Recording only the red or green or blue channel for black and white video

Started by eYeLicker, October 05, 2014, 10:41:39 PM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

eYeLicker

In film photography it was common when working with black and white to put a red or blue filter on the lens to achieve a certain look. A similar effect can be duplicated in Photoshop. Open a RGB color image in Photoshop and in channels turn off all but one channel of red, green, or blue. One of the single channel images is usually more interesting than the composite RGB converted to grayscale. It would be interesting to have the option in ML. But even more interesting, if only one color channel is being recorded couldn't the image be recorded at three times the resolution?

Walter Schulz

Canon's cameras are using sensors with bayer matrix. The colour filters cannot be altered by software manipulation and therefore resolution cannot not be tripled.


eYeLicker

Interesting - So it is not possible to capture a high resolution image; one that would require an impossible bandwidth but transfer the image data for only one of the color channels. It seems to me that if the final format is known it could be filtered on save. If byte 1 is red and bytes 2, 3 are blue, green then skipping bytes 2 and 3 and replacing them with other red byte data should allow three times the image size to be saved with the same amount of bandwidth.  That's not a filter manipulation, the filter works as normal but the unwanted data is skipped or truncated on save. The bottleneck is the bandwidth limit as i understand it. Is the image file saving method a black box? The data being saved can't be processed in any way? The saved file may require post processing but so does RAW.

Walter Schulz

Quote from: eYeLicker on October 05, 2014, 11:50:49 PM
Interesting - So it is not possible to capture a high resolution image; one that would require an impossible bandwidth but transfer the image data for only one of the color channels.

Sorry? What does bandwidth has to do with it?
Please read the documents about CFA and you will see you have a basic misinterpretation about what is going on inside a sensor with bayer matrix.

eYeLicker

I will look at the CFA docs. I was mainly interested in the video recording of a single color channel. The resolution was a side-effect. If the data being written to the CF card cannot be processed then recording a single color channel is impossible anyway.

dmilligan

Quote from: eYeLicker on October 05, 2014, 11:50:49 PM
If byte 1 is red and bytes 2, 3 are blue, green then skipping bytes 2 and 3 and replacing them with other red byte data should allow three times the image size to be saved with the same amount of bandwidth.
No. If you skipped for example the red and blue pixels and only recorded the green pixels, you would record 1/2 as much data (what you refer to as "bandwidth"), and the resolution would also be reduced by 1/2.

Red, green, and blue pixels are not co-located in raw data, they are not separate channels as you are probably thinking of it, they are in a bayer matrix. When you record 1920x1080, you are only recording 960x540 green pixels, 480x270 red pixels, and 480x270 blue pixels.

You're probably thinking of an image being like this:
{RGB} {RGB} {RGB} {RGB}
{RGB} {RGB} {RGB} {RGB}
{RGB} {RGB} {RGB} {RGB}
{RGB} {RGB} {RGB} {RGB}

Where each pixel is enclosed in {} and made up of three colors.
This is not at all how raw data is, there is only one color for each pixel, it looks like this:
{R} {G} {R} {G}
{G} {B} {G} {B}
{R} {G} {R} {G}
{G} {B} {G} {B}

PaulB

You can still use colour filters when recording. When you convert to B&W it should give roughly the same effect as with B&W film.

Levas

Did you know that Canon has a "monochrome" picture style for this.
It even let's you choose between none filter or filtering for Yellow, orange, red or green