Spaceship based telescopes

[David L Pulver]

Originally Posted by Ulzgoroth

Depends what you're trying to spot. Camouflaged facilities, probably not. Counting ships in a port and tanks or trucks in a parking area, very possibly.

Of course, the Hubble is not a spy satellite.

There is some anecdotal evidence that a lot of technology is rather similar, the optics are similar size, and NASA histories indicate technology transfer occurred between the Hubble and KH-11.

My current working option for visual light TV sensors based on this discussion at TL8 is:

Up to 16x magnification: weight = magnification/16.
Over 16x magnification: weight = (magnification/16) squared.

* You can then spend extra money to halve the weight, or take a big discount but double the weight.
* Assumes Tunnel Vision. Double weight for 120-degree restricted vision; 3x weight for human vision.

Thermal imaging sensors are about 4x heavier (effectively on level of telescopic vision worse.)

Night Vision are complex (Anthony has argued they should be much worse than GURPS portrays them; Kromm has disagreed, I think)
but probably 1-2x heavier? and costlier; I guess it mostly depends on the type of night vision

Thoughts?

[Anthony]

Originally Posted by David L Pulver

Night Vision are complex (Anthony has argued they should be much worse than GURPS portrays them; Kromm has disagreed, I think) but probably 1-2x heavier?

Realistically, you can't put night vision on an electronic telescope operating at peak resolution, they are already sensitive on the scale of single photons.

If you want figures of merit, the fully dark adapted human eye under good conditions of darkness can spot stars of magnitude 6. According to an exposure calculator, a 1m telescope with a 1s exposure wanting a S/N ratio of 5 can detect objects of magnitude 16.7. Since 5 points of magnitude is a factor of 100, that's about 20,000 times as sensitive, or equivalent to a lens of size (1m/sqrt(20,000)) = 7mm.

Which is about the size of the dark-adapted human pupil. Arguably average human vision should be a limit of more like magnitude 5, and that scope may not be quite maxed out, but in general a maxed out 2.5-3mm lens will match the human eye in both resolution and sensitivity.

Night vision gear has a couple of benefits. First, it adapts much much faster than the human eye (which takes on the order of half an hour, and can be blinded very fast). Secondly, by being indiscriminate about colors (including collecting near IR), it collects several times as many photons as the retina. Third, it has grossly oversized optics relative to its resolution -- a 30mm operating at 10x magnification will have about the same night vision as a human, one operating at 1x will have 100x the night vision. Reasonable TL 9 sensors can probably swap between night vision and telescopic vision on a 1:1 basis, and can add one extra level of night vision by switching to black and white with NIR. They can also take extra time, which the human eye generally cannot do.

[David L Pulver]

Originally Posted by Anthony

Third, it has grossly oversized optics relative to its resolution -- a 30mm operating at 10x magnification will have about the same night vision as a human, one operating at 1x will have 100x the night vision. Reasonable TL 9 sensors can probably swap between night vision and telescopic vision on a 1:1 basis, and can add one extra level of night vision by switching to black and white with NIR. They can also take extra time, which the human eye generally cannot do.

Well, the human brain can take extra time to study a scene, so it probably events out in terms of bonuses...

Would it be realistic enough for the sensor design rules to just say you can trade off night vision and telescopic vision levels at 1-1 at TL7-8 while designing the sensor, but at TL9+ the same sensor can electronically switch between them each turn?

[Anthony]

Originally Posted by David L Pulver

Would it be realistic enough for the sensor design rules to just say you can trade off night vision and telescopic vision levels at 1-1 at TL7-8 while designing the sensor, but at TL9+ the same sensor can electronically switch between them each turn?

If you've got access to variable zoom in the first place, you can do it at lower tech levels -- a conventional camera zoom lens has a different minimum F-stop at different resolution. For electronic zoom, usually not, but electronic zoom has other problems (basically, it's like enlarging a digital image: you don't get any more resolution, it's just easier to work with).

A simple lens setup with lens size L and magnification M produces an image at the eyepiece with a size of L/M. If this is equal to or larger than your pupil (assume 7mm), you have full night vision. If it's smaller than your pupil, what you see through the optics is dimmed (you're getting the same number of photons, but they're spread out over a larger area, and thus brightness per unit area is lower). If you replace the eye with a sensor, you get the same effect. This is why binocular sizes are what they are -- if you want to use binoculars at night you should get ones with a lens that's 7mm* its magnification (e.g. 7x50), ones mostly intended for daytime use are usually more like 5mm*magnification.

Note that the main reason you do any of this in the first place is because of pixel counts -- for a camera with a wide field of view and high resolution, the pixel count gets out of hand (10x mag, 180 degree field of view, is 7.4 gigapixels) and your system weight is actually dominated by your camera, not your lens.

[Eukie]

It is generally considered easy to spot something at the example 100 yards, but it should be remembered that this typically includes a search period.

The probability that a human eye's saccades passes over the object-of-interest can be given by the equation:

P(look at) = 1 - exp(-700/G*A_f/A_t*t)

Where A_f is the angular size of the eye search field, A_t is the angular size of the region being searched in, t is the search period in seconds, and G is a factor accounting for the environment typically equal to 10. The field of view of the eye that's actually active in searching/spotting is about 5 degrees wide. The full human field of view is 180 degrees horizontally and 135 degrees vertically, though for targeted searching this can be made smaller. This assumes the object being searched for is narrower than 5 degrees - for large objects replace A_f with the angular width of the target.

This is for a random search; trained searching makes the movement of the eyes less random, allowing for about a fourfold increase in searching efficiency.

In GURPS terms, most measurements are logarithmic, so the angular widths searching time, searching efficiency, and environmental factors can be described in terms of additive and subtractive factors, and the 3d6 curve roughly corresponds to the 1-exp(x) probability curve.

What makes visual search so "easy" in real life is that we usually have time to spot things, and furthermore, that movement will trigger the eye's wider field of view, greatly increasing A_f and the ability to spot in general. To account for the increased detectability of moving targets, multiply t by a factor (1+0.45*v)^2, where v is the angular velocity relative to the eye, in degrees per second. Sudden changes in brightness are also easy to spot because they benefit from more of the field of view.

In addition to this, there are two more factors to account for; when the eye passes over the target, it must be distinguished from the background, and when distinguished, it must be recognized.

The former depends on the target's contrast against the background; this is where darkness and camouflage comes into play. When a target has no contrast against the background in brightness or colour and must be recognized otherwise, it's considered to have a 50% chance of being distinguished. A blatant target, like man in a suit in a white room, has a 100% chance of being distinguished. The recognition is where the resolution comes into effect the greatest; the resolving power to tell which direction something is pointing must be about 3 times as great as to simply say something is there. The factor increases to 8-ish for telling what something is ("It's a tank", "It's a person", etc.) and 12-13 to tell exactly what it is. ("It's a T-72", "It's an enemy soldier", etc.)

Empirical data on the exact effects of camouflage (affecting contrast and colour contrast) and clutter (affecting G) is hard to find.

A very simple system to handle this would be three-part:

1: A Per-based roll against a relevant skill, affected by distance, target size, search field size, and search time.
2: A "defence roll" by the target, using their Stealth and/or Camouflage, aided by darkness and opposed by an unforgiving hiding place and good night vision.
3: Check how well the target is recognized by comparing Visual Acuity to Range+Size. Sometimes all you can say is "there's something in those bushes".

This is just some vague notes based on my memory of some papers I read once, mind.

[Anthony]

A lot of the problem with perception rolls is that realistically, a lot of the modifiers are too small, but tracking everything that applies is unplayable. Under controlled conditions, the difference between never recognize and always recognize is something like a factor of 3 in size (compare 12 point text to 4 point text), corresponding to a modifier of 3 on the range/speed chart, and skill makes very little difference. On the other hand, under actual 'adventuring' situations the variance is much larger (due in large part to modifiers that would be a pain to track), and the skill of knowing where to look is pretty important.