w.i.p New spotting is great / New spotting is overcooked feedback thread

[SharpeXB]

1 hour ago, James DeSouza said:

2-10 miles with a rough average in the 3-5 range is actual spotting ranges. 

It doesn’t matter how much real life data you cite. Players have been convinced by video games over the years that they’re supposed to easily see aircraft at great ranges. Cause that makes video games “fun”

[Tippis]

6 minutes ago, SharpeXB said:

It doesn’t matter how much real life data you cite. Players have been convinced by video games over the years that they’re supposed to easily see aircraft at great ranges. Cause that makes video games “fun”

Here, it sounds like you're against that. And yet, you're arguing in favour of even greater ranges — contrary to all data and scientific inquiry quoted to you — and against having those ranges reduced to far more sensible numbers. Why is that?

[BIGNEWY]

folks please knock it off with the back and forth, both of you need to keep to the forum rules they can be found at the top of the forum. 

[James DeSouza]

12 hours ago, SharpeXB said:

It doesn’t matter how much real life data you cite. Players have been convinced by video games over the years that they’re supposed to easily see aircraft at great ranges. Cause that makes video games “fun”

I am more curious about that guy who could only manage to see an A-4 at 0.38 nmi.  I know A-4's are small, but that sounds kind of silly.

[SharpeXB]

55 minutes ago, James DeSouza said:

I am more curious about that guy who could only manage to see an A-4 at 0.38 nmi.  I know A-4's are small, but that sounds kind of silly.

This discussion certainly has run its course. Any further “I can’t see the planes” anecdotes need to be accompanied by tracks and screenshots as well as game settings. Show us an actual screenshot of an “invisible” plane at .38 miles. Let’s see a label floating over empty sky to prove this claim. This is in fact a bug thread so show us an actual bug with documentation. Otherwise it’s apparent that some people just have poor eyesight. 
And if ED can’t arrive at an actual solution then just give us a genuine OFF option that’s server enforceable. Then we can all move on. 

[71st_Mastiff]

25 minutes ago, SharpeXB said:

This discussion certainly has run its course. Any further “I can’t see the planes” anecdotes need to be accompanied by tracks and screenshots as well as game settings. Show us an actual screenshot of an “invisible” plane at .38 miles. Let’s see a label floating over empty sky to prove this claim. This is in fact a bug thread so show us an actual bug with documentation. Otherwise it’s apparent that some people just have poor eyesight. 
And if ED can’t arrive at an actual solution then just give us a genuine OFF option that’s server enforceable. Then we can all move on. 

I cant see why you would label everyone with poor eye-sight, when in fact its something to do with setting; Antialiasing, MSA, Resolution size, etc., etc., etc. So the best option is 3D modeling of sizes and renderings. which they used to do back in the day, I don't understand why it cant be done any longer?

Also take into account that, in the near future VR will have double maybe tripled in FOV and clarity.

[SharpeXB]

34 minutes ago, 71st_Mastiff said:

I cant see why you would label everyone with poor eye-sight,

Eyesight is the most simple explanation of why one person can see something that another person can’t. In order to prove that this isn’t the reason we need screenshots etc. 

[Why485]

On 1/20/2024 at 1:02 AM, James DeSouza said:

Spotting aircraft IS HARD.  That's the whole point.  "Lose sight lose the fight".  It's hard to lose sight when the opponent is a giant black blob.

This isn't so much aimed at you, more so whatever audience for this discussion might be left, however it's a great point to respond to as it's a common sentiment I've seen.

In a good spotting system, you will still lose sight against a noisy background and ground clutter. Keeping track of the target in a fight would still be a skill that takes practice. A good spotting system aims to provide realistic results, and I've given examples of what those realistic results should look like many times in the past. E.g. for WWII planes, these are good reference numbers. Even with the current (new) and flawed system, you still regularly see people talk about how they lose sight or have difficulty spotting and what people like me are advocating for will in all likelihood make spotting more difficult than it is now.

My favorite discussion on the topic is somebody saying to me (paraphrasing), "Back in ye olden days of a certain WWII flight simulator from the turn of the millennium, we didn't need any kind of dots. Why do we need them now?" And like, that was pure *chef's kiss* because not only did this beloved WWII flight sim have dots, but it also used a basic form of scaling through creative LOD rendering. For the record, the distances planes became visible in ideal conditions matched the above cited numbers.

That's what I'm after. When these sorts of systems are done right, you won't notice it's doing anything at all.

Edited January 22 by Why485

[Tippis]

4 hours ago, Why485 said:

My favorite discussion on the topic is somebody saying to me (paraphrasing), "Back in ye olden days of a certain WWII flight simulator from the turn of the millennium, we didn't need any kind of dots. Why do we need them now?" And like, that was pure *chef's kiss* because not only did this beloved WWII flight sim have dots, but it also used a basic form of scaling by through creative LOD rendering. For the record, the distances planes became visible in ideal conditions matched the above cited numbers.

That's what I'm after. When these sorts of systems are done right, you won't notice it's doing anything at all.

And that example also highlights another crucial component in all of this: “spotting” isn't a single thing, and it cannot be served a single system.

There are (at the very least — arguably there could be more) three aspects that a spotting mechanic needs to deal with, each within its own range band.

1. Edge-of-visibility, long-range target detection. When a plane crosses over from BVR with just within WVR range.
2. Medium-to-long range aspect detection and target tracking. When a plane is WVR and you're trying to anticipate where it's going so you can keep track of it and get a suitable pursuit solution on it.
3. Medium-to-short range target identification. When it's time to figure out whether you should shoot or not.

DCS has always struggled, and pretty much consistently failed, to deal with all three aspects.

Spotting dots are, and can only ever be, an answer to the first of these. As described earlier, you have this problem on modern display systems that they are capable of rendering objects way smaller than the pilot should be able to see, and thus to render them at far too long distances. And that's before we even bring zoom into the equation, which will obviously make that much worse. To maintain a good perception simulation these days, we have to actively degrade what can be seen and what is drawn on the screen, but that in turn means we need a mechanism to cover up the gap between when we're in “this is the smallest thing you can see in-sim” mode, and when we're in “pure perspective gives an accurate target size” mode. And that mechanism needs to avoid pop-in, it needs to avoid being inequitable (and I use that word very specifically as opposed to “unfair”) for no good reason, and it needs to be responsive to zoom so that “smallest thing” remains the smallest no matter how you manipulate FoV.

I can't remember seeing any suggestion other than dots that can properly solve problem #1. Scaling might, but feels like total overkill for something that, by very definition, should always result in a very small group of pixels. The two tricky parts are determining equitable size across different resolutions, and figuring out a good fade-in/out function as the target moves across the visibility threshold.

For problem #2, there is an obvious solution — scaling. That's pretty much the purpose for which it was invented to begin with, after all. It also needs to solve the issue where we were looking at a dot as a representation of “smallest thing” that needs to seamlessly transition into “accurate size for what your display is capable of”. Zoom could conceivably do it the detection bit, but we're at ranges where sufficient zoom on its own would most likely also create over-sensitivity to viewpoint jitter that it's hard to keep track of where you are and how your own plane is oriented in relation to the target, which sort of defeats the entire purpose. But it couldn't really handle the dot-to-model transition in a reliable way.

For problem #3, zoom should probably take over as the best solution, possibly aided by scaling, but we're now at ranges where those two need to counteract each other to not make it too easy or look too silly. That's easy enough to handle with some of the later scaling equations.

 

I think a lot of the problems we're having with DCS' solution, and in this thread in general, comes down to the fact that people want to use one tool to solve all three things. Be it an extension of dots to cover the medium-range visibility problems (which they can't really do —they wouldn't be dots then), or just zoom (which can't solve the long-range problem, and is barely adequate for medium-range), or just badly made dots, solely for the long-range problem, but with no consideration for the other two visibility issues. Or just using dots for everything because scaling is verboten for political reasons…

In light of all of that, it's hardly surprising that the complete solution that is most fondly remembered is one that not only uses more than one tool, but does so in such a mix as none of them are even apparent — it just flows together with one taking over where the other no longer works for the problem at hand.

[Why485]

5 hours ago, Tippis said:

There are (at the very least — arguably there could be more) three aspects that a spotting mechanic needs to deal with, each within its own range band.

1. Edge-of-visibility, long-range target detection. When a plane crosses over from BVR with just within WVR range.
2. Medium-to-long range aspect detection and target tracking. When a plane is WVR and you're trying to anticipate where it's going so you can keep track of it and get a suitable pursuit solution on it.
3. Medium-to-short range target identification. When it's time to figure out whether you should shoot or not.

This is a good post, thank you. This is one of the things that I feel like I haven't been getting across. "Spotting" is a complex subject with a lot of nuance but in these discussions it all too often gets turned into cartoon strawmen and people completely talking past each other.

[James DeSouza]

On 1/20/2024 at 8:04 PM, SharpeXB said:

This discussion certainly has run its course. Any further “I can’t see the planes” anecdotes need to be accompanied by tracks and screenshots as well as game settings. Show us an actual screenshot of an “invisible” plane at .38 miles. Let’s see a label floating over empty sky to prove this claim. This is in fact a bug thread so show us an actual bug with documentation. Otherwise it’s apparent that some people just have poor eyesight. 
And if ED can’t arrive at an actual solution then just give us a genuine OFF option that’s server enforceable. Then we can all move on. 

It's not a quote from this game.  It's one of the results from an realworld experiment I mentioned.  I am curious how such a low result came about.

On 1/20/2024 at 8:33 PM, 71st_Mastiff said:

I cant see why you would label everyone with poor eye-sight, when in fact its something to do with setting; Antialiasing, MSA, Resolution size, etc., etc., etc. So the best option is 3D modeling of sizes and renderings. which they used to do back in the day, I don't understand why it cant be done any longer?

They always used to use spotting dots back in the day.  But the excuse back then was because people were rocking 800 or 1280 screens and dots genuinely were the only approach to get practically reasonable spotting distances (though they were of course still ugly).  That isn't the case anymore, which is the point, but people who have become accustomed to having functionally superhuman sight in flight sims because all combat flight sims have kept this mechanic in place in some form or another are averse to having it removed.

[Tippis]

3 hours ago, James DeSouza said:

They always used to use spotting dots back in the day.  But the excuse back then was because people were rocking 800 or 1280 screens and dots genuinely were the only approach to get practically reasonable spotting distances (though they were of course still ugly).  That isn't the case anymore, which is the point,

What would you propose instead, given the issues described above, where we now have the exact opposite problem? Displays are now too high-res and we need a mechanism to ensure that we only see as far as we should, which means we can't rely on simple geometry since that would massively exaggerate how far out things show up. How do you get around that without relying on a dot solution?

[SharpeXB]

12 hours ago, James DeSouza said:

It's not a quote from this game.  It's one of the results from an realworld experiment I mentioned.  I am curious how such a low result came about.

Well that makes sense. A small aircraft nose-on presents a very small silhouette. It could be nearly invisible even at relatively close range depending on the conditions. 

Edited January 22 by SharpeXB

[James DeSouza]

On 1/22/2024 at 4:11 PM, SharpeXB said:

Well that makes sense. A small aircraft nose-on presents a very small silhouette. It could be nearly invisible even at relatively close range depending on the conditions. 

 

But hey, I have it on good authority that a nose on aircraft should be a highly visible blob out to 40km...  hmmmmmmmmm.....

[oCMANo]

Think I partially fixed what had been bothering me.  My issue was I could see the blobs super far out, but in mid to close range it was extremely difficult to make out what they were.  Even in the middle of a WW2 furball in multiplayer it was very hard to distinguish between friend and foe.

For my setup it came down to MSAA 4X vs DLAA.  I'd been playing on DLAA since it was the 'new thing' so thought it would be better.  Nope.  DLAA was adding some 'ghosting' to planes up close so all I saw was a blur going by, and yes motion blurring was off.

Here's a couple of examples of nullwhat passing a Mustang in 4K looked like with DLAA on.  It was like it was rendering a partial 2nd plane almost on top of a single plane.  Second pic is more of what I'd often see, which was beyond frustrating.

null

 

null

Moved back to MSAA 4X and the ghosting was gone and things in general seemed a lot clearer.  Didn't feel like I needed to put glasses on.  Did lose 15-20fps, but as long as it stays over 80 I'm totally fine with that hit if I can see better.

And here's a Mustang Merging with MSAA 4X on.  Only just the change from DLAA to MSAA 4X.

Distance blobs aren't as bad in MSAA but are still blobs.  But at least in medium to close I'm not seeing a blurred target now.  Hope this helps.

Edited January 31 by oCMANo

[Chaffee]

Been following this topic. I'm enough of a n00b at DCS that I'm dead no matter what the spotting distance...

So let's do some science:

At 20/20 vision, the human eye can distinguish something at 1 arc minute in size. Ok, so what does that mean for spotting aircraft?

1 arc minute is 1/21,600th of the circumference of a circle (360 degrees * 60 minutes).

From that very simple number, you can calculate a fundamental limit of visibility (which doesn't tell the whole story, of course, but we'll get to that).

So, an F-16C is 49 feet, 5 inches long (15.06 meters, for those who speak French). Beam on (or plan view), that's 1 arc minute at a range of 27.9 nautical miles (51.6km).

Here's the math:

49.416 feet * 21,600 arc minutes = 1,067,385.6 feet of circumference
1,067,385.6 feet / 3.14159 (pi, to get the diameter of the circle) = 339,759.676 feet
339,759.676 feet / 2 to get the radius of the circle = 169,879.838 feet
169,879.838 feet / 6076 (feet in a nautical mile) = 27.959nm

Can you see a plan-view F-16 at 27.9nm with the naked eye in real life?

No, you can't, (except under some very specific conditions, which I'll mention later), and there's a reason (or really several) for this. If you're still here, let's investigate those:

The first problem is that F-16, even with ~575 square feet visible in plan view, only takes up about 30% of the space of the arc minute at 27.9nm. That means 70% of the light you see at that distance is coming from whatever is behind the F-16. In other words, it's more than 2:1 sky, at the very limit of the resolution of your vision.

The second problem is that, when you double the distance to something, you halve its apparent brightness and quarter its color intensity. That means the F-16 that looked normal to you at 400 meters is 112x dimmer and retains about 0.02% of its apparent color at 27.9nm.

The third problem is something called "aerial perspective." This is what makes far-away objects look bluer to you: atmospheric light scatters, reducing contrast and clarity.

So, at 27.9nm you have an object (the F-16 in plan view) that's only taking up 30% of the space your eye can resolve, at less than 1% of its apparent brightness and 0.02% of its color saturation that it has at the reasonable distance of 400m, and it's blue-shifted behind the scattered light of the atmosphere...

Never mind haze, dust, what your blood pressure is... etc.

At 27.9nm, an F-16 looks exactly like whatever is behind it... usually.

There are 2 things that can change that:

The first is backlighting. Remember all that atmosphere between you and the F-16? If you get the correct angle, with a nice low sun behind it (think dawn or dusk), you can get that F-16 casting a nice shadow through the scattered atmosphere, which makes it effectively much larger and increases contrast. At the correct angle, that little invisible F-16 can cast a shadow several arc minutes long.

The second is light emission. Turn on the landing lights, and you're visible for a very long way -- well beyond the arc-minute resolution -- as a point of light. This, by the way, is why you can see stars at night, even though you cannot resolve the solar disk with a good-sized telescope (you know, like the Hubble). Points of light are visible as far as their brightness will make an impression on the retina, irrespective of resolution limits.

Here are some numbers relating arc minutes to resolution, BTW:

3840x1440 1 arc minute = 1 pixel @ 64 degrees FoV

2560x1440 1 arc minute = 1 pixel @ 42.67 degrees FoV

1920x1080 1 arc minute = 1 pixel @ 32 degrees FoV

Anyway, do with all that what you will. Small fighter jets are unequivocally invisible at 27.9nm except under very specific conditions and, given aerial perspective, haze, color diminishment, etc., they are practically invisible at closer ranges than that.

Some additional trivia:
An F-16 making a knife edge turn in plan view to you at 13.4nm will be 2 arc minutes long by 1 across: it'll actually be dimensional! Even better, its surface area will be reflecting more than an arc minute of light back at you, so now it's only a matter of haze, dust, aerial perspective, apparent brightness, etc. But the situation is massively better.

Head on, that same F-16 only has 1 arc minute of surface area at 4.9nm, however, even if its wingspan is almost 4 arc minutes across. The good news is that most of the other issues of visibility have been solved, and it's now close enough that its countershading/camouflage is useful

 

Edited February 24 by Chaffee

[SharpeXB]

18 hours ago, Chaffee said:

At 20/20 vision, the human eye can distinguish something at 1 arc minute in size. Ok, so what does that mean for spotting aircraft?

Just because something might be technically mathematically barely visible IRL doesn’t justify putting a giant icon or dot on it thus making it very visible. 

[Tippis]

1 hour ago, SharpeXB said:

Just because something might be technically mathematically barely visible IRL doesn’t justify putting a giant icon or dot on it thus making it very visible. 

…and that's why DCS is getting rid of that system and we're getting this much improved one instead. Everyone wins. Except those who get too used to being able to see planes at insane distances and who want it to stay that way.

You're making a good argument why we also can't really rely on pure perspective rendering since that doesn't actually match how perception works.

Edited February 24 by Tippis

[Chaffee]

3 hours ago, SharpeXB said:

Just because something might be technically mathematically barely visible IRL doesn’t justify putting a giant icon or dot on it thus making it very visible. 

I agree. Please don't cherry-pick quotes and use them in an attempt to indicate the opposite of the argument I made. To wit:

 

21 hours ago, Chaffee said:

Can you see a plan-view F-16 at 27.9nm with the naked eye in real life?

No, you can't, (except under some very specific conditions, which I'll mention later), and there's a reason (or really several) for this. If you're still here, let's investigate those:

21 hours ago, Chaffee said:

Anyway, do with all that what you will. Small fighter jets are unequivocally invisible at 27.9nm except under very specific conditions and, given aerial perspective, haze, color diminishment, etc., they are practically invisible at closer ranges than that.

 

Q.E.D.

Edited February 25 by Chaffee

[Chaffee]

2 hours ago, Tippis said:

You're making a good argument why we also can't really rely on pure perspective rendering since that doesn't actually match how perception works.

Correct.

If I were to make a recommendation, it would be that zoom do nothing for targets beyond a certain range. In the above example, an F-16 at 27.9nm would appear to be a sky-colored pixel no matter what the zoom setting/FoV/screen resolution.

That pixel would be invisible for all practical purposes since it's effectively the same color as the background. Rendering this invisible pixel would start at this limit, but only if the player's FoV was set according to the data I gave at the end of that post.

 

Edited February 25 by Chaffee

[SharpeXB]

16 minutes ago, Chaffee said:

I agree

So what’s the point of all the mathematics then? With the spotting dots turned off, very distant aircraft are actually depected quite realistically as it is now. Just make the ability to turn these dots off a mission/server setting and I think we’re done with this problem. 
There’s no use in trying to calculate what should be visible in the game. If you take away the dots what you get is indeed realistic by virtue of the same mathematics. 

Edited February 25 by SharpeXB

[Chaffee]

Just now, SharpeXB said:

So what’s the point of all the mathematics then? With the spotting dots turned off, very distant aircraft are actually depected quite realistically as it is now. Just make the ability to turn these dots off a mission/server setting and I think we’re done with this problem. 

1) To demonstrate the absolute limit of human capacity
2) From this basis, to demonstrate that, under most conditions, nothing can be seen at that absolute limit, so that someone else doesn't use this calculation as evidence that something can be seen through an atmosphere at this limit
3) From this basis, to demonstrate that, when emitting light, resolution limit is no longer a factor (in order to cut off arguments like "why can I see this plane 60 miles away at night IRL")
4) to provide an objective basis for any argument made on this topic

All of that requires a reader to read, however, not to jump off 10% of the way in. I can't control that.

[SharpeXB]

3 minutes ago, Chaffee said:

1) To demonstrate the absolute limit of human capacity
2) From this basis, to demonstrate that, under most conditions, nothing can be seen at that absolute limit, so that someone else doesn't use this calculation as evidence that something can be seen through an atmosphere at this limit
3) From this basis, to demonstrate that, when emitting light, resolution limit is no longer a factor (in order to cut off arguments like "why can I see this plane 60 miles away at night IRL")
4) to provide an objective basis for any argument made on this topic

All of that requires a reader to read, however, not to jump off 10% of the way in. I can't control that.

Well the realistic limits to spotting aircraft are well understood and have been cited many times in these discussions. The trouble is that the preponderance of dots and icons in these games over the years have muddied the understanding of what this is. Again DCS without the dots is close enough. But good luck convincing gamers they shouldn’t be able easily see these things. 

[Chaffee]

1 minute ago, SharpeXB said:

Well the realistic limits to spotting aircraft are well understood and have been cited many times in these discussions. The trouble is that the preponderance of dots and icons in these games over the years have muddied the understanding of what this is. Again DCS without the dots is close enough. But good luck convincing gamers they shouldn’t be able easily see these things. 

I read a lot of this discussion and saw people talking about ridiculous spotting distances with zero evidence or calculation, etc.

But mate, I'm on your side here. The post was written in objective style, making no recommendations. I have some thoughts on the topic (which I've put in later replies), but I find that the objective mathematical details are more convincing, especially for people who might have better ideas than I have.

[Tippis]

On 2/25/2024 at 2:47 AM, SharpeXB said:

Well the realistic limits to spotting aircraft are well understood and have been cited many times in these discussions. The trouble is that the preponderance of dots and icons in these games over the years have muddied the understanding of what this is. Again DCS without the dots is close enough. But good luck convincing gamers they shouldn’t be able easily see these things. 

Again, you realise that the only one who has ever tried to convince anyone that they should be easy to see is you, right? You're were the one who once upon a time offered a screenshot of a plane being visible at 50nm as proof that DCS spotting was fine under the old system. It wasn't until you realised that it gave people advantages over you that counter-balanced the advantage you had over them that you wanted to see it changed.

Now we're getting a new system that is scaling back where that upper limit is; where it starts out being faded completely into the background rather than popping in quite harshly; where it is at least attempting to be hardware-agnostic; and where we know that the upper limits could stand to be scaled back a bit more (although, tbh, it's probably the near limit that is the bigger problem in how the fading happens), even if they are sort-of-reasonable for larger airframes given the right display settings. And with that, you have come full circle to arguing that we should actually go back to that old rendering system. A system that you, for a while at least, decried as being horribly broken. You have even posted new images showing the improvement, only to then immediately argue that it should be made worse, and that somehow, for no clearly explained reason or rationale, this reversal to ridiculous visibility is supposed to be more realistic. These new images show planes clearly visible at 30nm using the old rendering, and now you're trying to say that that's “close enough” to the realistic limits. Only roughly an order of magnitude off.

The realistic limits are well understood, yes. That's why actual numbers have been thrown at you for quite some time now: to demonstrate how your intuitive, yet ever-changing sense of what's supposedly right is actually objectively wrong. Especially on the topic of anything to do with display technology and how it relates to properly and realistically simulating perception, which is the ultimate goal here. And let's not forget, just about every time any kind of research or numbers have been offered you to demonstrate why your intuition is wrong, you have decried it as “silly” or, like now, questioned its relevance…

Edited February 26 by Tippis