 

This means that it is utterly trivial to add a scale transform to the rendering process that changes what size a unit will be on-screen, and since that data is not being used anywhere else, the process that determines “sensor size” can go on its merry way because its base data remains untouched (even if it uses the same 3D mesh because the mesh is not scaled after all).

 

In favour of this is that such a solution would make infinitely more sense than having the rendering pipeline contaminate unrelated processes such as sensor simulation.

 

 

Those are the only two options: either rendering parameters — including resolution — affects sensors, which makes unit scaling difficult, or they are separate and unit scaling can be done without affecting sensors. Pick which one you believe in.

Screen resolution has nothing to do with the actual size of the 3D models in the game, why would it? The whole analysis above therefore doesn’t make sense.

 

And it’s moot because scaling or changing the actual or apparent size of other objects in the game is a terrible solution. It’s the sort of thing that was done in primeval games and is out of place in today’s world. The solutions for today involve higher resolution displays and better rendering. Looking forward, not at past solutions.

[Tippis]

Screen resolution has nothing to do with the actual size of the 3D models in the game

That's not what I'm saying. The logic goes in the other direction.

 

And it’s moot because scaling or changing the actual or apparent size of other objects in the game is a terrible solution.

It really isn't if you want to make a proper simulation of perception or if you want to do any kind of size normalisation between different displays. And you kind of what to do both of those.

 

You can no longer rely on the antiquated notion that resolution is a static thing or that you can rely on some assumption about hardware — you have to design for a wide array of different setups. The gaps and differences will only grow larger as we go forward.

 

 

Above all, it is not moot for the simple reason that, if sensors are related to rendering in the way implied, then this is a awfully, horribly, utterly and completely broken implementation that must be fixed promptly. There simply is no going forward with such an antediluvian monstrosity lurking in the code. The other benefits such a fix would unlock are almost just a marginal happy coincidence compared to that basic necessity.

Edited March 19, 2019 by Tippis

[SharpeXB]

That's not what I'm saying. The logic goes in the other direction.

 

It really isn't if you want to make a proper simulation of perception or if you want to do any kind of size normalisation between different displays. And you kind of what to do both of those.

 

You can no longer rely on the antiquated notion that resolution is a static thing or that you can rely on some assumption about hardware — you have to design for a wide array of different setups. The gaps and differences will only grow larger as we go forward.

 

 

Above all, it is not moot for the simple reason that, if sensors are related to rendering in the way implied, then this is a awfully, horribly, utterly and completely broken implementation that must be fixed promptly. There simply is no going forward with such an antediluvian monstrosity lurking in the code. The other benefits such a fix would unlock are almost just a marginal happy coincidence compared to that basic necessity.

Honestly none of us know how this game engine really works. If Wags says ”smart scaling” can’t be done then it can’t be done. Rendering objects out of scale was already tried, turned out to be terrible and abandoned. So no point in revisiting it in any form.

[Tippis]

Honestly none of us know how this game engine really works. If Wags says ”smart scaling” can’t be done then it can’t be done.

 

 

Quite. The point is that his saying that comes with a some necessary implications that desperately need to be fixed irrespective of how you want unit sizes to work. The fact that those fixes would also allow smart scaling to be done is incidental.

[SharpeXB]

Quite. The point is that his saying that comes with a some necessary implications that desperately need to be fixed irrespective of how you want unit sizes to work. The fact that those fixes would also allow smart scaling to be done is incidental.

And really scale has nothing to do with this issue. Look at the screenshots of the tanker above. It’s huge. Scaling it bigger doesn’t make the image quality any better.

 

Besides, if you want objects on your screen bigger the solution is easy. Get a bigger screen and sit closer to it! A big 4K TV costs as much as a joystick these days. That’s a problem which is easily solved.