The "Whatever Theory" Identifying The World...

[Arete]

I am just checking Arete's claim based on WT's theory, a smart phone app I just got the other day, and the undercolor reduction method newly described, to come up with an ACTUAL R/G number for the bands on bees that Arete brought to our attention as being of the same color, and as examples of why speices can not be IDed by color.

 

Except you've developed a strawman test based on a flawed interpretation of the point I was making. Of course you can pick apart individual bees by color if you use a high enough resolution method - that was never in question.

 

The issue, and I'm repeating myself again here, is that the individual variation within species overlaps with the variation between species and vice versa - rendering delimitation using color alone insufficient to assign individuals correctly to species in a large number of cases.

 

As an analogy, I could decide to use height to bin people into age classes. I could develop a laser measuring tool accurate down to 0.0001mm. Despite being able to very accurately measure the height of individual humans and distinguish between them, my method would still result in poor classification and thus fail. Why? because once humans reach adulthood, height is a poor indicator of age. Similarly, color alone is a poor indicator of species boundaries.

 

Whether or not you can tell apart the individual bees in the aforementioned photograph is a moot point.

[tar]

String Junky,

 

You don't find anything interesting about the data points I presented? No interest in seeing what 12 points and 18 points and 24 points will bring?

 

Regards, TAR

[StringJunky]

String Junky,

 

The fact that species cannot be accurately delimited by color alone was decided upon years ago. Possibly before 24bit color.

 

I am just checking Arete's claim based on WT's theory, a smart phone app I just got the other day, and the undercolor reduction method newly described, to come up with an ACTUAL R/G number for the bands on bees that Arete brought to our attention as being of the same color, and as examples of why speices can not be IDed by color. If we can successfully read the ACTUAL R/G number off of bee 3 and off of bee 8 and they are different numbers, then we can take the test to the field and see if a different picture of each species, 3 and 8, yield likewise, different identifiable numbers in that band.

 

A reasonable test of Whatever Theory's claims against Arete's claims. One claim or the other will have to be adjusted, based on the outcome of the challenge.

 

I take offense at you saying I am blindly soldiering on, in the face of scientific fact already established. I am trying to falsify WT's claim.

Is Arete's claim somehow immune from falsification, just because nobody has yet falsified it?

 

Regards, TAR

I'm sorry you feel offended but it is patently obvious that the idea has no traction... it's like people are trying to find a grain of gold in a lorry load sand just to try and justify the effort spent now that's been made clear. if the idea worked, and Arete has made clear he's familiar with the concept and it's limitations, don't you think he'd be using it?.

[tar]

Arete,

 

I realize you announced yourself as having been in the species IDing business, so I am sure you know what you are talking about, and color alone will not do, to ID species.

 

I am just interested in finding out how far WT's theory goes, and WHERE it fails. And where there might be a useful application of the ideas we are coming up with here. I just took offense at String Junky talking like everything we are doing here is a waste of time. Whatever Theory has learned a few things, I have learned a few things and seen some interesting hue consistency, even where shade varies greatly. Whatever theory's theory is faulty in several ways, as we have pointed out to him. He is adjusting his claims, accordingly and has not even clearly stated yet, what those claims are. And several others have gotten something out of this as well. I don't find the blind soldier comment useful or sensible.

 

Regards, TAR

For instance, it made me think about the cuckoo bird and other mimickers. If a mutation of the Mclr molecule and amino acid variations and so on are responsible for color variations down through evolution, how does one species arrange to have the same mutation that the species they are mimicking had? Are there just a finite amount of Mclr mutations and amino acid arrangements, that every species would possibly cycle through, and the ones that worked to some advantage would stick and the ones that didn't would just die out.?

or are there ways a cell can manufacture a protein, or copy a protein that was ingested by the host, or otherwise introduced to the organism, not through the genes but through the environment in some fashion?

 

After all there might be a reason for the ole saying, "you are what you eat".

[whatever theory]

Hello Tar I have sent you a private message. Once we agree on all of the exact details I will provide all of the data on all of the bees.

 

StringJunky there is plenty of material on the net that I do not find interesting so I just stay away from it, but I do not go to all of these sites and tell them that there work is useless and try to disrupt there conversation. You do not have to look at this thread, nobody is forcing you.

I am sure that a lot of the exact same arguments were brought up when they were discovering DNA, like if we already know what a human is then why do we need to find out what their DNA looks like and there are other ways of determining a species that are already known to be the best ways, so why are you wasting time trying to figure out DNA.

Like DNA if CGG is correct then yes a lot of the science books will have to be rewritten along with many of the claims by scientist in there fields, but to just stop trying to figure this out so that these same ideas, which are so comfortable to science do not need to be changed, is bad science.

Would you rather we go to another science forum to figure all of this out? Or figure it out here?

 

Sure my theory my be wrong but at the same time it looks to me like it is rite.

 

Arete

There maybe a common color that say 10 different species of a bee share in a certain area, which would be great because that could be the first value that determines it is a bee from a certain area. Now all you have to do is find out the differences in each bees color to determine it exact species.

I think Tar is rite, only till recently has this technology been available to do these kinds of tests and the only way to test them is to capture their image. Until a new type of photograph is invented we have no choice but to use the current photographs available.

To me arguing about whether there is a consistent color code in nature is like arguing that the Earth is Flat.

You can go to "Google Images" and see pictures from outer space that the world is round and so to can you go to "Google Images" and type in pair of birds, fish, reptiles, etc. etc. and see that these color codes are consistent threw out the natural universe.

 

Here is a picture of different bees from the same species and from the same geographic location, in each picture.

I think it is clear to see that each bee is an exact twin to the rest in its colony yet each colony appears to be a little different from each other. These are the first 12 images I found in Google images when typing "many bees"

Thanks

I think something is wrong with this site. I can not upload this picture.

I will try again later.

[StringJunky]

...StringJunky there is plenty of material on the net that I do not find interesting so I just stay away from it, but I do not go to all of these sites and tell them that there work is useless and try to disrupt there conversation. You do not have to look at this thread, nobody is forcing you.

I don't go to different sites and criticise ideas, only this one. Sorry, but, when you present an idea, purporting to be scientific, it's not our job to help you make it work. if an idea survives criticism and can make accurate predictions, then it's a feasible idea.

Edited July 10, 2015 by StringJunky

[tar]

Whatever Theory,

 

You don't have to speak to String Junky that way, on my behalf. In fact you shouldn't speak to String Junky that way at all. In fact I should not have spoken to him the way I did.

 

He and Arete are long time, well respected experts on this board. If they are trying to keep a little order and sense in here, it is perfectly proper, and we should listen and learn and not let our "prides" get hurt.

 

After my initial reaction to String Junky's comment I have had time to think about it, and Arete's comment, and they are right. We, you and I are trying to make a silk purse out of a sow's ear. The horse is dead, and we are trying to beat it back to life. After many pages, we really have not gotten anywhere. I am of the mind to give it a rest, for now.

 

I still like your method. I still like my additions to it, and the hue number determination. But we are about done here, and I do not wish to alienate people I like and respect.

 

I will not be testing the bees.

 

Regards, TAR

[StringJunky]

Whatever Theory,

 

You don't have to speak to String Junky that way, on my behalf. In fact you shouldn't speak to String Junky that way at all. In fact I should not have spoken to him the way I did.

 

He and Arete are long time, well respected experts on this board. If they are trying to keep a little order and sense in here, it is perfectly proper, and we should listen and learn and not let our "prides" get hurt.

 

After my initial reaction to String Junky's comment I have had time to think about it, and Arete's comment, and they are right. We, you and I are trying to make a silk purse out of a sow's ear. The horse is dead, and we are trying to beat it back to life. After many pages, we really have not gotten anywhere. I am of the mind to give it a rest, for now.

 

I still like your method. I still like my additions to it, and the hue number determination. But we are about done here, and I do not wish to alienate people I like and respect.

 

I will not be testing the bees.

 

Regards, TAR

+1 for seeing the light. _As flattering as your comment is, I am no more than you are in the expert stakes. But I hope you've seen enough of my posts over the years that I don't post like that often and when I do, I am sincere... it was clear that the thread had it's own momentum with no facts to support its continuation. The idea likely has application elsewhere, and it is very interesting how colours are made up in computers of itself.

[whatever theory]

No body is helping "me"

This is not about me. I will always remain anonymous and I am no scientist, so I have absolutely nothing to gain other then knowing that I made a contribution to society.

If this can help science, then science is what you are helping by figuring out a proper technique, not me.

I have already proven this to myself, I do not need any more proof.

The proof is in the pudding and for every argument against this, I could provide 100 pictures that show proof that this is true. You can see it all around you everyday and that should be proof enough in itself.

I feel like I have already done my service here and I know that there are some people out there that get what I am saying and some may put this to use in their own field.

At the very least I know that everybody that has read this thread is now thinking and looking at nature in a whole new way, whether it be one way or another.

I am sure this thread will spawn many others like it in the future.

"Whatever" technique you decide to use or create is up to you. We have showed many examples on how to use this photograph technique and I am sure that others may come up with better techniques.

If you decide to match colors from one to another then there is a 1 in 16+ million chance of them matching.

You choose to use Tar's formula or if you want to combine all of these techniques together that might work to.

It is really up to the individual person to figure out how much proof do they need in the work that they are doing.

If only matching a few pixels is enough proof for you then just use that.

Well since Tar was the only one suggesting on how to repair the techniques that others did not like, I do not see how I am going to continue here, knowing very little about science and math.

So now it is up to you all to decide what happens to this thread.

"Whatever" happens here now, I will continue to search for answers.

I hope that you enjoyed watching and have learned something, like I have.

Take Care

Edited July 10, 2015 by whatever theory

[tar]

Whatever Theory,

 

Everybody here, helped you a lot. Challenging an idea is the best way to improve it, and make it more workable. That is my self imposed job on this board. To make good ideas work better, and to challenge ideas that don't work.

 

I have this theory on ideas. They spread like wildfire. Even in the old days, when all we had was telephone and ham radio, you would hear a joke from somebody, and then later that day hear the same joke from somebody that had no connection to the first party that you knew of that lived in another state. But the theory goes on to say that good ideas are instantly incorporated into everyone, that is exposed to them's thinking. And bad ideas are simply discarded as not workable. A good idea will come around again, strengthened and more workable. A bad idea will just go away. Because its not workable. It doesn't fit.

 

So test your ideas. If they work. Give them a tougher test.

 

Regards, TAR

besides...you are not the first to notice the importance of color, nor investigate how and why we see it, and how to make equipment that records it, and play it back

 

A lot of good ideas and thinking about color had to have happened BEFORE you put your cursor on a pixel and saw a number...don't you think?

[Strange]

I have already proven this to myself, I do not need any more proof.

 

And that attitude is why you are not a scientist.

[MonDie]

There should be standardized lighting and camera specs for biological photography. It would simplify color comparisons greatly.

Edited July 10, 2015 by MonDie

[StringJunky]

There should be standardized lighting and camera specs for biological photography. It would simplify color comparisons greatly.

You have to ensure that you preserve all the pixels, when copying, as well. Many formats use or are set for lossy compression..

[tar]

String Junky,

 

That is an excellent point. As I am seeing a pattern develop, looking at pixels, I am not necessarily looking at the pattern in the animal or plant I took the picture of. I could be picking up a pattern imposed by the image processing of the camera or the computer building the screen presentation on the computer, not to mention the potential influence of a rebuild of the data, after being compressed for efficient transmission and storage as you so aptly point out.

 

If one was to actually study the pixels for their biological meaning, one would probably NOT want to do it on a smart phone, where the meaning behind the pixels has been processed and may well not contain the original information.

 

If one wanting to study the meaning of the color presented by the original subject, and one had the choice, they should go with the largest uncompressed image available, and suffer the longer download time, and larger storage requirement to gain a more accurate picture of the wavelengths absorbed by the subject, without picking up the characteristics of the lossy compression systems.

 

For instance lets say the compression system looks at an area, finds an average WT-TAR R/G hue value for the area and just stores and transmits that number along with a short description of the limits of the area. Then at the other end, during the rebuild process the program knows to fill that area with randomly generated pixels that fulfill a normal distribution around a transmitted avg RGB pixel, yielding an area that "looks the same hue" but might not have a single pixel that is of the same value in the same position as it was on the subject. In such a situation it would not be surprising that you could deduce a pattern of an average hue that all the pixels distributed around, because that is exactly what the decompression algorithm created for you. In short, with the loosy compression, the exact arrangement of data is sacrificed for a shorthand version of the data and the aspects of the original data that you find the most important might not be faithfully reproduced. You might wind up making a claim about the pattern extant in the subject that is actually an artifact of the system that brought the image of the subject to you.

 

Regards, TAR

[StringJunky]

String Junky,

 

That is an excellent point. As I am seeing a pattern develop, looking at pixels, I am not necessarily looking at the pattern in the animal or plant I took the picture of. I could be picking up a pattern imposed by the image processing of the camera or the computer building the screen presentation on the computer, not to mention the potential influence of a rebuild of the data, after being compressed for efficient transmission and storage as you so aptly point out.

 

If one was to actually study the pixels for their biological meaning, one would probably NOT want to do it on a smart phone, where the meaning behind the pixels has been processed and may well not contain the original information.

 

If one wanting to study the meaning of the color presented by the original subject, and one had the choice, they should go with the largest uncompressed image available, and suffer the longer download time, and larger storage requirement to gain a more accurate picture of the wavelengths absorbed by the subject, without picking up the characteristics of the lossy compression systems.

 

For instance lets say the compression system looks at an area, finds an average WT-TAR R/G hue value for the area and just stores and transmits that number along with a short description of the limits of the area. Then at the other end, during the rebuild process the program knows to fill that area with randomly generated pixels that fulfill a normal distribution around a transmitted avg RGB pixel, yielding an area that "looks the same hue" but might not have a single pixel that is of the same value in the same position as it was on the subject. In such a situation it would not be surprising that you could deduce a pattern of an average hue that all the pixels distributed around, because that is exactly what the decompression algorithm created for you. In short, with the loosy compression, the exact arrangement of data is sacrificed for a shorthand version of the data and the aspects of the original data that you find the most important might not be faithfully reproduced. You might wind up making a claim about the pattern extant in the subject that is actually an artifact of the system that brought the image of the subject to you.

 

Regards, TAR

That's it. This is the second major weak link in the idea ... the first one being the actual intraspecific variation within a species, as Arete explained. Preserving the signal chain wouldn't be a trivial matter in this sort of exercise and is one that you definitely cannot ignore.

[tar]

Like and RGB value. A subject doesn't "have" an RGB value, we assigned it one.

Thread,'

 

Looking for the answers in the photos displayed on a smart phone might be like looking for flawed steel girders in a bridge by finding the cracks in a model of the bridge. The one does not properly represent the other. The one looks like the other, but the model isn't even made of steel much less have a flaw in the wood in the exact spot that the flaw in the steel is.

 

However, there may be some useful information obtained by the camera, after the light from the image has gone through filters and processes. The light absorbing molecules in the subject still may give themselves away based on what the camera records. And our technology is pretty good, wonderful engineers and designers have be working on ways to get that picture on the screen looking just like the real thing, in the fastest times. So even though you are going to have to loose something with an analog to digital conversion and the image processing and "optimization" and so on, there is still a lot retained. After all, you can still look at the thing and say its an oak tree. There might be enough of the original data represented in the image to work with it, still.

 

Regards, TAR

[Klaynos]

There should be standardized lighting and camera specs for biological photography. It would simplify color comparisons greatly.

That's almost impossible. Even under the sabre lighting conditions you can find species that will go all the way from under to over exposed due to reflectivity differences with any digital camera. It's not such a problem for human eyes because the sensors are dynamic and non linear therefore changing the "camera settings". I've said it several times above this idea is fundamental flawed.

[tar]

Klaynos,

 

Fundamentally flawed, but not because of shade. I showed the glare and shadow did not matter, and you have a particular hue, a particular two color ratio in a band on a bee....or I might have showed that, had we done the 36 point check. Arete said that was a moot point though, since we would find a range of hues within the same species, that would overlap with the range existing in another species.

 

The problem is the color of the light illuminating the subject, and the processes in the camera, network, computer and screen that arrive at an RGB value for a pixel in the first place. The R/G value will change if there is just a little more short wavelength light in your flashbulb then in mine.

 

Regards, TAR

[Klaynos]

Klaynos,

 

Fundamentally flawed, but not because of shade. I showed the glare and shadow did not matter, and you have a particular hue, a particular two color ratio in a band on a bee....or I might have showed that, had we done the 36 point check. Arete said that was a moot point though, since we would find a range of hues within the same species, that would overlap with the range existing in another species.

 

The problem is the color of the light illuminating the subject, and the processes in the camera, network, computer and screen that arrive at an RGB value for a pixel in the first place. The R/G value will change if there is just a little more short wavelength light in your flashbulb then in mine.

 

Regards, TAR

The exposure will effect the response of the CCD or CMOS. You can't avoid that. There are physical problems caused by changing the amount of light on the sensor. Even having the same object with the same light source will rarely give the same response on the same sensor after you've moved the object. This gets even worse when you start using different areas of the same array or different sensor arrays. You cannot do normalised spectroscopy using digital photography.

[MonDie]

Klaynos, I can't follow you. Could you introduce us to some of the underlying concepts? I recall that Cambridge in Colour was a good online resource for cameras and lighting.

Edited July 11, 2015 by MonDie

[tar]

Klaynos,,

 

But if you have an absorption band at 620 nm and you shine white light on the subject, the hole is still going to be at 620 nm, it does not matter how much white light you shine on the subject

 

Regards, TAR

or how wide your aperture is open 235 135 35 is still the same hue as 200 100 0

those numbers are not supposed to show what a hole at 620nm would look like in RGB. I will have to figure that.

[Klaynos]

Klaynos,,

 

But if you have an absorption band at 620 nm and you shine white light on the subject, the hole is still going to be at 620 nm, it does not matter how much white light you shine on the subject

 

Regards, TAR

 

or how wide your aperture is open 235 135 35 is still the same hue as 200 100 0

 

those numbers are not supposed to show what a hole at 620nm would look like in RGB. I will have to figure that.

Yes but your sensor isn't looking at 620nm it's got 3 color filters per pixel. If you start over loading any of them you make it even harder to distinguish between colours. That's why you need to use a spectrometer.

 

I'd suggest reading about how CCD cameras work as applied to color spaces and then extending that to over and under exposure. The killer is that they're linear with quite a small range of how many photons they need to kick enough elections to work.

[tar]

saw a chart that had red LEDs sensitive to light in the 645to620 nm range and red-orange LEDs in the 620.5 to 635 range so whatever the system of LEDs sensitive to certain wavelengths or LEDs sensitive to wide ranges of wavelengths with color filters infront of them, if the subject is absorbing 620nm wavelengths which is sort of a slightly orange red, the RGB value is going to be high in blue low in red, and sort of low in green. Lets say R32 G64 Blue 228. This is the same hue as 0 32 196 which has an WT-TAR value of B/G 6.1 More exposed you might get 52 84 248 but that is still the same hue with a blue to green ratio of 6.1 to 1

my mistake I looked at LEDs and thought CCDs. You are right, its done with filters infront of CCDs. I will have to think out that linear aspect. Overexposed is probably an issue because 255 255 255 wouldn't tell you a darn thing.

once any one of the CCDs, like if its the blue in my example, is maxed out, more light will reduce the B/G ratio, as adding 50 50 50 to 52,84,248 gives you 102, 134, 255 (no 298 possible) which is B/G 3.8

 

So Klaynos, you are absolutely right. Over exposer is a big issue.

[MonDie]

Here is what they mean by under/over exposure.

http://www.cambridgeincolour.com/tutorials/camera-sensors.htm

 

The problem is that moving toward brightness or darkness will drown out the hue. Hue reflects the relative strength of different wavelengths, but these wavelength proportions still exist even in things we see as white or black.

The actual camera sensor, however, only sees in RGB (red green blue).

http://www.cambridgeincolour.com/tutorials/camera-sensors.htm

I think TAR is talking about cells on the sensor getting maxed out (at 255).

[Klaynos]

Here is what they mean by under/over exposure.

http://www.cambridgeincolour.com/tutorials/camera-sensors.htm

 

The problem is that moving toward brightness or darkness will drown out the hue. Hue reflects the relative strength of different wavelengths, but these wavelength proportions still exist even in things we see as white or black.

The actual camera sensor, however, only sees in RGB (red green blue).

http://www.cambridgeincolour.com/tutorials/camera-sensors.htm

I think TAR is talking about cells on the sensor getting maxed out (at 255).

With the caveats that you also have noise and leakage around the two ends and the individual sub pixels will not all be identical in their response. Both on a single pixel and compared between pixels.