Visibility: 1920x1080 vs 2560x1440

[Tippis]

ED already tried “scaling” in the form of Model Enlargement. And although that wasn’t exactly the same thing as Smart Scaling

…aaaaand since it wasn't the same thing as smart scaling, it tells us nothing about the viability or results of a proper scaling method such as Serfoss. The reason model enlargement and impostors failed is because they were bad solutions. A bad solution being bad tells us nothing about how a good solution will fare.

 

The trouble then is nobody is going to be able to agree on how much scaling to use or whether to use it at all.

Fortunately, the research has already answered that question so there's nothing to disagree on.

 

So there’s no way to make it look realistic or believable.

Equally fortunately, the whole point of the methodology is that it looks more realistic and believable than relying on simple trigonometry. That's why it was invented.

 

Target size alone isn’t the most important factor and there are many better solutions.

If target size isn't an important factor, then that's just another argument against implementing proper perception simulation that falls by the wayside. As for better solutions, what would those be that don't actually just make the disparity worse?

[SharpeXB]

Fortunately, the research has already answered that question so there's nothing to disagree on.

If there’s one certainty it’s that nobody here will be able to agree on anything. They’ll want the scaling to be optional or adjustable and then it will cause the same mess as Model Enlargement.

[Tippis]

They’ll want the scaling to be optional or adjustable and then it will cause the same mess as Model Enlargement.

Optional is non-issue. Adjustable is the same as optional since there is a right answer that you can either apply or not. As such, it is also a non-issue. Either way, there's nothing to disagree on.

[SharpeXB]

the whole point of the methodology is that it looks more realistic and believable than relying on simple trigonometry.

Serfoss didn’t have ground units or low flying aircraft that can be seen in context against unscaled environment. That won’t look realistic or believable at all.

[Tippis]

That won’t look realistic or believable at all.

Prove it.

[SharpeXB]

Prove it.

How would aircraft parked on a carrier a mile away look if they’re scaled up 2x?

You would see the aircraft hanging over the edge of the deck. It would look completely stupid. And you’d see it clearly on the big Ultra HD screens people have today.

Edited December 5, 2019 by SharpeXB

[Tippis]

How would aircraft parked on a carrier a mile away look if they’re scaled up 2x?

You would see the aircraft hanging over the edge of the deck. It would look completely stupid. And you’d see it clearly on the big Ultra HD screens people have today.

This is not proof. This is just more things you need to prove.

So, again: prove it.

[NineLine]

I don't think anyone is asking for that

 

Based on this discussion between 3 or 4 people, you can't really make that determination, and if labels were greatly improved, with much more customization options, you would be surprised at how many would want that. People have been reporting posts in this thread because others are discounting their opinions, so we should try and stay away from that. No one person speaks for everyone here, not even me. At this point, it's just a discussion as we really, again, have no intention of pursuing scaling of any sort.

 

Enlargements, in my humble opinion, are ugly, worse than a good label. We don't have a good label in the game now. We have the foundation for it, but it needs many improvements. Better than that is graphical polishing to improve what we have, and to be honest, it's not that bad. I know its bad for some, I have a great set up, and visibility is where I expected, but still needing the tweaks I have mentioned more times now than I can remember. I know as I typed this, more charts, graphs and such are being posted about this scaling system, but I just don't believe it works for what we have... I don't want to see some infantry dude towering over a house... sorry.

 

ED has played with scaling, they are not interested in pursuing it further, so while this discussion is fine, it's not changing any minds. You need to think about what makes something more or less visible in the real world, and what about that is missing or needs improvement from ED, enlarging things isn't one of them.

[NineLine]

This is not proof. This is just more things you need to prove.

So, again: prove it.

 

It seems legitimate to me, how does scaling deal with an object that is next to, or in his example parked on top of something that isnt scaled? Infantry next to a house... do they look bigger than the house? It ends up being not all about an aircraft 2nm out with blue sky behind it. So many many more variables not being accounted for.

Edited December 5, 2019 by NineLine

[Tippis]

Based on this discussion between 3 or 4 people, you can't really make that determination, and if labels were greatly improved, with much more customization options, you would be surprised at how many would want that.

Maybe not as a statistical sample, no, but pretty much every time this discussion comes up, it has nothing to do with the longer ranges (where, if anything, the advocates for change tend to want to see less visibility than we have today) but rather the mid-range to just-out-of-gun-range segment. One curious thing that tends to happen is that the “no change is needed” camp starts rolling out screen shots showing planes appear at absolutely absurd ranges where they should not be visible, and this would prove that spotting isn't broken — it always does the exact opposite. :D

 

Also, I would not be surprised at all at how many would want improved labels. I'm one of them. I think you might be confusing two intertwined but essentially separate discussions there: one is the lack of functionality and even outright buggines of labels that everyone wants to see fixed; the other is the notion that labels are a good solution to spotting.

 

Enlargements, in my humble opinion, are ugly, worse than a good label. We don't have a good label in the game now. We have the foundation for it, but it needs many improvements. Better than that is graphical polishing to improve what we have, and to be honest, it's not that bad.

Graphical polishing only goes so far. No matter what, you end up having to simulate perception and visual cues in ways that simply are not real. The only way not to do that is to wait for universal adoption of hardware and software that isn't even on the horizon at the moment — rending pipelines, graphics cards, and monitors that go way outside the RGB gamut and which produce actual overbrights, not just silly HDR trickery, all that on top of at least double or triple the pixel density we have today (in VR, add an order of magnitude). And relying on hardware just makes the discrepancies between the have:s and have-not:s worse.

 

But here's the funny thing, though: the problem right now is actually the opposite of what you'd expect, where worse hardware in many ways makes spotting easier. The fear some express that solving spotting in a simple way would make the game look bad is actually already the case: by reducing graphical quality, things show up more easily. That is (part of) what the OP was trying to show. This is another flaw that needs to be addressed, and as it happens, the methods for doing so would tie in well with the methods for making spotting more realistic. The displays being used to play games today are more divergent than they have ever been, and that's not likely to go away any time soon. Graphical polish will not be able to gloss over that ever-increasing gap.

 

I don't want to see some infantry dude towering over a house... sorry.

That is not how a proper implementation would work anyway. And at the distances we're talking about, that infantry dude would be all of 4 pixels high. He's not going to tower over anything other than random noise in the ground texture. ;)

 

ED has played with scaling, they are not interested in pursuing it further, so while this discussion is fine, it's not changing any minds.

ED played with the type of scaling that is universally accepted as being a bad solution. There are other solutions that have been empirically proven to be good instead. The two are not the same and conclusions from one don't really apply to the other (especially not ones that the other is specifically designed to change up). If you try to bite into an orange and find it unpleasantly bitter, should that keep you from eating melons (or from trying to peel the orange the next time around)?

[nighthawk2174]

if labels were greatly improved, with much more customization options, you would be surprised at how many would want that.

 

Yes, better and fixed labels would be a good interim solution but ultimately it has some serious flaws. The whole point of smart scaling is to allow you to perceive the apparent orientation of an aircraft at realistic ranges. This is the problem sefross set out to solve. Such things as glint and/or contrasts adjustments to account for pixel color bleed and LOD loading ranges more deal with the ranges you can see targets at and the ability to track them at realistic ranges. Labels though doesn't help with any of this really, not only does it sit over the target hiding its orientation but it would more than likely interfere with any glint, color, and contrast changes you make to try and improve long-range spotting.

 

Enlargements, in my humble opinion, are ugly, worse than a good label.

 

I really don't see how? Have you tried out the scaling demo? From my own observations made over the course of a few days, I've only become more and more impressed with the scaling demo and how realistic it makes stuff look compared to what we have now.

 

 

it's not that bad. I know its bad for some,

 

I disagree strongly I consider DCS to be the worst in this regard and not just for a few people the amount of people who have trouble seeing in this game is quite extensive.

 

I don't want to see some infantry dude towering over a house... sorry.

 

There are several solutions to this:

 

A) Don't scale ground units

B) Have it so the scaling factor takes into account your current FOV aka zoom level and adjust the sizing down. Also, make it so when viewed through TGP's don't scale the unit.

C) Remove the unrealistic level of zoom we have right now. And by the time your close enough to easily see the target, the scaling factor will be low enough it won't be noticeable.

 

Easy solutions making this a non-issue.

 

ED has played with scaling, they are not interested in pursuing it further,

 

What ED did was a straight upscale such as X2 across all ranges and well obviously it was sub-optimal even Sefross points this out in his paper. To small at some ranges and far to large at others.

 

they are not interested in pursuing it further,

 

well then I guess we'll just have to keep trying till they change their minds

 

 

You need to think about what makes something more or less visible in the real world, and what about that is missing or needs improvement from ED, enlarging things isn't one of them.

 

Yes scaling is one of them, sefross did the scientific research for us and scaling is a way to improve not necessarily how easy you can see something (although it helps a lot at the 2-4nmi range)

[NineLine]

If you try to bite into an orange and find it unpleasantly bitter, should that keep you from eating melons (or from trying to peel the orange the next time around)?

 

I still see no information on how the oranges or melons look next to the guy holding them if the oranges and melons are enlarged and the guy isn't.

[David OC]

Not sure how far this is away? You would not want that scaled tho, look funny next to all the houses / buildings. 2560 x 1440

 

Time 7:30

 

 

Edited December 5, 2019 by David OC

[nighthawk2174]

I disagree, again I think your overestimating the amount bigger stuff looks right at gun range.

[SharpeXB]

There are several solutions to this:

 

A) Don't scale ground units

B) Have it so the scaling factor takes into account your current FOV aka zoom level and adjust the sizing down. Also, make it so when viewed through TGP's don't scale the unit.

C) Remove the unrealistic level of zoom we have right now. And by the time your close enough to easily see the target, the scaling factor will be low enough it won't be noticeable.

Ground units and A2G action are a big reason why a Serfoss system isn’t workable for DCS World

 

A) What about airfields where ground and air units are seen together?

B) If the objects scale up and down with the zoom level this creates a visual oddity that was already seen in Model Enlargement. An “inverse zoom” effect where objects don’t get smaller as you zoom out or vice versa. It makes judging range difficult and confusing. It also over enhances distant targets. Plus it would just look odd. As you close with a bandit only 1 mile away you’ll see it change size clearly.

C) The zoom level in DCS isn’t unrealistic. Your original post shows you’re using a 24” monitor. At that screen size in order to see a life sized image you need to be zoomed in almost fully. The problem with your results is that you’re looking at a wide FOV where everything is smaller than life.

DCS already has a tool to make distant objects larger. Zoom View. This is also used to make up for the low resolution of your display and is used in VR for the same reason.

And this feature is not unique to DCS. All flight sims and many first person sims have a zoom view for the same reasons.

[Tippis]

Ground units and A2G action are a big reason why a Serfoss system isn’t workable for DCS World

This has yet to be proven. Where's your proof?

 

B) If the objects scale up and down with the zoom level this creates a visual oddity that was already seen in Model Enlargement. An “inverse zoom” effect where objects don’t get smaller as you zoom out or vice versa.

No, they'll get smaller — just at a different pace. The problem here is that zoom is a highly unrealistic function to begin with so it looking wrong is to be expected. It won't make judging range any more difficult than it already is — indeed, one of the main points of the methodology is that it improves the ability to judge relative position and movement. Since it's a proper continuous function, you have a smooth and unique mapping between distance and size, same as without the scaling, so you won't be confusing one range for another.

 

It might have become a problem for optical gun sights… except that those will tell you to get far closer than the distances where it would become a problem.

 

The zoom level in DCS isn’t unrealistic.

Yes it is. Your eyes can't zoom. The zoom also isn't dependent on screen size, but rather operates on a fixed range. It is not a fixed FoV calculated from actual frustrum size and therefore by very definition not realistic.

[SharpeXB]

This has yet to be proven. Where's your proof?

The proof is self evident. This shouldn’t need explaining.

Visual oddities were even important to Serfoss

“It was believed that the target could be enlarged enough to do this, yet not be so large as to appear unrealistic or give the pilot a false sense of how the target would look in real combat.”

Perhaps on the displays he used he felt that he wasn’t seeing anything unrealistic but on today’s screens you’ll see unrealistic sizes. If you acquire a target at 1 mile and it’s 2x the size and as you close with it, it changes in size by -100%. That’s not realistic in anyone’s definition. Nor is seeing aircraft 2x the size next to objects that aren’t scaled.

 

Yes it is. Your eyes can't zoom.

This feature is universal across all flight sims. It’s there for some very logical reasons

If you don’t like it or can’t get around using it, maybe you should try VR. Although VR sorta needs a zoom view depending on the headset’s resolution, all the issues of size and FOV related to a monitor don’t apply anymore.

[Tippis]

The proof is self evident.

That is an assumption, not proof. Doubly so since you have expressly stated to never have seen this methodology in action. Again, you have no proof. You only have assertions based on admitted wilful ignorance of the topic and an apparent unwillingness to actually investigate it in any way.

 

You'll note that the research concludes that no only is the scaling does not appear unrealistic even as it approaches factors of 2x and 2.5x, but it actually creates a more realistic representation of how pilots empirically perceive targets at those distances.

 

If you acquire a target at 1 mile and it’s 2x the size and as you close with it, it changes in size by -100%.

No. That's not how percentages work, nor how perception of size works.

 

This feature is universal across all flight sims. It’s there for some very logical reasons

Yes. Realism isn't one of them.

Funnily enough, the arguments you bring forth against scaling are actually far more relevant as arguments against zoom, but fortunately, since they are almost universally factually incorrect, zoom is safe for the moment… :P

Edited December 5, 2019 by Tippis

[SharpeXB]

That is an assumption, not proof.

I’m not sure what sort of “proof” you need. We all have this game, go fly one mile away from an aircraft carrier and imagine all the planes on the deck shown at 2x their size. What you’re asking for is going to look extremely odd to the vast majority of players.

 

No. That's not how percentages work, nor how perception of size works.

According to Serfoss’s table a target aircraft is scaled up 2x at 1 mile. That’s really huge and will look unrealistic on a screen. Scaling isn’t needed to see a target at that range, see the attached screenshot. The FOV level in this shot is about life sized.

Then as you close with this target, even at 900 yards which is well within gun range, Serfoss would still apply a magnification of about 40% so you would actually see the target’s scale factor shrink quite rapidly as you close on it. Serfoss applies scaling at any range above zero, so it would still be shown to you at an unrealistic size at gun range where you can easily perceive that and the gun funnel on the HUD would be off by 40%

This doesn’t correlate with his own statement about not wanting pilots to see anything unrealistic.

Edited December 5, 2019 by SharpeXB

[NineLine]

If it is truly 2x at 1 mile, someone with Photoshop could show that real fast... expand a deck full of Hornets x2, leaving the carrier at normal size, unless scaling effects the carrier... then tiny Elvis says, man, look at that carrier, man that carrier is huge.

[Tippis]

I’m not sure what sort of “proof” you need.

Oh, almost anything will do.

Research papers. Correctly calibrated screen shots. Correctly adjusted unit sizes superimposed over existing assets. Pixel-count maths. It's quite easy to do if you just know what you're talking about.

 

What you’re asking for is going to look extremely odd to the vast majority of players

You keep saying so without being able to prove it. Repetition is not evidence.

 

That’s really huge and will look unrealistic on a screen.

Prove it.

 

The FOV level in this shot is about life sized

Really? On what size screen at what distance, and why is it cropped? Also, I think you might be confused about what problem Serfoss scaling is meant to solve…

 

e: But hey, let's assume you're right about both the FoV and the scale factor involved. Let's see what an abomination that would generate. To not break the forums for people, I'll just be kind enough to link it: scale comparison at 0.9nm. To check that you're at the right FoV, put the monitor at such a distance that the wingspan is is 6 mils across (just under 11 mils in the scaled frame) — i.e. that at 60cm for a classic laptop slump, it is just under 4mm across, and/or adjust your browser zoom to compensate.

 

Then as you close with this target, even at 900 yards which is well within gun range, Serfoss would still apply a magnification of about 20%.

It may be inside gun range but it is not be inside optical sight distances so that small adjustment will not throw off how large the target should be to fit within the gun reticle of the planes that rely on such sighting methods. They'll be looking at 3–400m, where the factor is so small as to not make any useful difference for the purpose of determining (via the sight) that they're in range.

 

So you would actually see the target shrink quite rapidly as you close on it.

Again, no, you wouldn't. You know that video you snidely linked earlier about being small vs. being far away? That's exactly the thing you're confusing here. As the target approaches, its apparent size will grow larger. Meanwhile, as it approaches, the scale factor will start to approach 100%, but at a much slower rate than the apparent size increase.

 

So no, you would not see the target shrink. It would just not grow large as quickly. That's a very different effect.

 

This doesn’t correlate with his own statement about not wanting pilots to see anything unrealistic.

…but as the research proved, they don't see anything unrealistic. That is the whole point of the thesis and is the reason why this method is suggested over the more simplistic methods like the one attempted in DCS in the past. This conclusion is further affirmed by a guy titled “United States Air Force Reserve Simulation Director” — someone who probably knows a thing or two about how simulations work…
Edited December 5, 2019 by Tippis
metric vs imperial is hard XD

[SharpeXB]

Oh, almost anything will do.

Research papers. Correctly calibrated screen shots. Correctly adjusted unit sizes superimposed over existing assets. Pixel-count maths. It's quite easy to do if you just know what you're talking about.

You can’t just look at your screen and tell what realistic looks like? Do I need to photoshop it for you?

 

It may be inside gun range but it is not be inside optical sight distances so that small adjustment will not throw off how large the target should be to fit within the gun reticle of the planes that rely on such sighting methods.

It’s still off. And since Serfoss applies a scale factor to anything over 0’ At 880 yards (.5 mile) it’s 40% enlarged (I was wrong before, it’s worse!), at 440 yards it’s 20% and at 220 yards it’s 10%. At 110 yards it’s still 5% too big. At WWII gun combat range the target aircraft is still 10% too big! Please don’t say you think that’s realistic.

 

As the target approaches, its apparent size will grow larger. Meanwhile, as it approaches, the scale factor will start to approach 100%, but at a much slower rate than the apparent size increase.

...

So no, you would not see the target shrink. It would just not grow large as quickly. That's a very different effect.

Corrected, it’s the scale factor that gets smaller. But the bottom line is it will look odd and unrealistic.

 

This conclusion is further affirmed by a guy titled “United States Air Force Reserve Simulation Director” — someone who probably knows a thing or two about how simulations work…

The world is full of “scientific thesis” which are wrong. And let’s start a new thread about failed Air Force projects...

And DCS is not a military trainer (some of EDs products are and they akaik don’t even use Serfoss scaling) Have you seen the graphics in the average military computer simulator? Those aren’t entertainment products like this is.

[Tippis]

Do I need to photoshop it for you?

That would definitely be something that could illustrate your opinion.

 

It’s still off. And since Serfoss applies a scale factor to anything over 0’ At 880 yards (.5 mile) it’s 40% enlarged

Eh, no. At 0.43nm, the scaling factor is somewhere between 1.0 and 1.2. Or are you deliberately reading from the wrong graph just to inflate the values for some reason?

 

Corrected, it’s the scale factor that gets smaller. But the bottom line is it will look odd and unrealistic.

…aaaaaaaaand again (and again, and again, and again) you have yet to prove this. The research proves you wrong and you have not been able to contradict it, disprove it, or in any way cast doubt on its veracity. If you want to prove yourself right again rather than just rely on some association fallacy, that is what you have to do: you need to provide your own research. “Other research has been wrong” is not even an argument — it's the explanation why you have to put in the effort here.

 

And DCS is not a military trainer

Even more reason to implement methodologies that counteract or smooth out the many differences that can be seen among the myriad of display systems used for this game. Now granted, just because it doesn't need the precision and the levels of realism that a military trainer would require, there's no reason not to strive for high realism anyway. If a specific methodology can be used to improve the realism of a trainer, why shouldn't it be applied to a sim that strives for realism as well? The argument that it shouldn't because it's not “the real deal” just reeks of laziness and doesn't belong in DCS, tbh.

[nighthawk2174]

I’m not sure what sort of “proof” you need. We all have this game, go fly one mile away from an aircraft carrier and imagine all the planes on the deck shown at 2x their size. What you’re asking for is going to look extremely odd to the vast majority of players.

 

 

According to Serfoss’s table a target aircraft is scaled up 2x at 1 mile. That’s really huge and will look unrealistic on a screen. Scaling isn’t needed to see a target at that range, see the attached screenshot. The FOV level in this shot is about life sized.

Then as you close with this target, even at 900 yards which is well within gun range, Serfoss would still apply a magnification of about 20%, so you would actually see the target’s scale factor shrink quite rapidly as you close on it. Serfoss applies scaling at any range above zero, so it would still be shown to you at an unrealistic size at gun range where you can easily perceive that and the gun funnel on the HUD would be off by 20%.

This doesn’t correlate with his own statement about not wanting pilots to see anything unrealistic.

 

Ok lets work this.

 

* Zoomed out is 90° FOV

* zoomed in is 40.8° FOV

 

Default:

 

 

 

 

Using raw Sefross:

----------------------

Zoomed out - factor:

1.505

 

Zoomed in - factor:

1.505

 

Using Chihirobelmo:

--------------------

Zoomed out - factor:

1.428

 

Zoomed in - factor:

1.126

Using Snapat V2:

--------------------

Zoomed out - factor:

1.591

 

Zoomed in - factor:

1.000 - Would look like the default pic

[SharpeXB]

Eh, no. At 0.43nm, the scaling factor is somewhere between 1.0 and 1.2. Or are you deliberately reading from the wrong graph just to inflate the values for some reason?

I’m looking at this chart here, the blue Cumulative Magnification line. What’s crazy is he ends up with the largest value from orientation (4) with the largest profile. How does that make sense?