Quote:

---

Another point IMO is image scale has a real effect on color perception....the greenish blue I saw observing M27 through my 25x150 binoculars was not seen though larger apertures at higher magnifications.

My take on this is spreading the total amount of light over a larger area (ie higher magnifications) naturally puts less energy into the same area (retinal receptors)

Also, the effects of two eyes has to be taken into account. Two 6 inch unobstructed apertures can cram a lot of photons into the eyes. There is 40% greater contrast using two eyes over one. The Fujinons are equivalent to a 7 inch refractor in perceived light gathering at a 6mm exit pupil!

Enjoying this discussion....this is what I look forward to on CN.

---

Post Extras:

Quote:

---

Quote:

---

Well,
Mauro's logic... But how to set up a test........

---

Hi Don. I am happy you read the paper. Now we can further discuss. The point you raise are all interesting. I have a lot of comments here are a few:

First we are speaking of a very bright nebula. The brightest parts are at magnitude 14 (surface mag). I have a SQM and can assure that 14 is bright. False color percetions at that level are mistakes due to the intervention of rods, but cones are still operative (we are tetrachromatic at that point but we do not see a 4-dimensional color space because apparently rod signals enter the same channels of the cones).
Given the brightness of M42 you have the potential (even theoretical) to see real colors and also saturated colors (not rudimentary). How much saturated you may find in the companion paper.

---

Real color perception variance is probably similar to visual sensitivity to stimuli, which, in my experience, covers about a magnitude and is partially due to differences in visual acuity. As is the case with stellar images, defocusing often makes colors more apparent. Since doing so spreads the light out, reducing unit intensity, size appears to be a bigger factor than intensity.
As an aside, the small % of women (only women) who have tetrachromatic vision (4 different color receptors) would see colors differently than the rest of us, but, since they use the same language, we'd never know how.

Quote:

---

The experiments were carried out with samples of uniform color which were seen under 8° apparent angle (this apparent angle is important... but again the authors know what they do). If there are regions of different colors compressed into a smaller apparent angle they cannot be recognized, even if the cones see the color. Thus it is not surprising that large scopes split areas of uniform colors in apparent patches large enough to be recognised (I still doub of H-alpha) but colors a 14 are seen provided the coloured areas are large enough (that is why your dark adaptation is disturbed, if the apparent area involve a large fraction of you FOV). If you look at a fractal image at distance (or a planet) you don't see vivid colors, but as soon as the details are resolved you see the color (provided they are large enough).

---

Therein lies a problem. I would rather the test be done with monochromatic light (say H-Alpha) in which the intensity is reduced to the point where no color perception is possible. I would guess different individuals would lose color perception at different luminosities. Judging from the studies on perception of grayscale brightness differences, it is likely that the differences between individuals could easily exceed 5 magnitudes. That no two individuals see a five-magnitude difference through the scope does not invalidate the point that there could be a 5 magnitude difference in color perception. The cited studies do not address this issue, but do discuss false perception of color. There would be a fine line between real and false at some brightness levels. It is quite possible that some individuals would be seeing "real" color when others are seeing "false". If they see the same color, how could you tell? A "threshold" test might help.

Quote:

---

There are areas of M42 that shine at 17 and below. Here is were rudimentary real color perception ends. The remaining are artificial colors.

---

That statement is implied by the studies, but not verified to my satisfaction. Some individuals see color in the zodiacal light and even Gegenschein, yet these are both fainter than the mag.17 figure. What is interesting is that the colors seen are verified by photometry. Could it bee that the threshold of true color perception varies a lot among humans?

Quote:

---

Little planetaries are very bright, and thus they are in the *real* mesopic range (for OIII). However the bright planetaries are bright because they are little, so even smaller ar the dim parts (the medium lights that should appear reddish).
The paper demonstrated reddish percepts for medium lights provided that they are large areas. Nothing is said about that effect for smaller areas. Maybe they are simply too small to be noticed (the ansae of M27 are instead large).

---

Another point to study. Small objects should also be subject to false color perceptions. Are they? Unknown.

Quote:

---

Another point you raise is the numerosity of the people who made tests. You are right when you say that the average response is identified but not the variance. However there are people who say seeing color at surface brightness of 22 (veil and M97, or in some regions of M42 that are at 22). From 17 (the average threshold) to 22 there a 5 magnitudes. A factor of 100. I know people that see stars half a magnitude dimmer than others, but nobody who sees 11.5 at the same place and time when the average see 6.5.

---

Stellar visibility and extended object visibility are not the same thing because of the radically different sized areas of the retina engaged.

Quote:

---

So even if we do not know the variance (but we have hints it is not large because the thresholds were similar for the small sample of persons who made the tests) it is very unlikely that the threshold is 17+/-5 (what is needed to see reral colors in the faintest parts of M42).

---

This is an assumption based on a small "N" and under conditions in which full dark adaptation might not have been applied. We know that rods grow in sensitivity by a factor of 90,000 times. We also know that cones turn off below a certain intensity of light. What we don't know (I've read hundreds of studies) is the response of the cones under low light conditions (except for the blue shift) and under conditions of maximal dark adaptation. I know that dark adaptation varies HUGELY in the human population--way more than 5 magnitudes--but we do not know about variances in the perception of different intensities of red (see my above comments). The cited studies only show what happens below the threshold for broadband-illuminated, large-scale, areas.

Quote:

---

Another point I did not notice is that the tests were carried out with broadband stimuli. Nebuale have line emission spectra. This might have some effect. However a pitfall is hiding here: our visual system uses three receptors in order to reduce as much as possible the phenomenon of metamerism (two different spectra and colors that produce the same stimuli and thus look like the same color). Three receptors are enough for smooth broadband spectra. Not enough for line spectra. L-cones are stimulated by OIII line. The mechanism of color constancy/compensation could produce apparent (metameric) red color.

---

Which would explain why a lot of the double star viewers in the past saw such vivid contrasting colors. Bear in mind that the perception of color in stars varies hugely with some individuals seeing more red than others.
I would like to see a red-light stimulus that is gradually raised in intensity from a very low level (say mag.24-25) which checks first where the stimulus is first seen (a test of dark adaptation and rod sensitivity) to the point where color is first perceived. Then, the identical test with a green light. If the cited studies are valid, the faint green light would first be seen as red and then turn greenish at some light intensity.
Starting out below the threshold of visibility would be critical in determining the variance in the visible threshold for light in general, and also for specific colors. If the blue shift holds, the threshold for blue light should be much fainter.
It would be an incredibly hard test to administer, since staring at a computer screen would ruin dark adaptation. The subjects would first have to sit in nearly-complete darkness for 45 minutes, and then look at a black screen on which a small square was projected. Looking at the light source or a computer screen would immediately invalidate the results. Because of huge variation, the projected images would have to start quite dim, say below 26th magnitude per square arc second.
This is similar to the hearing tests in which I participated. Not only the range was large, but the minimum threshold varied by a factor of over 1000 among our subjects. We even had some individuals who could hear the pulse of each heartbeat as the blood moved through the ear's blood vessels.
We have no reason to believe the threshold of perception in vision does not have a wide range. The cited studies purport to measure the perception of color at low intensities. I would immediately suspect results that show a 1-magnitude range in visual perception of color. Why? Because I routinely see large variations in the ability to see at all: acuity, faint images, colors in stars, limiting magnitude (which we know varies more than 4 magnitudes--see Schaefer).
Further studies are needed. That doesn't invalidate the cited studies and they are interesting.

--------------------
Don Pensack
12.5" Truss Dob, 5" Maksutov
Sustaining Lifetime IDA member, TeleVue junkie

Post Extras:

Quote:

---

We even had some individuals who could hear the pulse of each heartbeat as the blood moved through the ear's blood vessels.

---

That's when I know I'm somewhere under world class dark skies....no humans, no noise, no man made light pollution.

Central Nevada

Post Extras:

Quote:

---

You *do* se colors. The fact that you se *real H-aplha* is questionable, provided that there are proofs of false color perception due to a plethora of different cognitive effects.

---

Quote:

---

You have not given any proof that what you see is real (you claim it is real because you and other see it, and if somebody else does not you point him as "unable") except for some anedoctcal report (a friend seeing 7200 angstrom and the OIII filter with leaks).

---

Quote:

---

I understand however that there is no room for discussion at leas until you refuse to even read the studies I pointed out.

---

I did *not* refuse to do *anything*! The nearest University library is over 40 miles away, so I do not have access to the papers (at least not yet). However, I still believe (as some have already pointed out to you), that to be useful, a study of low-light color perception *must* be done with very similar light conditions and sources to those we view on the sky. Until this is done, I remain somewhat skeptical that they rigidly apply to all visual observing of faint sources. Clear skies to you.

--------------------
David W. Knisely
Hyde Memorial Observatory
http://www.hydeobservatory.info

Post Extras:

Quote:

---

Real color perception variance is probably similar to visual sensitivity to stimuli, which, in my experience, covers about a magnitude and is partially due to differences in visual acuity. As is the case with stellar images, defocusing often makes colors more apparent. Since doing so spreads the light out, reducing unit intensity, size appears to be a bigger factor than intensity.
As an aside, the small % of women (only women) who have tetrachromatic vision (4 different color receptors) would see colors differently than the rest of us, but, since they use the same language, we'd never know how.

---

Post Extras:

Quote:

---

I have used Hydrogen gas discharge tubes and spectrometers, so I can see what the H-alpha line looks like visually. Even when not using a spectrometer, I can see the gas discharge tube as a nice bright pink, which is very similar to (but quite a bit brighter) the shades of pink I see in M42.

---

This is contradictory in itself: you say that H-alpha at high intensity looks exactly the same color (hue and saturation) that when you supposdely see it at very faint level!! There is a huge amount of proofs that hue perception and saturation and chroma varies.
You simply do not read anything because you already know. I will no longer reply to you at least until you accept to read and discuss the points as is doing Don.

PS have you ever studied the sensor fusion processes that happen in the brain? Here there are lots of false color detection that happen at full illumination (figure at dim lights). But... you are sure that the dime faint pink is exactly 656 nm.... and therefore you se H-alpha.
http://www.purveslab.net/seeforyourself/ Look the effects of color dominants. why do they happen? because the brain, in attempt to evaluate the reflectance curve of an object has to subratct the spectrum of the incident light, which the brain evaluates with an algorthm that makes a long different processing path until high level object shape color recognition. BTW red is the color that happens when compensating a green dominant. A similar algorithm (google retinex) is used in automatic white balance in many camera. Look what happens throu a OIII filkter with NO red leak:

This is a scene in full color.

http://bp2.blogger.com/_rzxz_YRzFn8/RyTnpJqofvI/AAAAAAAAAEc/xp7ekvx8_VE/s1600-h/IMG_5181.jpg

This is the same scene through a OIII filter without retinex algorithm (no color constancy effect).

http://bp3.blogger.com/_rzxz_YRzFn8/RyTnkZqofuI/AAAAAAAAAEU/qp7XzXy7nyo/s1600-h/IMG_5182.jpg
It is monochromatic OII right? (I bet you see the wavelenght is 500.7 nm)

This is the same scene with the retinex algorithm on.

http://bp1.blogger.com/_rzxz_YRzFn8/RyTne5qoftI/AAAAAAAAAEM/4n4RWcv608Y/s1600-h/IMG_5186.jpg

Surprise surprise you can see (and visually that happens) colors through the filter!! Not only: there is red, the black "K" is RED. I bet you would say that you saw the real red, in fact the letters are red... you see red, hence you see the real color... (well let us ignore the discrepancies in the K).

Edited by Mauro Da Lio (11/28/07 04:25 PM)

Post Extras:

Quote:

---

This is contradictory in itself: you say that H-alpha at high intensity looks exactly the same color (hue and saturation) that when you supposdely see it at very faint level!!

---

Quote:

---

There is a huge amount of proofs that hue perception and saturation and chroma varies. You simply do not read anything because you already know.

---

Quote:

---

I will no longer reply to you at least until you accept to read and discuss the points as is doing Don.

---

That is your choice. I have read what Don has written, and he is doing just fine without any input from me. It would be nice if you would at least accept the points he made which you seem to be doing your best to downplay by going back repeatedly to your paper sources. If you don't want to believe that people have seen pinkish or reddish colors in M42, that is fine, but please don't imply that they are illusionary when they clearly may not be.

--------------------
David W. Knisely
Hyde Memorial Observatory
http://www.hydeobservatory.info


Edited by David Knisely (11/28/07 09:12 PM)

Post Extras:

Quote:

---

Are the colors real? Well, I can't say the tints perceived are EXACTLY what shows, but close enough to say that Occam's razor would say that the assumption that real colors are seen is a simpler explanation than false colors.
Why? Because the tints are too similar to photographic evidence to be dismissed out of hand.

There is a way to test this, though. If the coloration disappears when an O-III filter is added, and reappears when the filter is removed, then real color perception is likely. If, on the other hand, the coloration is still seen with the O-III filter in place, then the colors are false and perception is guided by factors other than perception of real colors.

---

Colors at dim lights are different than at full illumination. Since you red the first paper, I think you could also read this one (full reference in a previous post).

Change of Color Appearance in Photopic, Mesopic and Scotopic Vision
Jae Chul SHIN, Hirohisa YAGUCHI 1 and Satoshi SHIOIRI1

It provides data about the changes in saturation (chroma) hue and lightness between 1000 and 0.01 lux. Data have been measured with a haloscopic technique (one eye looking at the sample and the other into a CRT which the test person controlled to produce a matching color.
There you may find quantitative evaluation of reduction in chroma as weel as an indication of variability between persons. It also shows the matching points at 0.01 lux.

Can you test with a H-alpha filter? If you do not see anything...

By the way, about the limit between real and illusory color perception you are right when you say it is difficult to asses (that is because real colors at mesopic limit are the same as false colors below (not surprisingly). However, if you look at figure globally (the big table of reported colors) you may easily see three different zones: mnsopic perception starts at level 2 (0.3 lux) where all colors for *all* test persons show recognition errors. Ten at level 0.01 and 0.003 lux another change happens (or is completed) for *all* people again: red samples become black (no longer seen by either cones and rods), orange becomeseither black orange or red, yello become blue gray or green, pink becomes gray, blue-gray, blue, green yellow orange yellow becomes gra or blue or green, blue becomes black blue gray green orange and red and so on. Confused? Yes that is what happens there!
The existence of three zones is clear if you consider the whole set of data. It is also clear that the three zones happens at the same levels for the test persons.

So, using the razor, I would say that the overall data points out the three regions: photopic (perfect color recognition) at 1 lux, mesopic (many mistake but still correlations with real colors) between 0.3 and 0.01 lux and scotopic below (no correlation with color but correlation with lightness as shown in the figures on top of the same page).

PS as soon as I can get a H-alpha filter I will try. I have made some (non rigorous tests) with samples in a dime room and measured the brightness with a SQM. My limits are consistent wioth the reported data.

PPS here is another interesting paper about how high level percetions are built (i put this link ito question the argument that "I have seen it with my eyes"):

Seeing is believing: A common sense theory of the adoption of perception-based beliefs
JOHN BELL and ZHISHENG HUANG
Applied Logic Group, Computer Science Depar tment, Queen Mary and Westfield College, University of London,
London E1 4NS, United Kingdom
(Received May 28, 1998; Revised October 20, 1998; Accepted December 8, 1998!
Artificial Intelligence for Engineering Design, Analysis and Manufacturing ~1999!, 13, 133 – 140. Printed in the USA.
Copyright © 1999 Cambridge University Press 0890-0604 0 99 $12.50

1. INTRODUCTION
It is a commonplace that seeing is believing. When asked to
justify a belief, people often reply “I saw it with my own
eyes.” Indeed, people are often not prepared to believe some-
thing unless they have seen it for themselves. Perceptions
also occupy a place of special impor tance in epistemology,
being considered as fundamental, incorrigible, etc. How-
ever, there is always the possibility that perceptions are il-
lusory....

...

It is not the task of perception systems ~whether ar tifi-
cial or biological! to enter tain sceptical doubts of a philo-
sophical nature. Their task is rather to produce the high-
level perceptions that the philosophers take as their star ting
point. Systems of this kind are thus mainly concerned with
the task of interpreting sensor data. For example, a vision
system might star t with a two-dimensional array of pixel
data supplied by its camera, and its task is to interpret the
image, inferring edges, then sur faces, then objects, etc.,
and perhaps finally producing a high-level symbolic de-
scription @see, e.g., Marr ~1982! and Ullman ~1996! for an
introduction# .

....

However, once this task is accomplished there remains
the question of the interaction between the agent’s vision
system and its belief system, and it is here that the problem
of illusion arises. Should the agent believe what it sees?

....

But, what if the agent’s perceptions and beliefs con-
flict?

....

it should depend on the comparative reliability of the perceptions and the be-
liefs in question. If the perception is more reliable, the agent
should revise its beliefs to accommodate it. But, if the beiefs are more reliable, the perception should be considered
illusory and the agent’s beliefs should remain unchanged.


Clear? Call it a implementation of the razor.

So the point now is: are perception at so dim light more reliable tha a huge amount of studies carried out with stringe rigorouds method by the worldwide community of sccientist in cognitive science?

PPPPS Do you believe i mmake typpo mistakes?... nahhhh they are illusions

Edited by Mauro Da Lio (11/29/07 03:59 PM)

Post Extras:

Quote:

---

I would read the papers *if* I had access to them

---

Quote:

---

The Oxygen III line filter experiment I described earlier *does* indicate that the reddish color I was seeing in the nebulosity was due to red light from M42 getting into my eye and not some illusion as you seem to allege.

---

Post Extras:

Quote:

---

several observers saw the same exact colors

---

Post Extras:

Quote:

---

Quote:

---

several observers saw the same exact colors

---

Don, all the observers have the same percetion system (same processing of data). If color percetion were related to lightness (as literautre says) they would have perceived similar colors. Theory 1 is not contraddicted by this observation.

---

That is logical.
But when the colors seen match the photographic evidence?
I guess you would argue that if faint reds are seen, that would be expected at the low levels of illumination, and that this happens to be the colors present in the nebula is accidental.
Especially if lab results show color perception is impossible at that brightness level.
Then the color is illusory.
But, taking the Orion Nebula as an example:
1) How bright does the illumination have to be to damage your night vision? Certainly a level that high would be adequate to activate Mesopic vision. And I KNOW that M42 in large scopes can damage night vision.
2) The color match in the experiments that used both eyes looking at different targets (one being a CRT) is fundamentally flawed. One cannot have dark-adapted vision when staring at a CRT screen. The overall size of the light, coupled with the fact it shines directly into the eye, precludes the possibility of having dark adaptation in that eye. Other tests show that one cannot have dark-adapted vision in one eye when the other eye is compromised. Hence, this study is not done at the limits of vision. Additionally, the dominant/non-dominant eye problem would throw the entire experiment into question. Color matching, using their experimental setup, would be quite inaccurate.

So here is my conclusion:
Because colors perceived match photographs, and the image is bright enough to activate mesopic vision and damage scotopic vision, and because the experimental methodology does not match the circumstances of seeing color images in the field, therefore I suggest that, at least in the case of M42, real colors are seen and the agreement of many observers indicates all are seeing real colors.

You would be correct to point out that the results of such viewing of M42 does not necessarily prove that the experiment's results are invalid.

But I would still like to see what a field experiment's results would be like.

By the way, I have used an H-Alpha filter on M42 (using an 8" scope). It was very dark, very red, and very much smaller than seen through broad-band light. But, I could see it, and I could see some of the "arches" section (basically the brightest parts). I tried a black cloth over the head and this was what was required to see any details at all. It was not an observing situation I would often repeat. Now that I own a 12.5" (31.8cm), though, I'd like to try again. I work someplace where I can take one off the shelf. I'll try at the dark of the Moon.

--------------------
Don Pensack
12.5" Truss Dob, 5" Maksutov
Sustaining Lifetime IDA member, TeleVue junkie

Post Extras:

Quote:

---

But when the colors seen match the photographic evidence?

---

Post Extras:

Quote:

---

If the "facts" don't match the experience of several veteran observers, maybe it's time to question the conclusions of the studies....

---

Post Extras:

Quote:

---

1) How bright does the illumination have to be to damage your night vision? Certainly a level that high would be adequate to activate Mesopic vision. And I KNOW that M42 in large scopes can damage night vision.

---

Post Extras:

Quote:

---

2) The color match in the experiments that used both eyes looking at different targets (one being a CRT) is fundamentally flawed. One cannot have dark-adapted vision when staring at a CRT screen.

---

Post Extras:

Quote:

---

By the way, I have used an H-Alpha filter on M42 (using an 8" scope). It was very dark, very red, and very much smaller than seen through broad-band light. But, I could see it, and I could see some of the "arches" section (basically the brightest parts). I tried a black cloth over the head and this was what was required to see any details at all. It was not an observing situation I would often repeat. Now that I own a 12.5" (31.8cm), though, I'd like to try again. I work someplace where I can take one off the shelf. I'll try at the dark of the Moon.

---

Post Extras:

Quote:

---

Quote:

---

2) The color match in the experiments that used both eyes looking at different targets (one being a CRT) is fundamentally flawed. One cannot have dark-adapted vision when staring at a CRT screen.

---

No. It is explained: one eye was dark adapted and the other is looking to the crt (with light shielding). It is a establisghed methodology (haploscope http://en.wikipedia.org/wiki/Haploscope). Do you really think they might have been so naive? T

---

Perhaps. Especially if they weren't experienced astronomers.
If the dark-adapted eye is the dominant eye, perception would be different than if the dark-adapted eye was not.
Also, comparing a bright and dim image simulataneously may change the response in the dark-adapted eye. Was that taken into account? Can one have one dark-adapted eye and one not dark-adapted? I'm not sure it's possible because of sharing of information from the retinas (remember, not all of the data goes to the opposite side of the brain--it's shared between the hemispheres).
Without backup experiments to determine the viability of this experimental model, it is likely to be, as I said, fundamentally flawed.

--------------------
Don Pensack
12.5" Truss Dob, 5" Maksutov
Sustaining Lifetime IDA member, TeleVue junkie


Edited by Starman1 (11/29/07 08:43 PM)

Post Extras: