Research and Real-Wolrd data on aircraft visibility

Hi all,

The issue of what should be the most realistic aircraft visibility range in a flight sim has been raised before in these forums. There seems to be a paucity of real-world data in this matter and a lot has been left to speculation. A lot of the available evidence comes from very old experimental studies (dating back half a century ago), and then again, this evidence is very hard to find. The multifactorial nature of this problem and the emergence of radar probably played a role in this.

Recently, a research paper came to my attention, that was published in PLoS one (a respected scientific journal) addressing this issue. The authors are affiliated with NASA Ames research center, and the results of their work may be of interest to flight simulator development.

Their paper is entitled : "Predicting Visibility of Aircraft" and the link can be found here :

http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0005594 (access is free).

In their work they describe a tool that they have developed, termed Spatial Standard Observer (which they have patented), which incorporates divergent parameters such as aircraft size, shape, lumination, colour, contrast, etc, and predicts various parameters, the most relevant of which -for flight sims- is the threshold range (or threshold distance). The latter is defined as "the largest distance at which an aircraft can be seen" (for the given parameters). Their model predictions are tested against human observer results, setting up an experiment with 3d aircraft images in a monitor. They also compare their results with older data from real-world experiments.

Their work highlights the fact that for given atmospheric conditions, threshold range depends largely on contrast (the difference in luminosity between aircraft and background), target size (i.e. wingspan, etc) and spatial frequency of the background (roughly a measure of its complexity).

The characteristic of the targets tested are shown in the image below :

And the model predictions for these targets (threshold range vs contrast) is depicted in this chart :

The authors do note that these results are for specific view and illuminations conditions, and smaller or bigger threshold ranges may arise if these are varied. However, they could prove useful experimental data for a flight sim, in my opinion.

According to this chart, the threshold range for an F-16 (wingspan 9.8 meters) is about 5 kilometers at an average level of contrast of 0.50, and reaches 7km at high contrast rate 1.0. A C17 (wingspan 51.7m), should be visible at 20km and 30km at the respective contrasts. Making an extrapolation, a WW2 fighter such as the P51 with 11m wingspan should be visible at slightly higher ranges than the F16, whereas a B17 (wingspan 32m) should be visible at approximately 15-20 kilometers.

In this paper there is also another interesting chart from a study in a real-world setting, where pilots where asked to spot a DC-3 (wingspan 29m, close to a B17), coming at different angles in a collision course (this is not in a monitor setup). The threshold distance, over the same conditions, averaged from 17.3 to 23 km.

And a final interesting paper referenced in the citations :

http://www.ncbi.nlm.nih.gov/pubmed?cmd=Search&doptcmdl=Citation&defaultField=Title%20Word&term=Contrast%20sensitivity%20predicts%20pilots%27%20performance%20in%20aircraft%20simulators.

In an experiment done in pilot instructors in aircraft simulators, their performance in detection of targets was found to be better related to their contrast sensitivity and not their visual acuity.

I hope some of you guys find this stuff useful.

cheers

Hmm, that actually looks quite useful.

What I can't really decipher from the paper is how that SSO (spatial standard observer) and the eyes of the test persons they used hold up against the eyes of real pilots (which are supposed to see better than average people).

I wonder how the game engine can really use those numbers though, especially if we are talking about MP (which is actually the only case where it really matters). Won't people with different monitors and different contrast settings have a completely different end result?

and different FoV also

The 3 participants are the actual authors, and are quoted as "experienced observers" in the paper. All underwent visual acuity test by an optometrist. However, you are right in that comparing their eyes to pilot's eyes may introduce a bias. They tried to adjust for this by comparing their results against the results of the earlier study by Howell for the DC-3 (which was done on pilots). As you can see, the results are not very different.

There may be a way that these data could be used by the engine, but only the devs could know if the impementation in code is feasible.

First off, they provide a data pool for extrapolations that is based on research and not speculation or sporadic pilot accounts. For example, if the current game engine starts drawing the pixel for the P51 at X meters, the data mean that it should start drawing the pixel for the B17 at 3X meters. By referring to the above chart, various aircraft sizes could be initially drawn at the respective distances.

Second, developers could adopt the same strategy as the authors did in their paper, in order to test their LODs. You are absolutely right that the game is played at different monitors, with different size, contrast, etc, and also FOV, which introduces great variability in the results. However, the mean matters most when it comes to estimations. So, if a consensus is reached that i.e. a player with normal (standard) FOV should be able to spot a P51 against the sky as a background, at 7-8kms (which comes close to real world data), then a simple mission could be designed to test this : the tester spawns at a static point looking at a specific direction in the sky where the enemy aircraft is coming from and pushes a button as soon as the target is spotted at his rig. The results of many tests and many testers (why not even the whole alpha backers who represent the player pool with vastly different rigs) would give a chart similar to the DC-3 study, which the developers could use in order to adapt their LODs so that the mean threshold range is similar to real world data. This way, they will also get feedback on the degree of variability and which factors affect it the most -i.e. resolution? warm colors? monitor size? graphics settings? Please note that variability is significant even in the real-world study by Howell: if we look at the DC-3 chart, some pilots spotted the aircraft at 13kms, whereas others at 23kms (under the same visual conditions). The results of each tester could be transferred to the devs in a similar way that the 1.2.7. beta did. However, this strategy may be too time consuming for the devs or not feasible at all. It's just a suggestion.

Aircraft visibility is crucial... So this is definitely an interesting topic. I am very curious if the rendering with the introduction of EDGE will improve our impression of a/c spotting distances and such within the sim.

Hmm, that actually looks quite useful.

What I can't really decipher from the paper is how that SSO (spatial standard observer) and the eyes of the test persons they used hold up against the eyes of real pilots (which are supposed to see better than average people).

Pilots are just like everyone else when it comes to eye sight. :smartass:

I would hope we are basing our theories/observations on persons with 20/20 or better vision (corrected or not)

Pilots are just like everyone else when it comes to eye sight. :smartass:

I'm not quite sure whether this is correct for fighter pilots. I can't imagine the military allows people to fly 9g-turns wearing contact lenses or glasses, do they? So if you got bad eyes they won't let you fly a fighter jet.

Therefore I thought it's safe to assume that (until a certain point of course) having good eyes is a requirement. Correct me if I'm wrong.

And yeah, of course in the real world not all of them see equally well. Some people can see extraordinarily well, others just normal. Chuck Yeager was famous for his ability to spot airplanes at twice the distance (several minutes earlier) than other pilots, and so was Erich Hartmann.

As for the study: As far as I can tell they did their research well, if there is a small bias it shouldn't be too important, a few miles give or take.

My two cents..

In my experience as a commercial pilot.

Looking out for traffic is very common, theres lots of planes in the sky.

and as a sim pilot i usually think is much more easier to find airplanes. because at distance they are usually a black spot on your monitor.

In real life, even using a TCAS (traffic collision avoidance system.. a little device that tells you exactly where to look for) its a little harder.

Sometimes we are looking for a traffic just in front of us, heading our way a couple of thousand feet up or down and we spot him just 5 miles or less.. sometimes we dont see him at all.

And also theres times we can spot them at 20 miles or so.

It depends alot! on the background.. a nice cloudy day makes much easier the job.. I think the eye can spot an object moving againt a "static" cloud much easy.

But when the traffic is over a city is much more difficult. If the traffic is a Helo at 500ft and your ar above, forget about it.. you are never gonna find it.

For me the tipical black spot in sims is much more easier to find.

At night... its completly different.. the strobe light in DCS are way to dim...in real life you can spot them 20 30 40 miles..

All real pilots I know agree with that.

My own personal experience is only based on flying gliders, which I admit is a bit different, but I also agree. Spotting a plane, even one quite close to you can be very hard.

One would assume that since a glider is white, and moving you _should_ see it fairly easy, but it is small, and it has a thin fuselage and body. If it is below you and the background is some forest, it is rather simple. But a cluttered background such as a village makes it really hard. And of course color and size are very important.

About the lights... I think in DCSW lights can only be shown on the screen when the model is actually drawn. As long as the plane is just a dot there are no lights. In real life the lights are seen much earlier then. Not sure though.

The problem with most games is that they instantly jump from not visible at all to a pure black or colored dot. I remember flying the first IL-2 in a LAN party with friends, and spotting someone was never a problem. You couldn't hide by attacking out of the sun's direction or such things. Regardless of the size or shape of an airplane you could spot it (and not lose it again, which is also a big problem in real life) rather easily.

Most newer games do it much better, with a pixel that gradually becomes more opaque as the plane moves closer.

My two cents..

 

Sometimes we are looking for a traffic just in front of us, heading our way a couple of thousand feet up or down and we spot him just 5 miles or less.. sometimes we dont see him at all.

And also theres times we can spot them at 20 miles or so.

This I think is the essence of the issue. Even when you know exactly where to look, with an optimal background, on a perfect day, you will never see a P-51D within DCS at 20nm...or 10. At least, that is what I've experienced.

This has been posted before in other threads but does use real life military aircraft done by real world fighter pilots.

http://www.dtic.mil/cgi-bin/GetTRDoc?AD=ADA241347 Though it is only for one specific set of conditions which I don't believe are mentioned. It is a good document.

thanks for sharing DGC, that's interesting

I'm not quite sure whether this is correct for fighter pilots. I can't imagine the military allows people to fly 9g-turns wearing contact lenses or glasses, do they? So if you got bad eyes they won't let you fly a fighter jet.

Interestingly although slightly o/t I'm pretty sure I saw a documentary about the Tornado crew shot down during the Gulf War and I think the pilot wore hard contacts and carried glasses and one of his first activities on getting to the ground was to switch between them. Might be mixing it up with the navigator but one of them definitely had contacts. It was also a concern when they underwent interrogation on capture.

Perhaps the entry requirements specify good vision but if your eyesight degrades after the system spends a lot of money training you the regs are more lenient. I can remember taking particular note of it as vision that wasn't up to standard was the reason I never entered the RAAF when I left school so I was surprised he was able to make it in the RAF as aircrew in something like a Tornado.