The "Whatever Theory" Identifying The World...

[studiot]

Pearls, with your full-on approach, you are missing something many here are trying to tease out.

 

Whatever may be wrong in the speculation about linking colour in photographs to specific geographic locations (I think so) but the study of perceived colour can still bear much fruit, especially if conducted scientifically.

 

My approach is to nudge in this direction and try to find out if there are any more modest goals with real scientific value.

 

Edited July 2, 2015 by studiot

[tar]

Whatever Theory,

 

The 6 x 6 grid idea was not considering neighboring pixels but conceived as a way to select 36 widely separated pixels, without "matching" them.

 

In your process, you look around the area with your dropper (what specific program is this?) and mentally notice the pattern and ratio as the numbers go up in the light areas and the numbers go down in the dark areas, but then the cherry pick happens. You choose the number you think is the most representative, and then go find this same number again, to verify that it is this species number. But you just ignored the other 100 triplets in the same picture, that could also be matched within the picture.

 

The 6 x 6 grid idea was designed to take YOU out of it. This, so anybody else could take the same picture, randomly select 36 pixels from all over the sample area (without "trying" to match anything) and average the numbers in an agreed upon way, to arrive at "the" number that best represented the sample area. I do not know if the method would work. Maybe you need more pixels or a different way to calculate an average. Maybe the averaging idea makes no sense, and you are coming up with a good number, by some subconscious calculation you don't even realize you are doing. But the way to check it out, is to arrive at some mechanical process that takes YOU out of it, and arrives at a number. Maybe your paint program already does something like this for us and you can just ask the program for an average pixel in a given area. I don't know, nor do I know the process it would use to figure such. So I suggested that you come up with a mechanical way to arrive at the number, that takes no human judgement to arrive at it.

 

The pixels that would be selected by the 6 x 6 grid would not be next to each other. Say you are looking at a square inch of picture at 300x300 resolution and you thought of it as rows and columns of pixels. You would select the pixel from the 25th column and the 25th row, and the pixel from the 25th column and the 75th row.... The selected columns would be 25th, 75th,125,175,225,275 and the rows would be 25th, 75th,125,175,225,275. My thought was not that you had to figure an exact pixel, but that you would conceive of a 6x6 grid the size of the sample you where testing and select a random pixel from each of the 36 smaller squares that the grid would lay out. Taking the pixel directly under a dot (corner where four grid boxes met) however, would avoid the possibility that you would be "selecting" a certain collection of pixels subconsciously (like always picking a mid range pixel from the group).

 

Regards, TAR

This "ratio" thing though, we need to fully explore.

 

There may be a way to describe it, and then this description would be the "number".

 

There would be less than 16 million such descriptions, as there would be all the different shades of the same tint, and the ratio description would just describe the tint and not the shade.

 

I am thinking though, in terms of IDing the species, the shade gives us additional information, as to the surface characteristics of the species, so it would not be good to loose the average shade of the tint, which you would lose, with a ratio description...so, nevermind. The ratio, if we were to describe it, might be helpful in determining what pigments are present, but we would lose some contour and surface information. Which is probably important in feathers and such.

And how exactly to handle iridescence and soap bubbles and photonic crystals and peacock feathers and the like, is still up for grabs.

Whatever Theory,

 

The pink pattern on the paper towel might affect your numbers. Pink is a light shade of red. If the flash is putting out white light, the pink design has pigments in it that are absorbing green and blue (cyan) light. Therefore, you don't have 100 percent white light bouncing around the inside of the box, your pink design is soaking up a little green and a little blue light, so my guess would be that a subject, say a grey subject that would read Red 50, Green 50, Blue 50 if the design was not on the paper towel, would read Red 50, Green 49, Blue 49, with the pink design in the box.

 

Regards, TAR

This is however a good test of any system you come up with to arrive at a number for the subject. If you put another object in the box, that is absorbing certain wavelengths of light, it should affect your numbers, accordingly. Or if you have come up with some consistent numbers using the paper towel with the pink design on it, does using a white sheet of paper, instead, change your numbers.

I wonder if you should not put multiple objects in the box as what light the one absorbs will affect what the other will reflect. I am not sure though if it makes a difference if the other objects are absorbing the same colors...I would have to think that out.

Which brings up another issue. Does the one color on a multicolored subject affect the color of a neighboring area of a different color. Do we want this difference in our number, or do we not want this difference in our number. If we want it, we take the samples as they show. If we don't want it, we have to cover the subject with a white mask and take a picture only of the target colored area.

Edited July 2, 2015 by tar

[Strange]

I am not trying to disagree with you but, I can not figure out what you mean by, cherry picking.

 

There are two parts to this: (1) finding things that match (and ignoring those that don't) and (2) not checking that the matches are unique to that species.

 

1) You are looking for photos you can analyse that will confirm your idea (rookie mistake No. 1). In addition to this being the wrong approach, there may well be some (subconscious) selection bias going on: you look at photos (or objects to photograph) and think "that's a good candidate to confirm what I think". So you shouldn't choose the photos/objects yourself (rookie mistake No. 2) you should get someone else to choose them for you.

 

This will produce results that falsify your theory, as already shown by the simple example of carrots. But you have chosen to ignore that contradictory evidence (rookie mistake No. 3).

 

2) You are only looking for things that confirm your idea (rookie mistake No. 4). You claim to show that the colours of two examples of a given species match, but you must also show that the colour (or colour range) is unique to that species. Without that test, your experiments are pretty much meaningless in terms of confirming your hypothesis. You need to show that no other species has the same colour(s) as each of the species you look at.

 

These basic flaws in methodology are on top of your almost total lack of knowledge of colorimetry, colour spaces, how colours are generated by animals and plants, how the human perception system works and what its limits are, and so on (rookie mistake No. 5).

 

Finally, despite the fatally flawed methodology, evidence that falsifies your idea and your lack of even basic knowledge, you are still absolutely convinced you are right (rookie mistake No. 6, and the one that confirms you as a crackpot).

 

To address that last point, answer this question: What evidence would convince you that your idea is wrong?

 

Having answered that, the next question is: Why aren't you devoting all your effort to finding that evidence instead of wasting your time (and everyone else's) looking for yet more confirmations?

 

https://en.wikipedia.org/wiki/Dunning%E2%80%93Kruger_effect

Also I am not saying that identification can be made simply by color, but you must have a good idea of what kind of species you are looking at, whether it be carrots or pearls.

 

I thought that is exactly what you were claiming.

 

If you are not saying that, can you state clearly what your hypothesis is?

 

You have to know it is a carrot before you can find it's specific species

 

So, first you have to identify it as a carrot (Daucus carota). Then you can use colour to ... what? Decide what colour carrot it is?

 

This doesn't sound very insightful. Certainly not the "huge benefit" you initially claimed.

[tar]

Whatever Theory,

 

Think as if you are building an app, for a smartphone, that would tell you the species you are looking at, just by taking a picture of it. You would need a database of data for the picture to compare itself against. What would be the nature of the data in the data base and what would be the nature of the data you would extract from the picture to compare against the data base. Have you designed the system in such a way that every species is always positively IDed? When would the system NOT properly ID a species? How do you fix this situation so that the system would positively ID the subject? What percentage of false positives would you accept? What can you do to decrease your projected amount of false positives? Does your system ID a species correctly enough times that a child would buy the app? An adult? A scientist studying little know species of wild flowers in the Andes?

 

Regards, TAR

Does such an app already exist? Is it faulty in any way that you can correct with any of your ideas?

Whatever Theory,

 

One thing I learned about scientific investigations is that if you have a good idea, and its a workable idea, someone has already thought of it. Not only already thought of it, but built a whole area of study around it, and investigated it from many different angles. And not only is it already being studied, but there is already standards and systems built and used in the field. And not only has someone thought the workable thought before, but they thought it, and developed it, and shared it 600 years ago.

 

I like your idea, and I think it has possibilities, but you are not the first to study color, you are not bringing wonderfulness to an ignorant world, Mike Smith Cosmos is not your team mascot.

 

You are on a science board, and people are trying to harness your energy and creativity and industrious nature, and get you to run in a sound manner. As opposed to running around in circles.

 

Mike Smith Cosmos and I are in the later 1/3 of our lives. We would love to present the world with a gift. Some thought or picture or direction that would make life better and more enjoyable for people. We would love to answer unanswered questions or show people how wonderful this or that aspect of the universe is. But we have had these desires for years, long before you came on the board. You are not our captain though. We are not following you. We are guiding you.

 

Regards, TAR

Edited July 3, 2015 by tar

[whatever theory]

 

 

Pigments are not the only natural colour, there's also photonic crystals.

 

In all of this you are yet to demonstrate how you could possibly normalise two different photos. The concept is fundamentally flawed.

I have read the Wiki definition of photonic crystals and I really can not comment much on this matter. I will have ponder on the subject a bit before I can make any comments on this.

I have thought of a test that I could perform in my black box to test this. I was thinking about taking a picture of the same subjects a few times and then comparing them to each other. I can put a number 1,2 and 3 in each photo so that you know I am not just using the exact same picture over again. Please let me know what you think.

 

 

 

My approach is to nudge in this direction and try to find out if there are any more modest goals with real scientific value

Thank you for your comments, I need all of the nudging I can get.

I have not forgot about trying your resolution tests, but it may take a few days, because I am trying to explain my techniques and catch up with "cherry picking" comments right now.

 

Tar, thank you for taking the time to explain exactly what I am doing wrong and for providing ways that may help people take this more seriously. I was going to do as you suggested, but I just want to catch up on comments and finish showing how I have been doing my techniques now. Hopefully by tomorrow I can prepare a presentation with the data collected n the ways that you suggest.

 

 

The 6 x 6 grid idea was not considering neighboring pixels but conceived as a way to select 36 widely separated pixels, without "matching" them.

 

 

And how exactly to handle iridescence and soap bubbles and photonic crystals and peacock feathers

I am not totally sure about soap bubbles, but is soap natural in origin? There is something called overtones on pearls. This is not the actual color, of the pearl but a group of colors that appear over the color in certain lighting conditions, these overtones do look very similar to the iridescence of soap bubbles. Even though many different pearls have overtones which usually range from pink to blue I have found that these shades of pink and blue are also unique to their species. So I would think that 2 different kinds of soap may be the same way. Just a guess. I will try to do some research on soap bubbles.

I quess I missed the question about Peacock feathers also. Can you please tell me what the question was?

 

 

The pink pattern on the paper towel might affect your numbers. Pink is a light shade of red. If the flash is putting out white light, the pink design has pigments in it that are absorbing green and blue (cyan) light. Therefore, you don't have 100 percent white light bouncing around the inside of the box, your pink design is soaking up a little green and a little blue light, so my guess would be that a subject, say a grey subject that would read Red 50, Green 50, Blue 50 if the design was not on the paper towel, would read Red 50, Green 49, Blue 49, with the pink design in the box.

Regards, TAR

This is however a good test of any system you come up with to arrive at a number for the subject. If you put another object in the box, that is absorbing certain wavelengths of light, it should affect your numbers, accordingly. Or if you have come up with some consistent numbers using the paper towel with the pink design on it, does using a white sheet of paper, instead, change your numbers.

I wonder if you should not put multiple objects in the box as what light the one absorbs will affect what the other will reflect. I am not sure though if it makes a difference if the other objects are absorbing the same colors...I would have to think that out.

Which brings up another issue. Does the one color on a multicolored subject affect the color of a neighboring area of a different color. Do we want this difference in our number, or do we not want this difference in our number. If we want it, we take the samples as they show. If we don't want it, we have to cover the subject with a white mask and take a picture only of the target colored area.

I never thought of this before.

I was thinking that while I do the test, I was just describing to Klaynos, I could perform this same test, in a series of 3 pictures, and put something colorful in each different picture to see if the colors in the subject change from picture to picture. Please let me know what you think.

Thanks again for all of the thought that you have put into this thread.

 

Thanks Strange for explaining what you meant.

Sorry it is taking me so much time to understand a lot of the lingo and concepts of modern science.

I read recently, but can not find it now, a Senior member was saying that no theory is ever really proved and never goes beyond a theory. This statement has had me puzzled seance then for a number of reasons, but now that I understand the sciences definition of confirmation bias I can see why. Because you will never really be able to prove something 100%

The only thing I can think of to compare what I am doing, is to DNA. I know that my wife and I have never been DNA tested so does that mean that the DNA theory is not yet excepted? Because not every human was tested, to make sure that we all share the same DNA?

And when they test a persons DNA do they test every strand of DNA in that person or just maybe a few from the hair or fingernail?

So at what point does a speculation turn into a theory?

I am sorry that you all have to put up with someone that is as non-scientific as me, and believe me I dreaded opening this thread for a long time, because I really know nothing about science. But at the same time there may be others that have good ideas, that are like me, that may come along in the future also.

I think I may now understand what you and Tar are saying though. Like testing DNA they do not test every strand but, they make sure that the test that they are taking is a good accurate sample that represents the rest of the strands. By me only picking one pixel I may not be getting an accurate idea of what that area is.

 

 

 

This will produce results that falsify your theory, as already shown by the simple example of carrots. But you have chosen to ignore that contradictory evidence (rookie mistake No. 3).

Did I miss something about Carrots? What was the contradictory evidence.

It was my mistake for bringing up carrots as an example. What I should have said, and what I will say from now on, is Natural (Wild) Carrots. There has been a lot of talk about carrots lately and most of the Carrots we buy are cultivated and not Wild.

I do not think I have falsified anything here though.

 

 

 

2) You are only looking for things that confirm your idea (rookie mistake No. 4). You claim to show that the colours of two examples of a given species match, but you must also show that the colour (or colour range) is unique to that species. Without that test, your experiments are pretty much meaningless in terms of confirming your hypothesis. You need to show that no other species has the same colour(s) as each of the species you look at.

I guess my ideas about proving/disproving are different then that of science.

To me, I have always felt like if you are trying to prove something, then if you do not succeed, you have really disproved it.

Say I was arguing with someone over a flower being pink color with someone.

If I can prove that the color is pink by holding a pink crayon next to it then I am right. But if the color is not pink and it is yellow and I hold a pink crayon next to it then I am wrong. If someone else holds a yellow crayon next to it first then they are right and I am wrong.

So I will try to look at things from your point of view but I am still uncertain exactly what you want me to do, to an extent.

In the example of the meat test I showed first the differences in color from Beef, Pork, and Lamb then the next test I showed how each animal shared the same colors with its own species. But I can not do everything in every picture.

In the case of the leaf, flowers, and purple pearl test I was able to compare the species with each other but I also compared them to the other species also.

 

 

 

These basic flaws in methodology are on top of your almost total lack of knowledge of colorimetry, colour spaces, how colours are generated by animals and plants, how the human perception system works and what its limits are, and so on (rookie mistake No. 5).

I believe in what is said in the little trailer line on Tar's Profile "There is not a one of us that knows more than all of us put together.

I think that most theories are worked on for a long time before they present all of their findings and theory, but then science picks it all apart and they have to go and start over because there was some basic flaws in the techniques for proving it. So if I would have gone this route and spend a long time testing everything in nature and then presenting it the ways I am doing it, it would be all for nothing and then I would have to go and start over.

I am trying to go this route because I would rather have science pick it apart now and help fix all of the problems along the way so that when we reached the finish it is all done and no one can pick it apart.

I know I lack a lot of knowledge in a lot of areas but no matter how much I study Tar's statement is always going to remain true and someone will always have some information that you are missing.

 

 

 

Finally, despite the fatally flawed methodology, evidence that falsifies your idea and your lack of even basic knowledge, you are still absolutely convinced you are right (rookie mistake No. 6, and the one that confirms you as a crackpot).

This is not true Strange. I have admitted in this site that I may be wrong and I have not tested everything in nature yet. This is only my speculation, but it is supported by other Scientific fields, as you have previously posted links to.

 

 

To address that last point, answer this question: What evidence would convince you that your idea is wrong?

Having answered that, the next question is: Why aren't you devoting all your effort to finding that evidence instead of wasting your time (and everyone else's) looking for yet more confirmations?

I am just answering this, given what I think, whether it is wrong in the eyes of science or not. Just trying to describe my thoughts...

What would convince me is that if I can not match the colors, but since I keep matching them I have not convinced myself that it is wrong. If one day I can no longer match colors then I would prove myself wrong. So for me I do not know how to prove it to myself other then keep testing.

I hope you can help me figure out what you think I should do.

 

 

I thought that is exactly what you were claiming.

If you are not saying that, can you state clearly what your hypothesis is?

No Strange. Maybe in my remarks some where I said I was trying to identify species by color, but no where did I ever say that you could do this with color alone. For instance if it is a pearl then you need to make sure first that it is a real pearl with all of the normal tests, xray, etc. Then once you know it is a pearl you can identify the species which has not yet been figured out in the pearl industry.

If you are testing a "wild" Carrot because you want to collect its genetic color for future for breading purposes or to compare it to other "wild" carrots then you would need to know that it is a carrot or at least a vegetable before you tested its color. I will try to rewrite my hypothesis and post it soon.

 

 

So, first you have to identify it as a carrot (Daucus carota). Then you can use colour to ... what? Decide what colour carrot it is?

This doesn't sound very insightful. Certainly not the "huge benefit" you initially claimed.

Let us say that you were a scientist exploring the amazon for medicinal plants. Some one has previously found a cure for cancer there, in a pant, and is sending you back there to find more. The original person found the genetic colors of this plant and you have the same camera with the same box that he did. There are many plants that appear to look very similar and some are poisonous. You already know you are looking for a plant so you have already categorized this thing as a plant.

So even though you can identify that something is a plant does not mean that you can identify each species with only your eyes.

You also would not want to test any lizards even though they may be the same color as the plant you are looking for.

So that is were this may become beneficial. Everyone is already able to see if something is a bird, or a lizard or a fish, but what they may not know is each of these things exact species just by looking at them. Hope this all makes seance.

 

In my most previous post I am only explaining how I have been doing this technique up until now and although I may change the technique based on everyone's suggestions, I am only trying to show how I have done things thus far.

 

If anyone has pictures I would be happy to test theirs, but it needs to be in good conditions as I have explained before. Many of the pictures on the web are not taking into consideration my technique and can not be used.

 

 

A part of the technique that I have not yet touch on much is how to try to find the color range of a certain area.

In the picture below I selected a random pixel from what appeared to be the darkest brown from the first snake which are represented by red, blue and green dots. Then I took an additional samples from another brown spot on the same snake (pink and yellow dot)

Then I chose 3 spots randomly from the second snake (Dark green, Dark blue, Orange Dots)

The areas I tested were small so I just put the dots around the edges so I did not cover up the area.

Originally when I took this picture my eyes were playing tricks on me and I did not realize until after I started testing it that the lighting and angle of the subjects is not so good, but I continued testing it to show an example of this technique. Normally I would have tried to match the second snakes color, but in this case all of the areas tested were completely random as Strange and Tar's suggestions.

Maybe something like this may be a better way of trying to find a specific color range of an area.

Eager to here what you all have to say.

 

 

Mike Smith Cosmos and I are in the later 1/3 of our lives. We would love to present the world with a gift. Some thought or picture or direction that would make life better and more enjoyable for people. We would love to answer unanswered questions or show people how wonderful this or that aspect of the universe is. But we have had these desires for years, long before you came on the board. You are not our captain though. We are not following you. We are guiding you.

I was just joking about Mike being a Mascot. I know I am not the captain and I am not trying to act like one. So sorry if it appeared that way.

I am trying to follow everybody advise the best I can. I feel like I am just trying to steer the ship and all of you are the captain.

I appreciate all of your help and I hope I did not upset you Mike. But like a Mascot you are helping to lighten up the mood. That's all I meant.

 

 

One thing I learned about scientific investigations is that if you have a good idea, and its a workable idea, someone has already thought of it. Not only already thought of it, but built a whole area of study around it, and investigated it from many different angles. And not only is it already being studied, but there is already standards and systems built and used in the field. And not only has someone thought the workable thought before, but they thought it, and developed it, and shared it 600 years ago.

This is one of the main reasons I have come here. I do not want to reinvent the wheel, becuase I did not know it already exists.

As Strange and others have pointed out in other fields are doing similar things, but from what I can tell most of it is in very generic terms. For instance I was recently on a tropical fish site and they used the colors of the fish to represent the species, but they were not the deciding factors.

It read something like, this fish will have blue, red, purple and yellow coloring.

Or in another site I saw how some bird people were describing the birds based on feather color but only again in very generic terms such as red, green blue.

I have found were there are some horse and dog hybrid breeders that test there breeds colors to a paper color swatch. But it is only for the coats.

In the case of pearl identification the only importance that they have yet placed on color is a way to identify what color it appears on a normal color wheel. They have yet to see any importance in color.

So if somebody knows if someone is already doing and saying what I am then please tell me so I do not waste any more of my time.

Thanks again Tar for your comments

Edited July 3, 2015 by whatever theory

[StringJunky]

....

So if somebody knows if someone is already doing and saying what I am then please tell me so I do not waste any more of my time.

I think you should stop now, you are wasting your own time. It is completely flawed because colour within any species is too varied and governed by too many endogenous and exogenous variables. Add the variables of equipment and techniques, it becomes a futile and meaningless exercise.

[Strange]

I read recently, but can not find it now, a Senior member was saying that no theory is ever really proved and never goes beyond a theory. This statement has had me puzzled seance then for a number of reasons, but now that I understand the sciences definition of confirmation bias I can see why. Because you will never really be able to prove something 100%

 

This doesn't really have anything to do with confirmation bias. But never mind.

 

So at what point does a speculation turn into a theory?

 

When there is overwhelming evidence replicated by many people in different ways. So far you have no evidence and it isn't even clear what your hypothesis is.

 

I think I may now understand what you and Tar are saying though. Like testing DNA they do not test every strand but, they make sure that the test that they are taking is a good accurate sample that represents the rest of the strands. By me only picking one pixel I may not be getting an accurate idea of what that area is.

It is not to do with picking one pixel (but that doesn't help). It is because you are not doing anything to show that the colours you see are connected to that species.

 

For example, lets say someone develops a hypothesis that you can identify a species by the number of legs. They show you a picture of two people: "look they both have two legs". Then they show you a picture of two horses: "look they both have four legs, therefore my theory works".

 

This is exactly what you are doing (apart from the fact that there is much more variability in colour than number of legs).

 

You would point out to your friend that dogs have four legs and they are not horses. "Ah, but first you have to make sure you are looking at a horse, then count the number of legs".

 

Did I miss something about Carrots? What was the contradictory evidence.

 

Your original claim was that a species (in a single location) has a unique colour. Carrots (like many other species) have a range of colours.

 

It was my mistake for bringing up carrots as an example.

 

You didn't. Others did, in order to show that your idea doesn't work.

 

What I should have said, and what I will say from now on, is Natural (Wild) Carrots.

 

Ah, you have added the "No True Scotsman" fallacy to the list.

 

So your method doesn't work with all species, just the species it works with...

Say I was arguing with someone over a flower being pink color with someone.

 

I guess my ideas about proving/disproving are different then that of science.

To me, I have always felt like if you are trying to prove something, then if you do not succeed, you have really disproved it.

 

 

Science works by looking for counter-examples, not continually cherry picking successful examples. You are not doing this, therefore you are deluding yourself and boring everyone else.

 

So I will try to look at things from your point of view but I am still uncertain exactly what you want me to do, to an extent.

 

As you have changed your hypothesis but are unable to say what it is now, I can't help.

 

I am trying to go this route because I would rather have science pick it apart now and help fix all of the problems along the way so that when we reached the finish it is all done and no one can pick it apart.

 

That is a good attitude. However, ignoring all the flaws and contradictory evidence isn't really compatible with that.

So even though you can identify that something is a plant does not mean that you can identify each species with only your eyes.

You also would not want to test any lizards even though they may be the same color as the plant you are looking for.

So that is were this may become beneficial. Everyone is already able to see if something is a bird, or a lizard or a fish, but what they may not know is each of these things exact species just by looking at them. Hope this all makes seance.

 

There are well established ways of identifying species by looking at all aspects of their morphology: leaf shape, habit, where they grow types of flower, types of seed, etc. And, ultimately, DNA testing.

 

Oh yes, colour might come into this (particularly colour of flowers, very occasionally colour of leaves) because this can be, but isn't always, a distinguishing feature. After all, most plants have leaves of the same shade(s) of green. Do you know why?

 

Maybe you should try one of those online guides to identifying wild plants: they lead your through a series of questions. They will often give you the right answer even if you don't know what colour the flowers are.

 

Obviously colour can be significant, that is why there are so many species names such as aurea, alba, glauca, etc.

 

So if somebody knows if someone is already doing and saying what I am then please tell me so I do not waste any more of my time.

 

It sounds as if you have just listed a number of cases where people are doing what you say, but in a practical way (i.e. looking at the general colouration as part of the species identification).

 

But we can't really go on until you state exactly what your hypothesis is (in one or two sentences).

 

And then state, in advance, exactly what the exceptions are (e.g. no domesticated animals or plants) so that you don't kepp adding ad-hoc excuses when things don't go your way.

[studiot]

You still have not appreciated the implications of what I said about digital phtographs and programs.

 

Programs change the actual RGB values of pixels to suit their purpose.

 

I would add to Strange's list that error of continually introducing new material (examples) before fully dealing with existing ones.

With every point you introduce new photos and new goose chases.

Small wonder you are 'running round in circles'.

I posted that one single photo as it contains examples of almost all the effect I and others have mentioned, including the most recent that of the influence of adjacent colour.

 

 

But don't follow string junkie's advice and give up.

Reappraise and regroup and follow at least some of the sound advice given here.

 

Here is a good site to find out a huge amount about the vagairies of digital photography and colour.

 

They have lots of tutorials.

 

 

 

http://www.cambridgeincolour.com/color-management-printing.htm

 

Look up

 

Antialiasing

Subpixel rendering

 

https://software.intel.com/en-us/articles/conservative-morphological-anti-aliasing-cmaa-update

[StringJunky]

....

But don't follow string junkie's advice and give up.

What possible utility is there in trying identify to species by fine measurement of colours? I agree that he should learn the physics of colour and the vagaries of human vision in interpreting them but the idea itself is a fool's errand.

[studiot]

What possible utility is there in trying identify to species by fine measurement of colours? I agree that he should learn the physics of colour and the vagaries of human vision in interpreting them but the idea itself is a fool's errand.

 

 

 

I have already proposed more modest goals , and this was accepted.

 

 

Quote

 

 

My approach is to nudge in this direction and try to find out if there are any more modest goals with real scientific value

Thank you for your comments, I need all of the nudging I can get.

 

[StringJunky]

 

 

 

I have already proposed more modest goals , and this was accepted.

 

Ok, fair enough.

[tar]

Whatever Theory,

 

In looking at the snakes there are areas in the brown spots that are darker brown, like around the edges. There is also a shade difference due to lighting, as can be readily expected by the presence of the heavy glare on the left snake. My 36 point average system was an attempt to come up with an average tint and shade, that would define the color area in such a way as you might expect to arrive at the same number, should you test the same area again. With the markings on the snake being definite different shades at different areas of the marks, averaging the two areas together, would not be helpful, instead would represent a loss of important identifying data.

 

Your program needs to have rules decided upon as to what areas on the subject are to be considered areas of a "different" color, where a number for that particular color should be arrived at, that will exactly match that particular color anywhere on the snake or anywhere where it is found on any member of the species in the same area. For instance, you may need a number for the border of the mark, and a different number for the area of the mark, inside the border.

 

Which introduces another type of issue. Let's say one species of snake has brown markings that have a darker border of the same tint as the color in the center of the marking, and this is also true of species number two, but in species one the inside color is uniform, but in species two it is a very light shade in the center of the mark and gradually darkens as you get closer to the border. The "average" of the center color would be the same in both species, the average of the border color would be the same in both species, but you would have the same numbers listed as the color of both species, so you would not be able to ID the snake. Just say that its either this species or that species which is what you already knew and were hoping that the color ID process would resolve for you.

 

This is a different type of issue than if one species had exactly the same inside color and border color of their marks, but the one species marks were big and few and the other species marks were little and many. Or if the one species had round marks and the other diamond shaped. Being as there are optical character recognition programs, you should be able to handle the shapes and sizes of the marks, but how are you going to handle transitions of color, where the average color is the same color in two species, but the way the pigment is distributed is the defining factor?

 

Regards, TAR

Edited July 3, 2015 by tar

[Strange]

This is a different type of issue than if one species had exactly the same inside color and border color of their marks, but the one species marks were big and few and the other species marks were little and many. Or if the one species had round marks and the other diamond shaped.

 

Very good points.

We can compare a zebra, a panda, a skunk, a Dalmation dog and a Holstein cow, for example. (Maybe two of those don't count as they are not "natural".) Or maybe a Heliconius charithonia and an Idea leuconoe.

[tar]

Whatever Theory,

 

I just noticed a greenish area on the body of the left snake, over toward your pinky. This would indicate to me, by the direction that the faces of the scales that are greenish are facing, that there is an arbor over your head, or you are wearing a green hat (or shirt).

 

Sampling the colors in this area would more likely result in you being able to identify the location of the source of the dyes in your hat, than you being able to ID the snake.

 

Regards, TAR

 

Might be good to follow some white dress protocol, especially with species designed to blend in with their surroundings.

Or maybe R128 G128 B128 gray outfits only.

Whatever Theory,

 

Also in regards to my OCR comment and the snakes in #130, I noticed that the markings on the two snakes are identifiably different. For instance the mark on the back of the neck of the bigger snake might be IDed as a carrot shape, and the mark on the back of the neck of the little snake, might be IDed as a bowling pin shape.

 

To Strange's mention of cows and zebras and other species I did not look up to see their common names, I would have to add whatever species of whale that was I saw on a tour North of the tip of Cape cod. Each member of the species had its very own pattern of markings. The tour guides knew each whale, by sight. Going by their color though, they could have easily have been very large amputee Holsteins with reverse tracheotomies and faces misshappened by the accident, that just really loved the water off the cape.

 

Regards, TAR

Whatever Theory,

 

The two latin names that strange just gave are black and white buttflies.

 

Enough with the easy stuff. Show us the difference in the numbers between these two species.

 

Regards, TAR

Edited July 3, 2015 by tar

[tar]

Whatever Theory,

 

Here is another challenge for you. Something my wife brought home from the grocery store and something I brought in from the garden. They have similar looking colors, but are not of the same species. By the way, I don't think you are allowed to consider only natural species. Many of the species that the users of your app are going to be taking pictures of are from the grocery store, or from the garden. In both cases, some heavy human hands have been adjusting the species to our needs. There might even be some "tricks" the food producers have, to make the colors look more appealing to humans...so you have to have ways to identify artificially engineered species as well. Also another point brought up by another member, was that the pictures that the smart phone takes, may be "optimized" for color and contrast, or the numbers may easily be otherwise manipulated, in manners you might not be aware of, that would throw off the "meaning" of the numbers you are working with.

 

Picture taken early morning on a cloudy day with kitchen window behind me to my left, under an under shelf light with an unknown type of bulb. Taken with a Samsung S4 smart phone, with unknown settings.

 

Regards, TAR

whoops, I meant butterflies in a previous post...I was not trying to be gross or crude

Whatever Theory,

 

I just noticed a greenish area on the body of the left snake, over toward your pinky. This would indicate to me, by the direction that the faces of the scales that are greenish are facing, that there is an arbor over your head, or you are wearing a green hat (or shirt).

 

Sampling the colors in this area would more likely result in you being able to identify the location of the source of the dyes in your hat, than you being able to ID the snake.

 

Regards, TAR

 

Might be good to follow some white dress protocol, especially with species designed to blend in with their surroundings.

Or maybe R128 G128 B128 gray outfits only.

Whatever Theory,

 

Also in regards to my OCR comment and the snakes in #130, I noticed that the markings on the two snakes are identifiably different. For instance the mark on the back of the neck of the bigger snake might be IDed as a carrot shape, and the mark on the back of the neck of the little snake, might be IDed as a bowling pin shape.

 

To Strange's mention of cows and zebras and other species I did not look up to see their common names, I would have to add whatever species of whale that was I saw on a tour North of the tip of Cape cod. Each member of the species had its very own pattern of markings. The tour guides knew each whale, by sight. Going by their color though, they could have easily have been very large amputee Holsteins with reverse tracheotomies and faces misshappened by the accident, that just really loved the water off the cape.

 

Regards, TAR

Whatever Theory,

 

The two latin names that strange just gave are black and white buttflies.

 

Enough with the easy stuff. Show us the difference in the numbers between these two species.

 

Regards, TAR

 

Another picture of the same subjects taken with the under cabinet light turned off. Diffused, reflected and refracted rainy day sunlight through two windows, around assorted glasses and pots, only.

[tar]

...OK mascot, take off that goofy chicken head, put on some pads and a helmet and get on the field, I have an idea for a play that just might work.

 

Take one part magenta and 1.83 parts yellow and paint a patch. It should be the color of those snakes. Even paint the snakes if you are in the mood, with a little black for the darker areas and a little white for the lighter.

 

I took an average of the 8 colors in the snake post and got red 176 green 110 and blue 55. These numbers had Whatever's "ratios" of -66 and -55. The ratios are just how much more of the one color there is than the next. Transferring the RGB value into CMY we have Cyan 80 Magenta 146 and Yellow 201. Using the under color reduction principle we can take Cyan 80, Magenta 80, and Yellow 80 (a certain shade of black) out of that and get Magenta 66 and Yellow 121. This is equal to one part magenta and 1.83 parts yellow, or if you want to bring the tint up in word or paint or whatever, using RGB values, use 121 Red and 55 Green.

 

Regards, TAR

 

So the snake's RGB value, based on those 8 samples could be thought of as R/G 2.2

or in CMYK, Y/M 1.83

Edited July 4, 2015 by tar

[whatever theory]

Hello

I have been pretty sick so I could not post yesterday. Because I was also sick on the day I posted the Albino Anaconda test I did not do a very good job of explaining some things and the picture Was not so good for a couple of reasons. I posted it anyway, but the top snake had a lot of reflection on it and the bottom snake had a lot of shadows on in. If you notice most of the pics that I have taken from the web a in good lighting condition and the subjects are the same distance from the camera and are facing similar angles to the camera. I usually search threw many, many pictures before I find a good one from the net and in some cases like the Zonkey picture it is very difficult to find what you need all in one picture. So with this picture it was the only one I could find with mother Donkey, father Zebra and baby Zonkey all in the same picture. So I tested only areas that were all facing the same direction and that were in the sun light and not the shade.

As for your suggestion Strange, I could not find any pictures of a Zebra and a cow in the same picture and I do not think that this one counts.

 

The most important aspect, I think, besides what the species is, is geographic location. So this is always the first thing that you need to consider.

If you had an App or software this would always be the first things that you noted and categorized.

So in the case of a cow and a Zebra they do not live in the same area and would never be compared to each other.

I am still going to try to find a picture of these 2 side by side however.

Another thing that I need to emphasize is that this theory has a lot of potential, but you need to know how to use it.

A natural (wild) thing is something that man has had no intervention with and the rules that apply to it are going to be different then rules that others things that are not Natural follow.

So while this technique can identify species that are natural from one another you would not want to put them in the same category as non-natural things.

So if you wanted to go compare food at your local grocery store then most of that food is going to be not natural. You can still use it to identify what breeds were bred to create this new food and you may be able to even find out what farm it was grown or if it is natural in origin originally, but you need to use this appropriately in each case.

 

Strange when you argue: why is this technique important because we already have DNA I will list 4 reasons, I think,why this is better then DNA.

1. DNA is expensive (This is Free)

2. DNA Takes a long time, you have to mail it to the lab and then wait for them to test it then wait for the results. (My technique is fairly instant)

3. No body that I know has a DNA testing machine (Everyone already posses a computer and camera)

4. You can not test DNA from a picture.

From what I gather, color is too hidden in DNA for them to be able to find any patterns yet, so maybe this will help DNA scientist figure out the missing pieces to their puzzle also.

Of course there is going to be reasons that DNA is better for certain things then my technique but this is just 4 reasons that makes my technique better, in my opinion.

Also when you make the point: if you already know what something is then why are you trying to identify it.

DNA had to do this same thing at first. How would they know what the DNA of another human looks like if they did not test a human being first to find this out.

After they did this first, extensive testing on humans they no longer need to do this same extensive testing each new time because they already have a good idea of what they are looking for.

I am also sure that DNA as not yet tested every species on earth nor have they tested every strand of DNA on every subject before it was excepted by mainstream science, so why are you insisting that I do this.

By saying that this is not worth accomplishing because we already have DNA is not a good attitude in Science, as far as I can tell.

 

I do not think I can summarize this theory into one or two sentences. We are now on page 8 and I am still just trying to demonstrate the basics of my original techniques.

In the beginning of this thread I was trying to start out simple by only testing one color, but my original technique is to find the entire color range of a species, which as not changed at all, since first posting in the pearl forum many months ago.

 

Tar thank you for all of your thoughtful comments.

I did not take the picture of the snakes I just found it on the web. I see what you mean about a tint of green.

I was so sick that day I could not concentrate and I would have normally never chosen that picture because many of the conditions were not good.

The snake on the top had a lot of reflections and the snake on the bottom had a lot of shadows.

I did a simple test on the Watermelon and Cucumber and although some shades of green are similar I could not mach any of the exact colors to each other.

I agree that the most common color that will collide when dealing with plants will be green. That is why I demonstrated that there are many different areas to consider when testing plants, in the previous leaf test.

If you were to open that Watermelon up I am sure that the fruit and the seed colors would not match the Cucumber and If you were looking at the whole plant I doubt that the flowers, stems and leaves would all match either.

 

I like how you have formulated all of the colors on the snakes to find an average color..

I would like to talk to you more about it tomorrow, but I am going to now demonstrate (from my original technique) how I find the color range of a species.

Before I go on I would also like to explain that while doing the simple test, that I have been doing previously that is being accused of cherry picking, I try to stay in the middle of the first species color range with my first test.

Example, if there is a light shade of brown which turns into a dark shade of brown, then you want to take this first sample, somewhere in the middle, in between the light and dark shades. This gives you the best chance of matching that same color to the second subject.

In the instance of the snake, one animal may have a slightly darker shade of brown then the other subject from the same species.

At the same time this other snake may have a slightly lighter shade of this same brown color then the first snake.

So this lightest shade from the one snake would be the lightest shade of that color that could appear in this species and the snake that had the darkest shade would be the darkest shade of brown that this species could produce.

Every time you found a new snake in the future that had a lighter shade, then the one already know, you would extend that species color range to a lighter shade and you would do the same thing every time you found a new snake with a darker shade of brown.

If 1 of of every 1000 snakes produce say green spots then you would add a new color range category (for freaks of nature) and then in the future when ever one of these freaks would show up you compare it to this other green color from the previous freak snake.

 

So with the previous shown technique, were I am matching one color to another, this is just a fast easy way of comparing color to each other.Just for example purposes alone...

I like this technique and if you look at the number of possibilities that Tar has provided, I do not think the term cherrypicking makes any seance to me especially when you are testing more then one area. So I am still going to argue my case and I would really like to hear from all of the members, their thoughts on this.

Matching how many pixels from one photo to the next would make you happy to show a good example?. Should it be 2,3,4,5,36?

I think that even if you only match 3-5 it should be more enough to show a good example of something, which will allow us to move threw many different species at a fast rate.

But if the time comes when (if) we get serious about actually measuring the entire species colors range, after all of the problems with the technique are fixed and we have figured out a standard of controls that should be used, then at that point we do the following procedure on each species.

I do not expect to accomplish this task here, but if we figure out all the problems then everyone in their field can use this next technique to categorize (properly) the species that they are studying.

After you have the entire color range of a species the first time It will be easy for you to quickly test a few pixels of any other species, in the future, and compare this to the original range to see where they fall on your original color range chart.

The only time you change the original color range chart is when new shades of darker or lighter colors appear in new species you test, of the same species.

 

By using this color range technique I am being completely taken out of the equation and the color and the Red Numeric value determines which pixel is tested.

For instance, Below is the 2 species of Butterflies that Strange suggested I test.

Although these 2 Butterflies are both Black and White, there is something that definitely stands them apart from each other.

One has yellow in its wings, closer to its body and the other has pink spots near its body.

In this first picture I want to find the lightest shade of yellow and I also want to find the darkest shade of yellow in these same areas on both wings.

First thing I do is move my Dropper tool around, inside this area, and find the smallest RED RGB value that I can.

When you are convinced that you have the smallest red number then you do this same thing and find the largest Red RGB value you can find in that area.

Now that you have your smallest and biggest Red RGB value, you can find a Red RGB value that is somewhere in between the smallest and biggest, so that you can appreciate the pattern that develops within this series of 3 numbers.

If you were studying a certain species, the more of these reading you took the better idea you would get of the full pattern that develops and will give you a better chance of fitting future species into this range.

 

If you notice the red,green and blue dots in the photo represent this yellow color range of the creature on the left, and the yellow, light blue and pink dots represent this yellow color range of the creature on the right.

Notice the pattern that is happening between the colors and their ratios.

So by taking the lowest reading of Red and the Highest reading of Red combined between these 2 butterflies gives you your Yellow color range of this particular species.

Which would be:

red 145 - red 174

grn 142 - grn 185

blu 42 - blu 64

You could also calculate your average ratios, but I am still too sick to think that clearly.

Below the picture, I have also found the Color range of the white, which match exactly from one butterfly to the next using a very small area on the body of these creatures. These are the represented by the Purple and Orange dots.

In all of the tests, I have ever tested this is the first time that a creature had perfect shades of grey threw out its entire range.

Example: red 233 green 233 blue 233

Below I have done this same color range technique on the other mentioned species of butterfly.

Finding the color range of the pink and white areas.

 

In the last picture I was ably to find both of these species in the same picture. I looked at many hundreds of pictures and this is the only one I could find.

Fortunately there wings are at the same angles and distance from the camera and the lighting is the same on both creatures.

Again as I did above I have let the smallest and biggest of the Red RGB values dictate which pixels where chosen.

I have found the white and black color ranges for each butterfly.

As you can see that even though these species color ranges was almost identical to each other in the previous pictures, these species colors do not match each other.

 

 

.

I know there is still a lot of comments I need to catch up with, so I will try to continue again tomorrow if I am feeling better.

Thanks

Hello Again

One more thing that I wanted to touch on, that I think there is a lot of confusion about, is this.

If say you did come across a (wild) Zebra in the same field as a (wild) cow and when you took there picture they had exactly the same color ranges, in all aspects to each other. (Which is highly unlikely)

If these are the only 2 creatures in the world that had these exact same color ranges then a picture of a Zebra and a picture of this same species of cow would appear on your screen. If you picked the picture of the Zebra it would tell you this is a Botswana, Zebra from Africa.

If you chose the cow it would tell you this is a wild species of cow from the plains of New mexico called the Speckled MooMoo Cow.

If the cow was a certain breed rather then a natural breed, it may say something like, this is a man made breed from California called the California cow.

If there where other creatures that also fit into this Cow and Zebras color ranges, a picture of them would pop up also.

So now it is up to you to pick from a few possible choices, but you already can see what your looking at so the chose would be clear.

 

If you were confused whether something was either a bird or if it were a fish because they had the same color ranges, you would simply click on the picture that best looked like what you were looking at and it would tell you there species.

Hope all this makes seance.

Edited July 5, 2015 by whatever theory

[Strange]

The most important aspect, I think, besides what the species is, is geographic location. So this is always the first thing that you need to consider.

 

Moving the goalposts again?

 

 

I do not think I can summarize this theory into one or two sentences.

 

If you can't define your hypothesis is, then this is just a complete waste of time. How can you test an idea that is undefined?

 

It seems to be something like: Some species in the same location might have some colours in common, apart from some exceptions which don't, and these might be different from other species in the same region except the ones that aren't - but I can't define "colour" or how closely the colours will match but look at these pretty pictures...

 

 

Strange when you argue: why is this technique important because we already have DNA I will list 4 reasons, I think,why this is better then DNA.

 

You missed out the fact that DNA works and your idea doesn't. (DNA can even identify previously unknown species - your idea couldn't even if it worked.)

 

 

If you were confused whether something was either a bird or if it were a fish because they had the same color ranges, you would simply click on the picture that best looked like what you were looking at and it would tell you there species.

 

And, presumably, the same is true if they were two species of fish or two species of bird? You would consider other things like the shape or size, and then use the colour?

 

So, basically, this comes back to the way colour is currently used to identify species? You look at all characteristics, including colour.

[whatever theory]

Geographic location has been the main reason why colors will be present in species, from day one in this thread and also my pearl thread.

Nothing has changed here.

Ok Strange I will try hard to fit my hypothesis into one or two sentences.

 

 

And, presumably, the same is true if they were two species of fish or two species of bird? You would consider other things like the shape or size, and then use the colour?

Yes you are correct color is not the only deciding factor. Great, now can we move on?

If after all of the research is in and if there are say 2 species with the same color ranges then it will limit your choices to these two.

Now all you have to do is pick which one your species looks like from the 2 choices given. Then once you have chosen it will tell you the information about the chosen species.

 

 

You missed out the fact that DNA works and your idea doesn't.

You are the one that suggested this butterfly test and in this technique I have been removed from the equation, so in this test, I think it "Works".

Edited July 5, 2015 by whatever theory

[tar]

Whatever Theory,

 

You showed with the butterflies, that it MIGHT work. But again, YOU picked the areas. You have not described how the camera/app is going to pick the areas to find a number for, nor how a particular number is to be arrived at, in a mechanical way, nor how an identifiable range of numbers for an area is to be decided upon. Without the details, you can't "test" the idea to see where it works and where it does not work.

 

For instance do you have a number for the white area of black and white butterfly number one, that you will always arrive at when you analyze the white area of any member of the same species, using any smart phone, in any weather conditions, regardless of what colors are behind or around the butterfly? AND is this number different from the number of the white area of species two?

 

You for instance gave us three numbers for each species white area. Do these numbers represent upper red mid red and lower red? Does this mean we will never take a picture of species number one, where any pixel in the white area of its wing will have a red value greater than 242? If there is one pixel found in the white area of species number ones wing that has a red value of 243, does that prove your scheme does not work? If we read a number of 242Red 220Green and 216Blue, in the white area of the wing of a black and white butterfly, does it belong to species one or species two?

 

I am thinking you need to come up with a way to average, or blend the numbers, that will give you the same number, within a certain small band, no matter what large group of pixels in the white area of that subject you randomly chose to analyze. Then, you have a number to work with. Not a large window that an elephant could walk though, but a tiny slit, that only a number arrived at by analyzing a large number of pixels off the white area of THAT species of butterfly, would fit through.

 

Since shadow and glare can vary your numbers greatly, I suspect you might fair better concentrating on your ratios, or on what turns out to be the hue, or tint. This makes sense to me, because the pigments that exist in the white area of the wing of one member of a species is likely to exist in similar quantities and distribution in another member of the same species. Not only would concentrating on the tint, help to eliminate shading issues, and allow IDing in various known lighting conditions, but there is also a good chance that a certain protein ALWAYS absorbs the same wavelengths of light, and the existence of that protein, would ALWAYS leave the same identifiable hole in the "white" light, or whatever known topline collection of wavelengths the subject was illuminated in.

 

Regard, TAR

[Strange]

Yes you are correct color is not the only deciding factor. Great, now can we move on?

 

You seem to have moved away from your "colour is unique and will identify the species" to "after you have narrowed it down to a type of rodent that lives in the desert then colour might help you say which species it is".

 

So how does this differ from any other method of identifying species by observation?

 

Especially when "there's an app for that": http://www.gardenista.com/posts/diy-identify-leaves-and-flowers-theres-an-app-for-that

(That's just the first thing on the search results, there appear to be many others)

 

Or this one, where (you will be pleased to see) the first two questions are about colour:

http://www.insectidentification.org/

Edited July 5, 2015 by Strange

[MonDie]

Color is a more trivial change than morphology, so perhaps it's more likely to become a hybridization barrier, as explained below.

 

Male mandrills and peacocks both use colorful displays.

 

WT, recently diverged species may have overlapping territory, a *hybrid zone*, but it's maladaptive to make hybrid offspring because of *reduced hybrid viability*. Because each species is adapting to its new environment, a hybrid offspring is adapted to neither environment. To top it off, hybrids are often sterile. Consequently, individuals will have better reproductive success if they don't waste time courting or mating the other species, a situation which may lead to hybridization barriers. Color probably often serves as one such barrier. For example, suppose one species is more brown, so its females start picking browner males to avoid hybridization, hence the browner males get more mates, hence that species gradually becomes browner. That would be color serving as a prezygotic barrier to hybridization.

By the way, some birds see ultraviolet light.

I have to say, this is a pretty cool theory. I doubt we're the first, but we should be proud of it. It could explain the origin of colorful sexual displays.

Edited July 5, 2015 by MonDie

[whatever theory]

Hello Strange

 

 

(DNA can even identify previously unknown species - your idea couldn't even if it worked.)

Ok, so that is 1 point for DNA...BUT WAIT!!!...

Below is a picture of fossilized bones, all of which appear to be from the same animal. It is clear to see that the manner in which the bones aged, at its location, have all aged the same way and all of the bones are the same color.

Below is a picture of many of the same species, all from the same location and appeared to die around the same time.

All of these creatures aged the same way and are clearly all the same color.

 

Below is another picture of 2 of the same species that both aged to be the same color, because of there location and the mineral content of the soil, along with what species this has determined what color there fossils will be.

 

Here is 2 leaves, from the same species that have also turned the same color, based on age, species type, location, and mineral content of soil.

 

Here is a picture of many different kinds of fossils all from different species found in different locations.

You can see that they are all different colors.

 

So, if you were studying fossils you would have to take a lot of things into consideration and you would have to do some original test to find out how to identify different soil conditions based on the mineral they contain and how old the different layers of earth would be.

Once you have a good idea of what a species of fossil would turn what color at what depth and at what mineral content of the soil.

You can begin to build a chart of what all of this is.

Then in the future if there was many different creatures all buried together you could take pictures along the way of digging and identify which species each bone belonged to.

You could also learn what color each layer of soil would represent. So the first time you would have to study each layer of soil to determine age and mineral content, but once you have this data, at any time in the future you or someone else can identify this with a picture.

Same with minerals. You have to first find out the difference in the color of minerals and then that would help you determine what effect these minerals have on the color of different fossils found in those minerals.

Once you have gathered all of your soils and minerals data and you have compared all of the different fossils in these different areas, you should be able to paint a pretty picture of what is going on. I heard Mike is an amazing painter, maybe he can help paint the picture. hehehe

Here is a picture of the earths layers. Notice all of the different colored layers.

 

I completely agree with you Tar a way to find the average color is important.

I still have not been able to get around to testing 36 pixels, but I will try to post it soon.

I like what you have done with the snake but I have a few questions.

I understand the first part of your formula:

 

 

I took an average of the 8 colors in the snake post and got red 176 green 110 and blue 55. These numbers had Whatever's "ratios" of -66 and -55. The ratios are just how much more of the one color there is than the next. Transferring the RGB value into CMY we have Cyan 80 Magenta 146 and Yellow 201. Using the under color reduction principle we can take Cyan 80, Magenta 80, and Yellow 80 (a certain shade of black) out of that and get Magenta 66 and Yellow 121.

But I am a little confused on some of the rest of it.

 

This is equal to one part magenta and 1.83 parts yellow, or if you want to bring the tint up in word or paint or whatever, using RGB values, use 121 Red and 55 Green.

Regards, TAR

So the snake's RGB value, based on those 8 samples could be thought of as R/G 2.2

or in CMYK, Y/M 1.83

I don't get what happened to Blue (R/G 2.2)

I do not understand what these letters mean either. (CMYK, Y/M 1.83)

I also do not understand how to convert (1 part of magenta and 1.83 parts yellow) into RGB scale. And again what happened to blue or cyan?

 

 

 

You showed with the butterflies, that it MIGHT work. But again, YOU picked the areas. You have not described how the camera/app is going to pick the areas to find a number for, nor how a particular number is to be arrived at, in a mechanical way, nor how an identifiable range of numbers for an area is to be decided upon.

I used the Dropper tool to find the largest and smallest Red RBG value of a certain color range. These values were my lightest and darkest shades of that species particular color range. So if a computer app was ever developed it would have to look for the same things.

But anyone can start using this to build a species color range at home with their computer now, you do not have to wait for an App or software to come out.

 

 

 

For instance do you have a number for the white area of black and white butterfly number one, that you will always arrive at when you analyze the white area of any member of the same species, using any smart phone, in any weather conditions, regardless of what colors are behind or around the butterfly? AND is this number different from the number of the white area of species two?

 

I think that this technique is already ready to go if you are using very controlled condition like a "Black Box", but comparing photo to each other is far off. This area is where I am having the most amount of difficulties with also, and this is why I am trying to talk to people and try to find out all of the variables that need to be considered.

Like the App that is already being used to identify plants, that Strange just posted a link to: http://www.gardenist...an-app-for-that

It looks like they have not yet also taken all of these factors into consideration also. It sounds like it does not work. Maybe if they were using a black box and sticking more to only identifying color I think they would have a better chance of accomplishing there goal.

 

 

 

If there is one pixel found in the white area of species number ones wing that has a red value of 243, does that prove your scheme does not work? If we read a number of 242Red 220Green and 216Blue, in the white area of the wing of a black and white butterfly, does it belong to species one or species two?

I am not really trying to find any creature whole entire color range. It is too much work, but I am convinced that the technology exists to accomplish this. There may be some small areas from one creature that match another, that is why this computer program would have to analyze all of the colors to find the averages and same colors between species.

I am just trying to demonstrate this color range between species, with my simple testes. If you were really serious you would want to take a lot of time on each species, but you would only have to spend all this time the first time. Then after that it would be easy to compare future species to your already known color code chart.

 

 

 

Since shadow and glare can vary your numbers greatly, I suspect you might fair better concentrating on your ratios, or on what turns out to be the hue, or tint. This makes sense to me, because the pigments that exist in the white area of the wing of one member of a species is likely to exist in similar quantities and distribution in another member of the same species. Not only would concentrating on the tint, help to eliminate shading issues, and allow IDing in various known lighting conditions, but there is also a good chance that a certain protein ALWAYS absorbs the same wavelengths of light, and the existence of that protein, would ALWAYS leave the same identifiable hole in the "white" light, or whatever known topline collection of wavelengths the subject was illuminated in.

Thank you for the suggestion Tar. I will ponder on this issue.

 

Thanks for for posting the video and for supporting my theory.

 

 

I have to say, this is a pretty cool theory. I doubt we're the first, but we should be proud of it. It could explain the origin of colorful sexual displays.

 

This is my main motivation behind all of my work "colorful sexual displays" OOOOoooo Yaaaa!!!...

 

Thanks for posting the links Strange. Nice to know that I am not the only out there thinking about this.

 

Hello Studiot

I have been planning on doing a test were I take pictures with the different cameras in different lighting conditions of the same objects. At different times of day. I am going to use different setting and I will also try your suggestions to try different resolutions also and I will also do as you say and compare all of these images using different software programs and computers. Please let me know if you have anything thoughts on how to better this test. It has been raining for a couple of weeks so I am just waiting for a sunny day to do this test.

[tar]

Whatever Theory,

 

CMYK stands for cyan magenta yellow and black, which are the toners or inks that are used in standard digital color printing. Toner for laser printers and inks for inkjet printers.

 

As that complementary color experiment, where you stare at a spot without moving your eyes for 30 seconds illustrated, Red's complementary color is cyan, green's complementary color is magenta, and Blue's complementary color is yellow. Thus as that upsidedown chart I came up with a while back, showed, having a lot of magenta toner, absorbs a lot of green light.

 

Thus, let's take a made up color R 172, G37, B122. The green number is low, which means the magenta number is high. It is possible to think of red 172 green 37 blue 122 as having a partner, or same tint, constructable out of CMY by taking White255-172red and getting 83 not red(cyan), taking 255white-37green and getting 218 not green (magenta) and taking 255-122 and getting 133 yellow. So if Mike would squeeze 83 mm of Cyan paint out of a tube, and 219mm of Magenta paint and 133mm of yellow paint and mix them up together he should get the same color purple we read as R 172, G37 B122. But, he does not have to use any cyan at all to make the same tint, as he can make black out of 83mm of cyan, 83mm of magenta and 83mm of yellow. So he should be able to make a lighter shade of the same purple, by never putting the black in. So we thought we could make the purple by using 83cyan 218magenta and 133yellow. so we should be able to make a lighter shade of the same hue by using 0mm of cyan(83-83), 135mm of magenta (218-83), and 50mm of yellow (133-83). We could check our logic by seeing if 255not cyan(red), 120 not magenta(green) and 205 of not yellow (blue) gives us a lighter shade of the same purple.

 

So we can make that same hue, of a lighter shade with 0 cyan, 135 magenta and 50 yellow. So that is where the cyan went (to zero), leaving us 135 M and 50 Y. dividing 135 by 50 we get 2.7 time more Magenta than yellow, so I made up a conventional representation of this ratio and called it M/Y 2.7. So, one part yellow and 2.7 parts magenta, should give us that hue, and we have a single ratio, M to Y at 2.7, that gives us the hue (but not the shade) of R 172, G37, B122.

 

But you want to work with the RGB, so lets see if the opposite logic works. Rather than taking black out to get a lighter CMY shade that we can make a two number ratio out of, let's take white out of the RGB number and get a darker shade of the same hue. The smallest number is the green number. But R37 G37 B37 is a dark grey, so we are just taking a little white out, and we should just get a little darker shade of R172, G37, B122 by taking 37 from each value and getting R135, G0, and B85. dividing 135 by 85 you get 1.59, so I made up the convention for the ratio of R/B 1.59. So to check it, to see if it make any sense and gives us a consistent hue, pick a number for Blue....38 and multiply it by 1.59 and you get 60.42 which we will round to 60 for the Red figure, giving us R60, G0, Blue38. which looks again like the same hue, only a real dark shade of it.

 

Regards, TAR

[whatever theory]

Strange

The links you provided are great.

It seems like there are many different names of each App, NatureGate, iPflanzen, Gardenista.

I have been wondering why no body ever thought of this before, now I know they already have.

Even thought there are many Different names for these Apps all of them claim to have a 700 species database, so I am wondering if they are not all the same program, just with different names???

It seems from al of the comments that it does not do a good job of identifying species, from the test that everybody has performed.

It also seems that they are just taking pictures in normal lighting and then comparing these photos to each other. But as we have been discussing we all know that they is many problems with doing this and that may be why there App does not work for anyone.

It looks like they do all provide a certain colored grid paper, on which you need to photograph your subject on. But it looks like some Apps do not even do this.

This is one of the issues we have been talking about here is that there needs to be some kind of control that is the same in all picture to compare to.

Also by using Shape, Size, hairyness, thornyness I think it may be too many factors to consider.

If you did not take your picture at the exact same distance from the subject every time or if your zoom was different in different pictures, then your sizes would be all over the place. So this may not be a good deciding factor.

If a species lost all of its hair or a bug ate the hair off or if it was missing petals, etc., etc. the shapes and number of petals etc. would be different.

Also the flower will change shape as it blooms.

The only things that I can see that would stay consistent would be color.

So by taking all of these other big problems out of the equation and just focusing on color, geographic location and time of year, may improve what they are doing also.

I see that there is also a bird and fish identifying App, but I can not find links to them. Anyone have links?

I hope that the developers of these Apps can join us here and maybe together we can help each other answer some questions.

Also, an App or computer software is only the ideal end goal and if this goal is never reached, each field of science can use these techniques and start building there own Color Code charts.

For instance the flower and plant App that already exist and it sounds like it does not really work, so for you this is not practical. Say the 700 species it contains are not even the same ones that you are working on.

You can photograph all of your flowers and plants in your own black box. You can now share these with others. If you were the first one to figure out these species color ranges you could open your own website. Anyone that wanted to compare there photos against your in the future would just have to take their pictures in the same condition as you did.

Hello Again

I wanted to comment again on this statement:

 

 

(DNA can even identify previously unknown species - your idea couldn't even if it worked.)

Another way my technique may identify previously unknown species is the example I have shown of the Zonkey.

While it would take a lot of money and time DNA these species you can do the test I performed very fast and free.

Also if my theory is correct it can also identify other things that DNA can not identify like minerals, gemstones, rocks and fossils.

'Whatever Theory" Score: 10 ---- "DNA" Score: 0

 

Thanks Tar, for explaining everything about your new formula for finding an average color. It took me a while, but I finally get it now.