PDA

View Full Version : Something to consider re. dot size



Lurch1962
03-09-2007, 06:12 PM
Sorry to begin a new thread, but after 5 pages threads get too cumbersome to wade through. So, here's some BRIEF notes to ponder...

1) Let's assume a monitor dot pitch of 0.28mm.

2) Say your screen is 15 inches (380mm) wide, and is running at 1600x1200. That's 0.24mm/pixel -- close to the dot pitch.

3) Let's take the resolution of the eye in bright light as being 1 arcminute (1/60th of a degree).

As seen from a distance of 963mm (38 inches) your the screen's array of pixels will be just resolved by the eye.

A fighter with a wingspan of 10m will occupy:
1 pixel at 34.4 km
2 pixels at 17.2 km
4 pixels at 8.6 km
8 pixels at 4.3 km

Gotta run...

--Lurch--

Lurch1962
03-09-2007, 06:12 PM
Sorry to begin a new thread, but after 5 pages threads get too cumbersome to wade through. So, here's some BRIEF notes to ponder...

1) Let's assume a monitor dot pitch of 0.28mm.

2) Say your screen is 15 inches (380mm) wide, and is running at 1600x1200. That's 0.24mm/pixel -- close to the dot pitch.

3) Let's take the resolution of the eye in bright light as being 1 arcminute (1/60th of a degree).

As seen from a distance of 963mm (38 inches) your the screen's array of pixels will be just resolved by the eye.

A fighter with a wingspan of 10m will occupy:
1 pixel at 34.4 km
2 pixels at 17.2 km
4 pixels at 8.6 km
8 pixels at 4.3 km

Gotta run...

--Lurch--

AKA_TAGERT
03-09-2007, 06:27 PM
that summary is 'spot' on

DGC763
03-10-2007, 01:22 AM
Nice post.

Now what is in low light (times other than midday and different backgrounds), lookin towards brighter objects (sun), looking at object the same colour as the ground (camoflague). What is the focus area of your eye? What portion of the screen in normal and wide view would be under direct focus? Should we only have close to zoomed in view (but then we lack periferal). etc etc etc.

grifter2u
03-10-2007, 03:31 AM
good post Lurch1962 http://forums.ubi.com/groupee_common/emoticons/icon_smile.gif

Lurch1962
03-11-2007, 05:16 PM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Now what is in low light (times other than midday and different backgrounds), lookin towards brighter objects (sun), looking at object the same colour as the ground (camoflague). What is the focus area of your eye? What portion of the screen in normal and wide view would be under direct focus? Should we only have close to zoomed in view (but then we lack periferal). etc etc etc. </div></BLOCKQUOTE>

The eye's resolving power depends on both overall scene brightness and contrast (brightness differences). From daylight to nighttime, resolving ability decreases pretty much along a continuum.

To illustrate just how much lower it can be, consider this (from actual experimentation conducted during WW2). At the lowest illumination levels and contrast thresholds detectable by the eye, an object must subtend an apparent angle on the retina of 5 or 6 *degrees* in order for the visual cortex to discern it from the "noise." But this is an extreme case, applicable to situations such as being in a large, darkened warehouse having only a feeble exit sign for illumination coming from the opposite side.

In a simulation, one can't come even close to representing the real range of brightness between day and night on any monitor in existence. To wit: the brightness difference between sun-lit ground and that same ground at night is about 20 astronomical magnitudes, or a factor of 100 million! Your monitor can display only a tiny slice of this brightness range, somewhere towards the brighter end. (Even a "black" screen is quite bright!)

About camouflage... even under bright light it's effective in making the target difficult to see because of lowered contrast. More important than matching the colour of camo paint to the ground and sky backgrounds is the matching of TONES.

The region of sharpest vision occurs on the central portion of the retina called the fovea. In angular extent it's not much more than a degree wide, so as a result your eye is constantly moving so as to bring the object of interest to that very confined point. No matter the screen size, resolution or field of view setting in-game, you'll only be scrutinizing a rather small area in order to see detail. As a test, try reading this text without moving your eye from &lt;this point&gt;. You'll find that you must always scan the words so that they pass across your fovea.

Lastly, I should have mentioned before the role that the FOV (field of view) setting has. In the first post I illustrated a more-or-less typical monitor set-up. To reiterate, a 15-inch wide screen running at 1600x1200, viewed from a distance of 38 inches (so as to about match the pixel size/dot pitch with the eye's resolution).

The screen's apparent width would therefore be 22.3 degrees. Even the most zoomed-in FOV is a wider 30 degrees. In such case, to match the simulated view with a real life view, the FOV setting would have to allow an even smaller angle of 22 degrees.

To go back to the short table of pixel sizes I presented earlier. Those values apply to the monitor situation as described, and for a true-to-life-size representation of a 10m wingspan A/C.

They would apply also if Il2 offered the corresponding 22 degree FOV setting. But seeing as the narrowest FOV is 30 degrees, multiply those values by 0.73. If the game's FOV is widened to 90 degrees, then the pixel sizes as listed must be multiplied by 0.24. So at 4.3 km the wingspan in pixels would be only 2 (instead of 8 at a 22 degree FOV)

So the most important point of all this is:

WE ARE VIEWING OUR SIMULATED WORLD AS THROUGH A TELESCOPE FROM THE WRONG END!

For a typical eye-to-monitor distance, our view of the Il2 world is "shrunk", even when the FOV is zoomed-in to the max.

That's why I'm such a strong advocate of auxiliary optics such as fresnel lenses and strong reading glasses. By getting in closer to the screen and magnifying it we can better match the view of our simulated world to what we would experience in reality.

--Lurch--

Tipo_Man
03-12-2007, 03:28 AM
So here is the question... how many pixels should this plane take? You have all the parameters in order to solve the equation, I'm playng on 19" LCD at 1024x768...

http://i18.photobucket.com/albums/b128/TipoMan/inv21.gif

My supposed answer is it should take two fifths from the gunsight reticle.

grifter2u
03-12-2007, 06:40 AM
Lurch1962,

an excellent post there, you obviously know a lot about human vision. there is interesting and relevant new information in what you posted.

with the last point you make about the 15 inch wide monitor for il2 i would see that a little different in practise, but your points are valid. the way i see it:
- the 15 inch (38 cm) wide monitor being viewed from 38 inches (96.5 cm) is probably not realistic for normal windows use(your basically describing a 19 or 20' lcd), but i understand the point you are making. since 17' to 22' lcd's have roughly the same pixels per inch,and viewing distance is determined by the ability to read 96 dpi fonts, the viewing distance can be roughly the same (60 cm is not a bad ballpark).
- when you use the closer viewing distance from the monitor, this brings it to roughly a 35 or 40 FoV for il2, which is available in il2 but very unrealistic to fly in constantly, or to dogfight in.

i havnt seen fresnell lenses, but to fly il2 on a normal view (70) you need one heck of a big monitor, about a 30 inch one. sadly at that size the dots representing medium and far away aircraft become nearly invisible on those big monitors because they are high resolution, so in trying to solve the view problem you create another.

Lurch1962
03-12-2007, 07:08 PM
Tipo_Man,

Bit of a puzzle!

You didn't say what your FOV setting was, but as an exercise I'll calculate a possible value.

Moreover, it would also have been nice to screen-capture the Yak against the sky so that it could actually be seen. In this way I could verify the angular size of the Revi gunsight reticle (which I can't find info on here at the shop.)

OK, let's proceed on the assumption that the Revi gunsight's reticle circle is 3 gradians, or 2.7 degrees in diameter (assuming each tick mark is 0.5 grad).

To find a possible FOV setting...
=================================
Based on a reticle diameter of 2.7 degrees, which in your screen shot is 50 pixels across, I get an image scale of 18.5 pixels/degree. At a screen width of 1024 pixels, that suggests a FOV of 55.4 degrees. But because Il2 uses a gnomonic projection, the image stretching off-axis means that the actual FOV is more likely a bit smaller--say, 50 degrees. (I could calculate the ratio, but I'm too lazy at the moment.)

To derive the Yak's wingspan in pixels...
=========================================
With a 10m wingspan and distance of 490m, the angular span would be 1.17 degrees. So at an image scale of 18.5 pixels/degree, the plane's span should be 21.6 pixels. As a fraction of the gunsight reticle diameter, that would be 0.432, or just over 2/5--as you'd determined!

Of course, if the gunsight reticle angular diameter differs from my assumption, a re-calc would be necessary...


Below are some handy relations applicable to small angles (object size &lt; ~20 degrees, and FOV &lt; ~60 degrees). The examples are the derivations of the preceding...

=============================

angular size = TAN^-1 (obj. size / distance)
[TAN^-1 means "inverse tangent"]

ang_size = TAN^-1 (10m / 490m)
= TAN^-1 (0.02)
= 1.169 degrees
=============================

Obj. size in pixels = ang_size / FOV * scrn_width_px

obj_pix = 1.17 / 55 * 1024
= 0.021 * 1024
= 21.8
=============================

--Lurch--

Tipo_Man
03-14-2007, 06:15 AM
Thanks for the well put reply Lurch.
Really very informative.
I'm impressed you guessed the FOV picture was taken with.

Actually my proposed answer was derived from the information that a fighter should fit into the REVI's reticle at 200meters...

The fact that it is invisible at that range is actually my concern, since Oleg should really find a way to draw these at least 18 pixels !!!

Lurch1962
03-14-2007, 06:19 PM
Tipo_Man,
I thought the (green camo?) Yak fighter was invisible because of its being located against the green forest background, which, if the tones were similar, would lead to apparent invisibility.

I know that at 490m I can quite well see any fighter at my end, and at the expected size on-screen.

Perhaps you can try another screen capture with a plane against the sky. Then we'll know with certainty if all's OK or not.

--Lurch--

grifter2u
03-14-2007, 07:06 PM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Obj. size in pixels = ang_size / FOV * scrn_width_px

obj_pix = 1.17 / 55 * 1024
= 0.021 * 1024
= 21.8 </div></BLOCKQUOTE>

lurch,

a minor point. shouldnt screen width be 1280 instead of 1024 ? thats about a 25% size difference for pixels then. i think 5:4 screens at 1280 x 1024 are always wider then they are high, same as 3 x 4 screens (like 1200 x 1600).

and when you calculate the pixel size for the object like you just did, do you find it corresponds on screen in il2 to the size it would be in real life observations ?

Lurch1962
03-14-2007, 08:45 PM
grifter,
I specified "1024" 'cause that's what Tipo_Man said his resolution was set at.

And this is the first time I did such a computation, and only 'cause I was asked by Tipo--I've never had reason to think that in-game objects were drawn at unrealistic sizes. The only issue as I see it is how distant A/C are handled when they get down to the single- and sub-pixel regime.

On the subject of unrealistic size, did you ever play Combat Flight Simulator (The original, not 2 or 3)? From the cockpit, other planes (and ground objects) are drawn at something like 4 times normal size, while from an external viewpoint they are drawn at correct size! And one of my old time faves, Chuck Yeager's Air Combat, drew objects at 8 times the real size!! (This was confirmed during a conversation with the developer, Brent Iverson. Imagine... fighters with 250 foot wingspans! http://forums.ubi.com/images/smilies/16x16_smiley-very-happy.gif

Lurch1962
03-14-2007, 09:44 PM
Mein Herren,

(The following is a repeat of a post in the related thread, "Dot size in high resolution LCD")

I'd surely like to have arrayed before me a whole bank of monitors of various size and resolution. A good test would be to display the very same scene showing a line of planes strung out over a wide range of distance, to see how each setup handles distant A/C dots.

I think an important aspect to consider is the fidelity with which a given monitor will display a single image pixel.

The 5 year old Dell 19" UltraSharp LCD I'm using here in the shop, when viewed with a direct measuring microscope, has a screen whose picture elements are composed of 3 sub-elements--one each for red, green and blue--which together make a 0.25mm square pixel. So right there you have one source of image infidelity because in order to assign a colour to a pixel, the three sub-elements have to be illuminated to varying degrees. For example, a pure red pixel will only have its red sub-element illuminated; green and blue will be dark.

I'm pretty sure the game's rendering engine is not quantizing the calculated locations of objects or polygon vertices so as to put (or force) them exactly on any given screen pixel. The dot (actually 2 pixels--light above and dark below, if I recall correctly) representing a distant A/C will more often than not be shared by several pixels. Anti-aliasing will then apply an appropriate "blending" so that the effect will still be of a single pixel-sized dot, not one much larger.

Running software at a resolution which isn't a match to the monitor's native resolution can introduce problems at the pixel level. Again, anti-aliasing comes to the rescue here, but I'd think it's less than ideal.

--Lurch--

grifter2u
03-16-2007, 11:35 PM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Lurch1962:
I specified "1024" 'cause that's what Tipo_Man said his resolution was set at.

And this is the first time I did such a computation, and only 'cause I was asked by Tipo--I've never had reason to think that in-game objects were drawn at unrealistic sizes. The only issue as I see it is how distant A/C are handled when they get down to the single- and sub-pixel regime. </div></BLOCKQUOTE>
you are right, i didnt look at that earlier part.

do you know a simple way to calculate what a real plane (for ex 10 meter wide) would look like to the naked eye at various distances (for ex 300 or 500 meters)? to me some of the sizes of the various LoD models in il2 look a bit odd at times.

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Lurch1962:
I'd surely like to have arrayed before me a whole bank of monitors of various size and resolution. A good test would be to display the very same scene showing a line of planes strung out over a wide range of distance, to see how each setup handles distant A/C dots.

I think an important aspect to consider is the fidelity with which a given monitor will display a single image pixel. </div></BLOCKQUOTE>

one would hope that Oleg and Co would have tested for that exact issue, but with the game engine being about 7 years old they probably didnt have the monitor hardware, and might not even have considered some of the dot size issues we are dealing with now. the most important thing is that the issue is identified and can be dealt with better in BoB. they probably took a coding shortcut to represent distant aircraft as 2 dots, and never could foresee the problems this would create later on huge 30' monitors.

personally i think they should be able to fix the issue about dot size in 1946 as well, but if no coders are working on this anymore it might be to late. sadly the dot issue is very important for high end users, and it makes the game very difficult to play in a fair way online (unless you set your lcd to a non native resolution, which makes it look very ugly)

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Lurch1962:A good test would be to display the very same scene showing a line of planes strung out over a wide range of distance, to see how each setup handles distant A/C dots. </div></BLOCKQUOTE>
i think the mission builder can be used to create that, but i dont know how to do it myself. a string of airborne 109's could be created,at 100 meter intervals from 100 to 3500 meters, that would make it easy to see at what range the visibility issue becomes a problem.

Lurch1962
03-19-2007, 09:54 PM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">do you know a simple way to calculate what a real plane (for ex 10 meter wide) would look like to the naked eye at various distances (for ex 300 or 500 meters)? to me some of the sizes of the various LoD models in il2 look a bit odd at times. </div></BLOCKQUOTE>
Grifter,

If you're asking to simulate what the eye would actually *see*, and aren't asking about simply cranking out angular sizes (which can be hard to visualize), here's what to do:

Make, in effect, a scale model!

1) Print out or draw a small picture of a plane, in whatever orientation you chose.
2) Measure the appropriate dimension of your "model", and note the corresponding measurement of its real-life counterpart. From this you get the scale.
3) Place your "model" at the appropriate scale distance to see what it would look like.

Example: Your little plane with its 10m wingspan is 2cm across. So the scale is 2cm : 10m, or 2 : 1000, or 1 : 500.

To see what it would look like at 2km, place it 2km (2000m) / 500 = 4m distant from your eyes.

I'll repeat here the formula to derive angular size:

=============================

angular size = TAN^-1 (obj. size / distance)
[TAN^-1 means "inverse tangent"]

ang_size = TAN^-1 (10m / 490m)
= TAN^-1 (0.02)
= 1.169 degrees
=============================

--Lurch--

M_Gunz
03-19-2007, 10:45 PM
Lurch, the zoomed in FOV is supposed to be "as real". Might I suggest that you figure out
how far back the players eye should be from the screen using that? Not that it matters since
screen resolution may differ and we do change FOV without moving our chairs.

At zoomed full out, less than 90 deg last I looked, everything is supposed to be 1/4 size though
I think that makes the ship models just a bit peewee'd -- just IMO.

When you fly sometime, watch the ground as you gain altitude. It changes just by how much air
and how dense/wet/dusty that air is. When I am at 8,000 ft AGL a plane that would blend well
500 ft or less AGL won't blend nearly as much at 3000+ ft AGL over the same ground from where
I am. Objects seen through much lower air loose color and definition, the closer plane will
look sharp by compare and yet can be hard to find at least to untrained eyes. It is the
movement that counts.

Night fighter and night bomber crew training segments have been posted on the web for over 10
years now. At night and other low light, low contrast situations the trained people say to
not look at center of view but let your eyes roam and pick up first on the edges of vision.
It's the rods that catch the object in the dark and eye nerve wiring that brings movement
view to bear but in low light you don't want to look direct until you get quite close.

Lastly, you might benefit from talking to an optician about wearing strong reading glasses for
however long, it is not really a good idea at all. How do I say that? One of my brothers was
an optician and had a lot to say on the subject. But they are your eyes.

grifter2u
03-20-2007, 03:28 AM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Lurch1962:
If you're asking to simulate what the eye would actually *see*, and aren't asking about simply cranking out angular sizes (which can be hard to visualize), here's what to do: Make, in effect, a scale model! </div></BLOCKQUOTE>

there is a simple formula to calculate this, but i cant remember what it is, school time is just to many years ago. the greeks solved this 2500 years ago, making scale models isnt accurate enough and would be to time consuming for our current purpose.

since you knew enough to remember that resolution of the eye in bright light is about 1 arcminute (1/60th of a degree), i thought you might have worked with those issues in real life, or remembered enough from our schooldays.

K_Freddie
03-20-2007, 06:49 AM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Lurch1962:
...
The region of sharpest vision occurs on the central portion of the retina called the fovea...
--Lurch-- </div></BLOCKQUOTE>

AFAIK, this region exists around the fovea, as the fovea is populated by 'colour' sensors which are larger in diameter than the 'black/white' sensors.
I cannot remember the exact names of the sensors, but the black/white one are compacted closer together than the fovea ones, giving higher resolution descernibility.

One uses the technique of looking a degree or so, to the one side to discern lower contrast objects.
http://forums.ubi.com/groupee_common/emoticons/icon_cool.gif

grifter2u
03-20-2007, 07:43 AM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by M_Gunz:
Lurch, the zoomed in FOV is supposed to be "as real". Might I suggest that you figure out
how far back the players eye *should* be from the screen using that? Not that it matters since
screen resolution may differ and we do change FOV without moving our chairs.

At zoomed full out, less than 90 deg last I looked, everything is supposed to be 1/4 size though
I think that makes the ship models just a bit peewee'd -- just IMO. </div></BLOCKQUOTE>

gunz, for each monitor with a particular screen size, you can calculate the field of view it represents (depending on how far away you sit from the monitor). when you then set your il2 to display this FoV, you *should* be seeing all objects 100% in their correct sizes. some examples are:
for a 19' lcd, with you sitting 60 cm away from the screen, it represents a 35 degree FoV
for a 30' lcd, with you sitting 1 meter from the screen, it represents a 70 degree FoV

this is usefull for the 30' lcd owner, because flying most of the time in il2 with a 70 degree field of vision gives you some peripheral vision. when dogfighting you also are able to see the aircraft near you in their correct sizes, so you have a much better impression of their distance from you.

for the 19' lcd, flying in a 35 degree FoV is very disorienting because it is as if you are looking only at a small part of the area around you, as if looking through a narrow tube (basically giving you tunnel vision, without any peripheral information). when you use a wider view, like 70 or 90 FoV, the objects around you shrink in size very significantly, around 100 or 150%. you then have the joy of trying to dogfight with a set of mini planes, and you dont have a good sense of their distance from you.


additionally, when you use the wider FoV on a smaller monitor (like a 19 or 20' lcd), some distant dots representing aircraft, for ex at 3600 or 4500 meters away from you, can become totally invisible. some screenshots posted in the first couple of pages of the other thread illustrated that well.ideally you would want to fly most of the time in a view that is "normal" for your monitor setup, so you can see il2 objects realistically.

LEXX_Luthor
03-20-2007, 10:59 AM
One thing that can help in the FOV transitions would be smooth and fast FOV changes. StrikeFighters offer this, although hopelessly slow and NOT helpful out of the box, it can be increased by the player to very fast FOV changes. One may change FOV continously very rapidly, while keeping focused on the target.

grifter:: <BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">one would hope that Oleg and Co would have tested for that exact issue, but with the game engine being about 7 years old they probably didnt have the monitor hardware, and might not even have considered some of the dot size issues we are dealing with now. the most important thing is that the issue is identified and can be dealt with better in BoB. they probably took a coding shortcut to represent distant aircraft as 2 dots, and never could foresee the problems this would create later on huge 30' monitors. </div></BLOCKQUOTE>
Oleg knew the exact issue, and he fixed the dots with FB 3.0 (the Pacific Fighters releace), allowing each player (or server) to make dots visible for all customers through using the mp_dotrange setting, but then he deliberatly porked it in, as either carguy or fordfan notes, in the fastest Patch releace to date, by crippling the dot's response to the mp_dotrange setting.

For the first time in the history of Forgotten Battles, we saw a brief flowering of No-Icon servers, as Online players began to enjoy the freedom from Oleg Maddox's Arcade Text Icons painted across the Online sky. These servers wilted and died after the next Patch when Oleg crippled the mp_dotrange and dot relationship.

I still think Oleg crippled his sim in response to a few loud Online players wanting everybody to play at low resolutions or with Text Icons for competitive Dogfight advantage, however, Oleg poasted here that he worked with "real life" pilots who claimed the tiny invisible dots were "realistic." But, Oleg naturally did not give the monitor resolution used by these pilots. I assume the pilots were using 1024x768 which the dots were originally designed for, and work quite well. My theory is that in Russia, gaming equipment is a few years behind the west. One thing to note, in the "real life" pilot thread, the "real life" pilots claim that the tiny dots are not realistic. http://forums.ubi.com/images/smilies/51.gif Hence my assumption that the western pilots are gaming with higher resolution than the Russian pilots, until Oleg provides that resolution information -- maybe they used Oleg's high tech equipment?? I don't know, but something's not matching up here.

Lurch1962
03-20-2007, 06:19 PM
M_Gunz,

The formula to derive the "correct" distance at which to place your eyes in order to see the view as life-size:

(NOTE: screen_width is NOT the diagonal dimension! It's the horizontal width of the actual viewable area. IL-2 bases the FOV on this dimension.)

view_distance = (screen_width/2) / (TAN[FOV/2])

Example: Monitor's screen width = 16 inches.
IL-2 FOV = 60 degrees.

view_distance = (16/2) / (TAN[60/2])
= 8 / (TAN[30])
= 8 / (0.577)
= 13.9 inches

"Correct" viewing distances for a selection of screen widths:
=====================================
FOV___14"____16"____20"____24"____30"
-------------------------------------
90____7______8______10_____12_____15
80____8.3____9.5____11.9___14.3___17.9
70____10_____11.4___14.3___17.1___21.4
60____12.1___13.9___17.3___20.8___26
50____15_____17.1___21.4___25.7___32.2
40____19.2___22_____27.5___33_____41.2
30____26.1___29.9___37.3___44.8___56
=====================================

Note that at a FOV of 50-55 degrees, in order to see a life-size view one's eyes must be as close to the screen as it is wide.

The reason for the blurring or "hazing" out of the landscape as seen from high altitudes is due the way the display engine renders the ground/water. It's not meant to represent haze/fog/dust, etc. (except in the case of morning fog); it's a method used to simplify drawing the earth in order to keep frame rates higher.

The low light technique of not looking directly at an object but instead some distance away is called "averted vision." We astronomers use it all the time, I can tell you. http://forums.ubi.com/images/smilies/blink.gif

Improperly used, strong reading glasses can possibly lead to a tendency for the eye's own lens-focusing muscles to relax overly much. I use such glasses only so that I can bring the screen into focus from rather close distances. It's the same case as when I'd like to read really fine print comfortably. Moreover, I'm quite attuned to my eyes and can to a certain extent self-diagnose. To illustrate, before I ever went to an opthamologist some 10 years ago, I knew already the extent and orientation of my astigmatism.

--Lurch--

Lurch1962
03-20-2007, 06:47 PM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">there is a simple formula to calculate this, but i cant remember what it is, school time is just to many years ago. the greeks solved this 2500 years ago, making scale models isnt accurate enough and would be to time consuming for our current purpose.

since you knew enough to remember that resolution of the eye in bright light is about 1 arcminute (1/60th of a degree), i thought you might have worked with those issues in real life, or remembered enough from our schooldays. </div></BLOCKQUOTE>

Grifter,

As to the first point, earlier in this thread I supplied a formula to calculate the apparent angular size--in degrees--of any object of known dimensions lying at any known distance. If you desire output in arcminutes instead of degrees, multiply the formula's result by 60.

A scale model need be nothing more than a simple silhouette cut out of paper (a printout of a plane's image?). Takes no more than a few minutes, and has the advantage of allowing you to truly appreciate the relative dimensions involved (plane size vs distance). And your long "ruler" can be a piece of string with tape bits placed at suitable intervals to scale.

--Lurch--

Lurch1962
03-20-2007, 07:06 PM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">AFAIK, this region exists around the fovea, as the fovea is populated by 'colour' sensors which are larger in diameter than the 'black/white' sensors.
I cannot remember the exact names of the sensors, but the black/white one are compacted closer together than the fovea ones, giving higher resolution descernibility.

One uses the technique of looking a degree or so, to the one side to discern lower contrast objects. </div></BLOCKQUOTE>

K_Freddie,

It's not the *size* of the colour cones and B/W rods which determine their ability to resolve detail, but instead their density on the surface of the retina.

The cones are crowded very much more densely in the region of the fovea (centre of vision), which is why one needs to place an object there in order to see sufficient detail. Cones are also distributed all over the retina, but much less densely in the peripheral regions.

The rods, on the other hand, are more numerous than the cones in the peripheral regions of the retina, and moreover are very much more sensitive to light. There are a number of reasons why peripheral vision is less sharp than foveal vision, but one cause is the insufficient number of nerve axions available for the number of rods. This is ameliorated by the way in which groups of rods are linked to an axion--now a larger unit of rods can contribute their signal, but at the expense of sharpness, or ability to resolve. But this results in an improvement in signal to noise.

In low light situations one uses the so-called technique of "averted vision" to detect a signal which is too faint for the cone-dominated fovea. The best angular distance at which to direct one's gaze from the object of interest is more like 20 degrees.

--Lurch--

grifter2u
03-20-2007, 07:53 PM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Lurch1962:
M_Gunz,

The formula to derive the "correct" distance at which to place your eyes in order to see the view as life-size:

(NOTE: screen_width is NOT the diagonal dimension! It's the horizontal width of the actual viewable area. IL-2 bases the FOV on this dimension.)

view_distance = (screen_width/2) / (TAN[FOV/2])

Example: Monitor's screen width = 16 inches.
IL-2 FOV = 60 degrees.

view_distance = (16/2) / (TAN[60/2])
= 8 / (TAN[30])
= 8 / (0.577)
= 13.9 inches

"Correct" viewing distances for a selection of screen widths:
=====================================
FOV___14"____16"____20"____24"____30"
-------------------------------------
90____7______8______10_____12_____15
80____8.3____9.5____11.9___14.3___17.9
70____10_____11.4___14.3___17.1___21.4
60____12.1___13.9___17.3___20.8___26
50____15_____17.1___21.4___25.7___32.2
40____19.2___22_____27.5___33_____41.2
30____26.1___29.9___37.3___44.8___56
=====================================

Note that at a FOV of 50-55 degrees, in order to see a life-size view one's eyes must be as close to the screen as it is wide.

The reason for the blurring or "hazing" out of the landscape as seen from high altitudes is due the way the display engine renders the ground/water. It's not meant to represent haze/fog/dust, etc. (except in the case of morning fog); it's a method used to simplify drawing the earth in order to keep frame rates higher.

The low light technique of not looking directly at an object but instead some distance away is called "averted vision." We astronomers use it all the time, I can tell you. http://forums.ubi.com/images/smilies/blink.gif

Improperly used, strong reading glasses can possibly lead to a tendency for the eye's own lens-focusing muscles to relax overly much. I use such glasses only so that I can bring the screen into focus from rather close distances. It's the same case as when I'd like to read really fine print comfortably. Moreover, I'm quite attuned to my eyes and can to a certain extent self-diagnose. To illustrate, before I ever went to an opthamologist some 10 years ago, I knew already the extent and orientation of my astigmatism.

--Lurch-- </div></BLOCKQUOTE>

excellent post lurch !

i'll fiddle with the numbers later to redo it in metric for my own use, so it is more intuitive for us euro's.

nice to hear you are working in astronomy. so we can now with certainty figure out the correct viewing angles represented by various monitors, and the right distances to sit from the monitors.

personally i think the min and max distance to sit from the monitor should be first identified based on our ability to correctly read text on a monitor in normal font sizes, and the right distance to view video (standard def and hi-def). this will give a right ballpark number to work from (for ex 60 cm for a 19' lcd). doing it that way will avoid getting people to sit to close to their monitors, and potentially damaging their eyesight, like the 20' owner who might want to sit 14.3 inches (36 cm) from his monitor in order to use the 70 FoV. that individual should really sit almost 2x further from his monitor for normal eyesight.

ideally in the future this could be represented on an excell spreadsheet type table, with all the numbers for the FoV's for the different monitor sizes, and a grey shading strip indicating the best viewing distances according to human vision and screen resolution (ie for a monitor running at 1024 x 768 you will need to sit far enough away for the pixelation not to be an issue, but the same monitor at 2048 x 1536 would allow you to sit relatively closer).

the last part of this puzzle will be to get a map in the full mission builder, which has a string of airborne aircraft (like the 109 with the useful 10 meter wingspan) at 100 or 200 meter intervals. they could be represented from 200 meters to 5000 meters (the il2 maps have grid lines so you can position them very precisely on the map. any il2 user could then bring up that mission, pause their play, and determine id the "normal FoV" for their monitor size and seating distance away, indeed provides accurate and realistic identification of these aircraft at the various distances.

doing it that way would conclusively quantify our visibility issues in this "simulator". once the FoV is set correctly for monitor sizes, and we know the degree of faults and problems it has in accurately representing human vision from a ww2 cockpit, we can then fiddle with things like AA and contrast to try and improve object visibility. that mission map could be used to compare visibility against open sky (which will be reasonably good) and visibility of those 109's against a ground terrain with forest cover (which will be very bad and unrealistic i believe, except if your monitor has 0.40 or 0.50 mm pixel sizees)

so what we need is somebody that is handy with the full mission builder. it would only take them 10 or 15 minutes to make such a map i think. personally i have no clue how that feature works.

Lurch1962
03-20-2007, 09:01 PM
Grifter,
One need not be overly concerned about eye strain resulting from a close viewing distance. How far from your eyes do you hold a novel when reading... 12 to 18 inches? Normal close focus for the eye is about 10 inches , assuming no presbyopia (farsightedness). So why not sidle up to a monitor, even though the pixels happen to be "over-resolved?" As long as one's eyes are able to comfortably focus without strain and fatigue, it should be OK.

As it now stands, the two primary causes of difficult-to-see plane dots at high resolution:
1) Such a dot is so miniscule amongst the vast acreage/greater number of surrounding pixels. It can't as easily "draw attention to itself." Kind of like a blade of grass in a field.
2) And by using a large screen to take advantage of that great resolution, one tends to sit farther back, which now makes the dot small in apparent angular size on the retina.

For myself, it remains to see if the dot size remains a constant 2x2 pixel block out to maximum ranges. I also wonder if at extreme range the dot is drawn a paler tone, to mimic visibility-reducing haze?

It would be a fairly quick affair to whip up a mission containing pairs of fighters and bombers strung out in a line at fixed intervals. I say both fighters and bombers so that differences in plane size can also be evaluated re. the dot visibility question. Perhaps one evening I should do just that... no promises, though!

Have you visited the link below? It contains a 4-panel image I cobbled together which addresses one aspect of plane dot size...

http://picasaweb.google.com/Lurch1962/Il2/photo#5043820460870562354


--Lurch--

grifter2u
03-20-2007, 11:03 PM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Lurch1962:
Grifter,
One need not be overly concerned about eye strain resulting from a close viewing distance. How far from your eyes do you hold a novel when reading... 12 to 18 inches? Normal close focus for the eye is about 10 inches , assuming no presbyopia (farsightedness). So why not sidle up to a monitor, even though the pixels happen to be "over-resolved?" As long as one's eyes are able to comfortably focus without strain and fatigue, it should be OK. </div></BLOCKQUOTE>

we have a somewhat different perspective on this (not to make a pun out of it) http://forums.ubi.com/groupee_common/emoticons/icon_smile.gif if you look generally at viewing distances for video (projectors, lcd and crt monitors, crt & lcd & plasma tv's ,etc..) there is a correct viewing distance for them, and this can be calculated. if you sit to close you see the pixelation of the screen, and if you sit to far you reduce the immersion factor and reduce your ability to discern detail.
- mostly this is defined by the technology of the monitor, screen size, and the resolution used, so that you can see the reproduced video in its optimal quality. when you look at a video being played on a tv or pc monitor, similar issues come into play and you mostly wouldnt want to sit any closer than 2x the hight of the screen (even if it is a high definition resolution). interestingly that is very similar to the distances you need for resolving 96 dpi for windows fonts, so the ballpark parameters are very similar for normal windows use with pc monitors (up to roughly 24' size).
- when you look at expert videophile advice they go into extensive detail on those issues, i am just giving you the main points. this is relevant for us il2 users, because on a good display medium the grafix quality and terrain modeling in the il2 sim is good enough to trick your mind into "suspending disbelief" and feeling immersed in the scenery you are flying in. but to achieve this video quality needs to be good !(this was possible to achieve on a good large crt monitor, but lcd and plasma is already a major compromise in that regard)
- the above info relates to a very good display medium for tv's and pc monitors, if that technology is less than high end (like some of the cheap TN based lcd's) you probably would need to even sit a little further away (so you cant see the imperfections as clearly). also for those that still use large crt monitors you would never want to sit closer than 2x monitor high, because of the intensity of the radiation.

that is what leads me to say that less than 60 cm distance from a 19 to 23' lcd is probably not appropriate. some might choose to however, that is up to them.

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Lurch1962: As it now stands, the two primary causes of difficult-to-see plane dots at high resolution:
1) Such a dot is so miniscule amongst the vast acreage/greater number of surrounding pixels. It can't as easily "draw attention to itself." Kind of like a blade of grass in a field.
2) And by using a large screen to take advantage of that great resolution, one tends to sit farther back, which now makes the dot small in apparent angular size on the retina. </div></BLOCKQUOTE>

agreed ! but as a famous french philosopher said some time ago "what is normal" ?

since aircraft in il2 are rpresented with LoD models and pixels, we first need to confirm these are indeed modeled in their correct sizes for the size aircraft and distance they represent. i am still not convinced that this is the case, but we are very close to solving this issue if we can represent a string of aircraft in the mission builder.

when a real ww2 pilot looks out his cockpit at the terrain below him and tries to identify targets (like moving planes or tanks and trucks) he has in real life a number of additional que's that add to his spotting ability, compared to the simple flat 2 Dimensional dots we have in il2 on a pc monitor. real objects are 3 dimensional and stand out more (even if painted in camouflage), and the human eye is extremely good at tracking a moving object against a stationary background.



<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Lurch1962: For myself, it remains to see if the dot size remains a constant 2x2 pixel block out to maximum ranges. I also wonder if at extreme range the dot is drawn a paler tone, to mimic visibility-reducing haze?

Have you visited the link below? It contains a 4-panel image I cobbled together which addresses one aspect of plane dot size...

http://picasaweb.google.com/Lurch1962/Il2/photo#5043820460870562354
</div></BLOCKQUOTE>
nice screen shots, it illustrates the problem well ! maybe some of that can be solved for the closer than 5 km models by adding more black dots to the multi pixel model, instead of having so many grey ones for the wings.

we have also succeeded in having a 2 page constructive on-topic discussion in a civilized fashion at ubi, there might be hope for mankind yet ! http://forums.ubi.com/groupee_common/emoticons/icon_biggrin.gif

Te_Vigo
03-21-2007, 05:40 AM
I'm glad you are now using some of the things you picked up from the other thread Grifter2u

grifter2u
03-21-2007, 07:23 AM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Te_Vigo:
I'm glad you are now using some of the things you picked up from the other thread Grifter2u </div></BLOCKQUOTE>
like what vigo ? did you think there was some hidden wisdom in something you said there ? its been going for 3 months now and you still havnt understood what it was about. you going to start the same thing here now ?

Lurch1962
03-21-2007, 04:49 PM
Grifter,
Visit the web site below to see just how much the "rules" regarding resolution matching can be violated. The product is a digital planetarium that projects a hemispherical (180 degree) image inside a dome, and whose circular image is only 768 pixels across! That's 4.3 pixels/degree, or 14 arcminutes/pixel.

Scroll to very near the bottom of the page and click the link "Dome Shots" to see just what you get. VERY BIG PIXELS!

I've always subscribed to the notion that for moving images--as opposed to stills--much loss of sharpness and a certain degree of oversampling can be tolerated. Especially if you feel like you're *inside*, as opposed to being a more removed spectator.

A common example of an over-sampled display... From the typical viewing distance, the 525 rasters of an NTSC TV display can be clearly distinguished by the viewer.


http://www.digitaliseducation.com/products-digitarium_alpha_2.html


--Lurch--

Lurch1962
03-21-2007, 05:05 PM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">nice screen shots, it illustrates the problem well ! maybe some of that can be solved for the closer than 5 km models by adding more black dots to the multi pixel model, instead of having so many grey ones for the wings. </div></BLOCKQUOTE>

Grifter,
The "grey" pixels which make up much of the distant B-17 (especially wings) are actually the result of anti-aliasing. In drawing the LOD model applicable for that distance, the rendering engine has to draw full-size pixels where in actuality the small dimensions (especially the thinner wings) would occupy only a fraction of a pixel. Hence the paler tones, and even some "dropped" pixels, depending on screen resolution.

Yet another limitation imposed on image fidelity by the software and hardware.

If one modified the draw routines so as to selectively add darker pixels to a distant pixelated plane, one might as well employ sprites (perish the thought!) http://forums.ubi.com/images/smilies/disagree.gif



Speaking of fidelity, note the difference between the B-17 representations depending on resolution (1600 vs 1024 horizontal pixels). At yet higher resolutions the plane would look even more realistic (within the limitations of the LOD model, of course!)

--Lurch--

grifter2u
03-24-2007, 09:03 PM
hiya lurch,

i have no idea what makes up the combination of grey and black dots for the distant low models. i thought it might have been a "feature" of the game grafix engine (in the same way it combines black and grey pixels for distant dots representing aircraft). if it is directly caused by a AA artifact trying to compensate for a low resolution having to draw fine detailed lines, then it is another major argument for using as high of a resolution as we can for il2 use. the high resolutions make the cockpit and the scenery look better as we all know, it seems logical it also improves LoD model detail. sadly it does the exact opposite for our ability to see distant aircrafts represented as dots (it makes them smaller, thus harder to see, hence the previous thread)

while doing some reading on human vision i found this interesting bit..
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content"> Visual acuity can be measured with several different metrics.

Cycles per degree (CPD) measures how much an eye can differentiate one object from another in terms of degree angles. It is essentially no different from angular resolution. To measure CPD, first draw a series of black and white lines of equal width on a grid (similar to a bar code). Next, place the observer at a distance such that the sides of the grid appear one degree apart. If the grid is 1 meter away, then the grid should be about 8.7 millimeters wide. Finally, increase the number of lines and decrease the width of each line until the grid appears as a solid grey block. In one degree, a human would not be able to distinguish more than about 12 lines without the lines blurring together. So a human can resolve distances of about 0.93 millimeters at a distance of one meter. </div></BLOCKQUOTE>

if what they are saying is correct (it is from wikipedia so add some grains of salt for critical reading), then this 0.93 mm at 1 meter distance is an interesting number and might help to explain some of the issues we are dealing with. when the il2 grafix engine jumps from using its smallest LoD model to drawing that distant aircraft as a 2 pixel item, then pixel size is very critical (and so is the viewing distance from the screen). that 2 pixel size ( 2x 0.25 mm, versus 2x 0.40 mm for ex) is already right at the limit of what the average human with good vision can detect, and any small change in the variables (dot size, viewing distance, pixel size etc..) can make all the difference between spotting that enemy at 3000 meters below you, or not seeing him at all.

the additional factor is of course whether representing that distant aircraft at 3000 meters with 2 pixels is realistic or not for representing the size that aircraft would be seen as in real life, and i think in most situations this is indeed way to small in il2 as well, which seriously compounds the above problem.

even after that, presuming in il2 we get the correct "pixel number' representing a distant aircraft size, having those 2 dimensional pixels as a flat object against a flat monitor terrain trying to recreate a 3 dimensinal real world that has depth and texture to it, to be realistic the 2 dot aircraft representation probably would need extra visibility added to it (like adding some pixels, or making all pixels black etc). but that would be the last step to figure out.

i reinstalled some of the old il2's last night to compare the way it handled dots in the various versions.some observations:
- the original il2 only had 3 FoV settings, normal (70 ?), wide (90 ?), and gunsight (30 ?). the distant aircraft "dot popup" was at about 9 km and it was a large 9 pixel blob immediately. very easy to see, but to large and unrealistic compared to real life.
- FB implemented the additional FoV's we now have, from 90 to 35, and the gunsight view. (i think the gunsight 30 FoV is not just a narrower FoV but also adds an additional zoom magnification from then on). "dot popup" was now reduced to around 5 km, and the dot it used was extremely small (2 pixels ?). even against the open sky the dot was fairly hard to see at 3 km, unrealistic in my opinion and Oleg swung the pendulum to far to the other extreme i think.
- PF still used very similar small dots and similar popup distance (i didnt try the AEP addon yet to check that step)

i think this helps to explain the 3.02 "dot fix" Oleg attempted. the FB and PF dots were just to small, and needed to be larger and easier to see (particularly against forest and terrain background). he was trying to find some middle ground between the 2 earlier extremes. the problem is that he was doing this around the same time people were transitioning from large crt's at low resolution, to much bigger lcd's that were also running at at much higher resolutions. so what might have been ok for one user would have been very incorrect for the other. if you add to that the "fake real" crowd attitude who believe that flying around with a myopic virtual pilot is realistic because they have never been in an aircraft, then you get the arrive at our current dilemma.

i'll try that 3.02 patch later and see how it compares to the earlier 2 or 3 versions of il2.

the game is afoot dr watson, we are getting closer to a solution http://forums.ubi.com/groupee_common/emoticons/icon_smile.gif

Lurch1962
03-26-2007, 04:42 PM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Next, place the observer at a distance such that the sides of the grid appear one degree apart. If the grid is 1 meter away, then the grid should be about 8.7 millimeters wide. </div></BLOCKQUOTE>

You're right when commenting that Wiki entries should be taken with a grain of salt. Indeed, there's an error in that tid-bit you got hold of. At 1m, 1 degree is 17.45mm, not 8.7mm.

--Lurch--

carguy_
03-29-2007, 05:39 PM
I don`t see why we should take this again into debating.

Bottom line is:

Oleg found a way,

Oleg had it fixed before,

Oleg turned down the fix,

end of story.

grifter2u
04-02-2007, 05:52 AM
nope, not end of story, unless you want to see the exact same problems in BoB again. if they design it the same way again from the start, it will be to late to complain then !

there is no "debating", the purpose is to try and understand the issue better and then based on facts to try and figure out what the right way of representing distant aircraft as dots is. all we want is to have distant dot visibility to be correct and the same size for all monitor users. probably this means il2 dot size should also be controlled by the FoV setting, just like LoD model size currently is.