LRGB imaging, how does it actually work ?

[Astroboffin]

Hi all,

This is going to sound like a really stupid question, but I need to ask it, sorry in advance.

Can someone explain in as simple terms as possible how you get a colour image from a mono camera.

By that I mean, I am struggling to understand the science behind it, if you take an image with a mono camera through a coloured filter, the image is still mono, no matter what colour the filter is, even a green filter with pink spots, the final image is still mono, blacks whites and greys.

So if you take three different images through three coloured filtered RGB all, three images are still mono, put them together and all you have is a more detailed mono image, so where on earth does the colour come from....?? And how is it allocated to each image, as the camera only sees blacks, whites and greys....!!

Why can't we just take say 30 "L" images, and then take 10 and colourise them red, take another 10 and colourise green, and same for blue, then combine, surely this will mean we don't need RGB filters

I can't see how this all works, I am genuinely trying to get me head around all this LRGB stuff, I know it all works, just need to explain to me how.

Told you it was a stupid question.

Regards

AB

[beamer3.6m]

I dont know the answer but I like the question. Looking forward to the answer.

[Robp]

The RGB filters only allow through the wavelengths that are specific to that colour of light.

So for a red filter, any light received will be red only. even though the image from the camera will be mono in colour, that is only light which is red.

In imaging software such as Photoshop, you assign each mono image to there specific colour in the channels tab. Photoshop will add the colour to each layer and then where the colours intersect, the colour will change to represent the intersection. so if you have a part of the image where both blue and red are present, it will show purple. the shades etc are based on the brightness on the mono images (I think)

An L filter allows light through on all of the wavelengths. If you were to assign L images to R,G and B then the image will be a single colour and there is no variation in light

Thats my understanding of it, sure someone else will be able to explain it better

[ncjunk]

The thing you are missing is that each mono image only contains information from a certain range if wavelengths specific to each filter.

Each pixel in the picture shows the intensity of the signal so put a red filter in front and it shows the value only for red light getting through and hitting it. Remove the filter and it will show a value of all light hitting it and will have a higher value as it records green and blue wavelength photons as well.

So a red image contains info on only the amount of red light that hits each pixel. The same with green and blue.

You then, for each pixel, take its red image value, then its green and blue image values and combine them back into a colour image.

I don't think i'm helping

[Big Jim Slade]

You may want to start with this: http://en.wikipedia.org/wiki/Light

[glowingturnip]

this is a good example:

http://en.wikipedia.org/wiki/Color_photography#/media/File:Rgb-compose-Alim_Khan.jpg

The Emir of Bukhara in a 1911 color photograph by Sergei Mikhailovich Prokudin-Gorskii. At right is the triple color-filtered black-and-white glass plate negative, shown here as a positive.

Sergey Prokudin-Gorsky - Taken from the Library of Congress' website and converted from TIFF to PNG. TIFF file from LOC

A picture of Alim Khan (1880-1944), Emir of Bukhara, taken in 1911. This is an early color photograph taken by Sergei Mikhailovich Prokudin-Gorskii as part of his work to document the Russian Empire. Three black-and-white photographs were taken through red, green and blue filters. The three resulting images were projected through similar filters. Combined on the projection screen, they created a full-color image. Prokudin-Gorskii also made color prints from some of his images and published a number of them as inserts in Fotograf-Liubitel, a photographic magazine he edited from 1906 to 1909. His 1908 color portrait of Leo Tolstoy was also published as a postcard. In recent years, the Library of Congress has made high-resolution scans of their collection of Prokudin-Gorskii's original glass plate negatives and contracted with outside agencies to produce high-quality color-corrected images from the black-and-white scans. This example is a simple color composite of the three original images shown at right and has not been color-corrected, retouched, or artificially enhanced in any way.

 

 

[Astroboffin]

The RGB filters only allow through the wavelengths that are specific to that colour of light.

So for a red filter, any light received will be red only. even though the image from the camera will be mono in colour, that is only light which is red.

In imaging software such as Photoshop, you assign each mono image to there specific colour in the channels tab. Photoshop will add the colour to each layer and then where the colours intersect, the colour will change to represent the intersection. so if you have a part of the image where both blue and red are present, it will show purple. the shades etc are based on the brightness on the mono images (I think)

An L filter allows light through on all of the wavelengths. If you were to assign L images to R,G and B then the image will be a single colour and there is no variation in light

Thats my understanding of it, sure someone else will be able to explain it better

Thanks very much for that, simples when explained so very well. :)

One thing I still don't quite get, and it is probably my ignorance or stupidity, I still don't really understand why we can't just use all L images and assign RGB to a batch of each and layer together, would that nor give a true colour image, as all the data is in all the images. Or is it that they would just cancel each other out, as it were....?

I know I am missing something

AB

[StuartJPP]

Or is it that they would just cancel each other out, as it were....?

Basically...If you just captured L then you will have a mono image with all the colour information mixed into the same pot. This is handy for the purpose of Luminance which is the "brightness" part of the image but not for discerning colour. If you don't split out the individual colours then the R, G and B will contain the same values, i.e. it will be mono again...

Look at Additive Colour (http://en.wikipedia.org/wiki/Additive_color). It is the combination of the individual monochromatic Red, Green and Blue to produce many millions of different colours. (Which was easily visible on an old CRT TV.)

[Robp]

As the L filter passes all three wavelengths at once there is no way to tell if the signal of a pixel is red, green or blue.

When you assign the individual colours, the pixel will have the same value of colour for all three of the RGB channels so it will essentially be shown as grey.

I think you can check this by using a colour picker tool in some imaging software, I have taken a screenshot to show you.

When you put R,G and B as 100 the colour is grey (left of image). If you assign a higher value to the R channel it starts to turn red (right).

So when you image in RGB you get three values for the same pixel. Red might be 160, Green might be 10 and Blue might be 20. when this is combined you get a really strong red colour

If you were to do the same with a L file, a pixel would be say 160 for all three channels, this would show as a grey.

[Robp]

Another way of explaining it would be to create a block of three different grey shades and assign them to RGB channels

As you can see you need the variance in levels of grey so that the software can determine the colour (in this case red).

If you were to pretend this was a pixel in an image, through the RGB filters the pixel would be brightest in the Red filter image and darkest in the green filter image.

If you were to attempt the same process with an L image, the pixel would always have the same value of grey, so will always equate to a grey (no color is strongest).

So what we do in LRGB images is use the RGB to gather the colour of the image.

We use the L to capture the details quicker as it captures all of the colour wavelengths at the same time (10 minutes of L equals 10 minutes of R+G+B = 30 minutes to capture all three)

As you cant tell the colour of an image from just an L frame we use the RGB data to provide the colour, and the L to provide the detail and brightness.

[Astroboffin]

Thanks for all the input, all straight in my head now, where would we be without this superb forum, for people like me to ask questions.

And big thanks to Robp for the superb explanations.

Regards

AB

[Gutross]

this is a good example:

http://en.wikipedia.org/wiki/Color_photography#/media/File:Rgb-compose-Alim_Khan.jpg

The Emir of Bukhara in a 1911 color photograph by Sergei Mikhailovich Prokudin-Gorskii. At right is the triple color-filtered black-and-white glass plate negative, shown here as a positive.

Sergey Prokudin-Gorsky - Taken from the Library of Congress' website and converted from TIFF to PNG. TIFF file from LOC

A picture of Alim Khan (1880-1944), Emir of Bukhara, taken in 1911. This is an early color photograph taken by Sergei Mikhailovich Prokudin-Gorskii as part of his work to document the Russian Empire. Three black-and-white photographs were taken through red, green and blue filters. The three resulting images were projected through similar filters. Combined on the projection screen, they created a full-color image. Prokudin-Gorskii also made color prints from some of his images and published a number of them as inserts in Fotograf-Liubitel, a photographic magazine he edited from 1906 to 1909. His 1908 color portrait of Leo Tolstoy was also published as a postcard. In recent years, the Library of Congress has made high-resolution scans of their collection of Prokudin-Gorskii's original glass plate negatives and contracted with outside agencies to produce high-quality color-corrected images from the black-and-white scans. This example is a simple color composite of the three original images shown at right and has not been color-corrected, retouched, or artificially enhanced in any way.

 

 

that is the best example of LRGB i have seen yet, thank you

[Gutross]

Another way of explaining it would be to create a block of three different grey shades and assign them to RGB channels

02.jpg

As you can see you need the variance in levels of grey so that the software can determine the colour (in this case red).

If you were to pretend this was a pixel in an image, through the RGB filters the pixel would be brightest in the Red filter image and darkest in the green filter image.

If you were to attempt the same process with an L image, the pixel would always have the same value of grey, so will always equate to a grey (no color is strongest).

So what we do in LRGB images is use the RGB to gather the colour of the image.

We use the L to capture the details quicker as it captures all of the colour wavelengths at the same time (10 minutes of L equals 10 minutes of R+G+B = 30 minutes to capture all three)

As you cant tell the colour of an image from just an L frame we use the RGB data to provide the colour, and the L to provide the detail and brightness.

awesome thank you!  and to everyone else in the thread, thank you as well. I guess a mono camera and a colour wheel are in the forcast for tax time!