Target visibility

[6S.Manu]

I would like to open a new discussion about this subject.

It's great having a sim where the aircraft's 3D models are close to reality, where performances are very similar to the real ones, where we act on the strumentation as the WW2 pilots did.

When we're infllight the complete picture is really photorealistic. The effects are great and it's almost like flying in a real aircraft (with physical limitation impossible to simulate, like G force,flames ect...).

But IL2 (1946 and CloD) is not only a flight simulator... it's a COMBAT simulator.

Here DMs are detailed enough and the weapons should be already modelled in the correct way.

But you can shoot at an enemy only if you see it and here we have a serious problem: visibility is the most important thing in WW2 air warfare.

I'm not talking about tracking a contact: I'm talking of DETECTION and IDENTIFICATION.

With the help of 3D Studio Max we have reproduced a picture of a 109 at various distances. Then I've taken a pair CloD pictures from the "screenshot" thread (I hope the authors will not complain about this) and from these I've built new images. The error should be in the range of 1 pixel (I'm working with a 24" monitor, 1920x1080)

As first we have the image taken with a 50mm (39.6 fov) to have what the human eyes see.



Below is what we have in IL2's normal view (fov 70):


Note that planes at 3km are already dots...

But at which distance can a pilot detect a flying object?
There are many variables: camouflage (already proved that it's not a magical tool http://www.dtic.mil/cgi-bin/GetTRDoc...c=GetTRDoc.pdf), human eye's threshold of acuity, eye's accommodation, glare, atmospheric haze, target speed differential, target profile, ect... some are really important, others are modest factors.

I was looking for an analysis and here's I found something interesting:
http://www.plosone.org/article/info:...l.pone.0005594

Howell [5] carried out a field study in which pilots attempted to detect another aircraft (DC-3) approaching on a collision course. Over various conditions, the average distance at which detection by the pilot occurred (“detection distance”) was from 5.5 to 8.7 km. Of greater relevance to this study, the subject aircraft also carried an experimenter who knew exactly the approach angle of the target aircraft, and “kept constant vigil with his naked eye” until he detected the intruder aircraft. This “threshold distance”, over the same conditions, averaged from 17.3 to 23 km, about three times larger than the detection distance. We will return to these results later in this paper. Analyzing these data, Graham and Orr concluded that see and avoid failures were due primarily to failure to detect the target [1]. No attempt was made to predict aircraft visibility.

It says "over various conditions"; we don't know the color of the dc3 and the background's color (that's the real factor in target detection, not the plane's color but the contrast between it and the background), we don't know if the persons are professional pilots or not (or better, trained military pilots), but the distances are really higher than the ones in IL2 and they still SEE contacts...

But what's the problem? Dots can replace the flight object but what about identification?
A Dot is always the same at +3km, it does not give to us the profile of plane, its direction, the model of that plane. And more Dots are not always plane... in CloD ships are showed as dots at very long distance.

Additionally dots are merging with the background and in the case of the detailed ground we don't have our eyes focusing on it and "excluding" the objects around it: the wider is the speed differential the easier is for our eye to focus on the target...

Don't misread me... target identification is not a easy thing to simulate: they are still studying it and there are dozens of variables. Maybe one we all will be playing on monitor with amazing resolutions and it will be a lesser problem.. but in these days I think that Luther and Co. need to find a solution to it.

Otherwise CloD will be a great flight simulator but an half WW2 combat simulator.

I think that Dots are not the answer. What your opinion?

Perhaps dots are not the answer but I doubt that one can compare the visibility of a comparatively large aircraft (DC-3) to the visibility of a small aircraft that is only half the size or less.

To be honest I sometimes look up when I see some condensation trails and I do see the plane but could not tell its type (well they all look the same anyway) nor do I distinguish many details knowing that they fly at 11 km and basically are huge compared to the planes we consider here.

[LoBiSoMeM]

In IL-2 1946 dots works. We neen bigger dots in CloD (fixed size, for ANY resolution) and better AA. Maybe some "reflection" too in far objects...

Just that.

Mh. Not sure. Actually I think the difficulty to see planes is an improvement over IL2.

There are numerous accounts that pilots just missed each other despite passing not too far from each other.

I understand the optical issue brought here by OP but IL2 is not a realistic example. With the old IL2 logic one could have as well made the points blinking with a pink arrow painted on top of them accompanied with the tag "Here! Here!"

One can spot planes from a quite a distance (4 km) in Clod if one is attentive and in a good position. Which btw is about in accordance with what Manu wrote on the experiments done by Howell remembering that the plane in Howell's experiment were at least double the size of a fighter. So my guess is visibility limit of 4-5 km for a small plane such as a fighter is fine.

What could be perhaps discussed if at 3 km one could not paint a small trait instead of a dot. I think this is the major concern of OP.

If a better solution of the optical distortion is found without going arcade I will be happy. Old IL2 is imho not the way to go.

[LoBiSoMeM]

To do some testing, we can have an option of "no artificial dots" at all. Maybe in next beta patch?

"Realistic Target Visibility"!

The first LOD will be REALLY pixel size. And people with lower resolutions and big monitors will have some advantage. But we can test.

[Insuber]

To ensure some playability, and avoid boredom killing this game, we need above all to improve DETECTION range by increasing the dot visibility and contrast to 15-20 km, with normal visibility (as in the US Navy document posted by Manu earlier), and also solve the issue of disappearing contacts - linked to the LODs probably. A darker dot would do, imho.

Then, as far as IDENTIFICATION, I agree that in RL you can tell a Spit from a Bf from a longer distance than in this game (I think to events like Duxford, where the plane's silhouettes and colours are identifiable from very far away, let's say 3-5 km). Color of camo helps a lot, and in old Il2 the LOD's at far range were typical of a certain model (more squared for Bf, thinner for Spit); working on LODs to differentiate them a bit may be a solution. Skilled players could then ID a distant contact from the predominant color of the camo and the form of LOD.

Cheers!

[Untamo]

I have never undestood the use of these artificial dots. Why not just use a really simple 3D model (just a few polys) when the aircraft is far away, the 3D engine (if it works correctly) would render the target as it should be seen. What we are after here is the right amount of degrees/radians the aircraft occupies on our screen at given FOV.

When an aircraft gets further away, eventually it is rendered as one pixel, and as it get more distant, this one pixel gets more faint (opacity drops).

At lower resolutions the aircraft would be one pixel at closer range and at higher resolutions further away. Detection wise neither one would have advantage against the other, as at lower res the pixel is larger and at higher res smaller but the target may still be several pixels at the same distance when the lower res screen shows just one pixel. But the size (the degrees/radians it occupies on the screen) of the target would be the same on both screens (assuming the physical size of the monitor is the same).

[ZaltysZ]

Originally Posted by cheesehawk

There's got to be a way for the game to grab your monitor size and resolution from Windows device manager. Then calculate 100% correct size for each object.

If you talk about real life scale, then you also need to know distance from observer to screen for this to work. Windows device manager does not know that.

[Untamo]

Originally Posted by ZaltysZ

If you talk about real life scale, then you also need to know distance from observer to screen for this to work. Windows device manager does not know that.

Indeed. Nor the physical size of the monitor. Say a 17" monitor vs. a 32" low res TV (used as a monitor) might have similar resolutions but a wholly different viewing distance.

Originally Posted by Insuber

To ensure some playability, and avoid boredom killing this game, we need above all to improve DETECTION range by increasing the dot visibility and contrast to 15-20 km, with normal visibility (as in the US Navy document posted by Manu earlier), and also solve the issue of disappearing contacts - linked to the LODs probably. A darker dot would do, imho.

Then, as far as IDENTIFICATION, I agree that in RL you can tell a Spit from a Bf from a longer distance than in this game (I think to events like Duxford, where the plane's silhouettes and colours are identifiable from very far away, let's say 3-5 km). Color of camo helps a lot, and in old Il2 the LOD's at far range were typical of a certain model (more squared for Bf, thinner for Spit); working on LODs to differentiate them a bit may be a solution. Skilled players could then ID a distant contact from the predominant color of the camo and the form of LOD.

Cheers!

Keep in mind that the aircraft to be spotted in the US Navy experiment was about or even more twice the size of the planes we discuss here (why do I have to repeat myself?). So I have very very strong doubts that you just can take the findings there and transpose it to fighters here.

I think the dot contrast is fine so I do not need a change here. If you have some difficulties finding the dots due to contrast please first tweak your own monitor settings or the graphic card settings before requesting a change in game that is fine for many others

I agree that something should be done with the presentation between first appearance as dot at 4 km and a presentation as a scaled down model at 2 km). I'd propose to use a something like a dash or so (could be a bold dash at a certain distance).

Sorry but I think it is absolutely unrealistic to see a fighter plane in 20 km distance even as a dot. These fighter planes have a wingspan of about 10m only. Do not compare it to airliners that have 6 or 8 times that wingspan.

The 4 km may be to pessimistic but the 20 is imho unrealistic. Please keep in mind that the 15-20 km in the US Navy experiments were obtained with a bigger aircraft and with the pilots knowing EXACTLY where to look. Both conditions (big aircraft, exact knowledge of position of plane to be spotted) do not apply for fighters in the situations we discuss here. So probably if you knew exactly where to look for a plane you could see a fighter at 7-10 km. But only if you knew exactly where to look.

The only real big issue that I see is that planes with a 3d model suddenly disappear into a dot or what I frequently experience that when zooming my view on a dot or a small plane (at about 2km) it gets invisible (perhaps a smooth zoom in function could work).