General information about monochrome conversions


  There are three things to consider when modifying a digital camera to monochrome. Color, light spectrum sensitivity and the installed AA (anti-aliasing) filter. Starting from the sensor level and working my way out, I will explain how modifying each of these affects the image captured on the Sony mirror-less cameras. 


BAYER pattern CFA

  First there is the color aspect. The sensor can only capture the level of light (photon quantity) that falls upon it, not the actual color. To capture the color of the light falling upon the sensor at any given spot, the sensor is covered with filters of the three primary colors, Red, Green and Blue. Each color filter covers one pixel site, or photo sensitive diode. There are 12 million of these sites in a 12 MP camera. The layout of the filters is done in one of several ways, but for this Sony camera it is the BAYER scheme that is used; 1 R, 1 B, and 2 G pixels. Using a mathematical algorithm, the processor converts the light levels of surrounding pixels into a color that best corresponds to the actual color captured at that pixel. This is done for the entire 12 million pixels. These RGB filters block a percentage of light from reaching the sensor. It's a trade off, one that most are willing to have to get color images from their camera. FIgure 1 shows how a sensor "sees" light across the spectrum. The portion of the spectrum that the human eye is sensitive to is the rainbow bar at the bottom of the graph. The area under the R G and B curves, and contained in the pink curve, is the light that reaches the sensor in an unmodified camera. By removing CFA layer from the sensor, all of the light's color and intensity hit each and every site on the sensor. The total area under the pink curve (hot mirror, IR cut) is the amount of light reaching the sensor now that the CFA has been removed. The overall sensitivity of the camera is increased, with the trade off of only being able to capture the luminosity of the scene in monochrome. This assumes the IR Cut filter and AA filter are still in place. 

color response of camera.jpg

Fig. 1   CCD sensor shown, but CMOS is nearly identical

  Second, there is the consideration of the sensor's sensitivity to each wavelength of light falling upon it. Sensors have a sensitivity range of around 300nm to 1000nm. The length of the light wave is in nanometers, that's one-millionth of a meter. The human eye is sensitive to the range between 400 to 700nm. The IR cut filter, or Color Correcting filter (CC), is used to limit the sensitivity of the sensor to that of the human eye. This is shown in the area under the pink part of the graph in Fig. 1. With the CFA in place, the sensor sees the area under the three curves colored red, blue and green, but contained in the pink region. By removing the CFA and leaving the CC filter in place, we now gather everything under the pink curve. Then by removing the CC filter, we gain everything under the black curve shown in Fig. 1. This is a big increase in sensitivity, not just from the standpoint of the amount of light, but also of the color of that light along the spectrum. Much more IR (infrared) light is now seen by the sensor. Most conversions simply remove this CC filter and leave the CFA in place. IR filters are used to block regions of the light, typically on the lower end (300 - ~650 nm), to let just the higher end wavelengths through to the sensor. This can result in a very eerie effect of things in the scene that reflect those wavelengths above what the human eye can see. Not all surfaces do reflect all wavelengths of light--a red subject only reflects red light--, nor do all sources of light have the full spectrum shown in the figure--think of a red LED emitting only red light. If the wavelength is there, and the camera is modified fully--no filters at all--, then it will be recorded. This is true of a non converted camera also, but with more restrictions. In an unmodified camera the amount and wavelength of the IR light reaching the sensor is limited. This is why it is necessary to have long exposures when shooting IR images with a unmodified camera. Without the CC filter the exposures are much shorter. I've hand-held a shot once with a 850nm filter over the lens. This filter is so dark, you can't see through it with your eye. But since the camera is sensitive to that wavelength of light and above, it captures an image of everything in the frame reflecting IR light from 850nm to 1000nm, but not below. The image was shot in the daytime, so there was plenty of IR light coming from the Sun. By placing any other IR filter over the lens, I can make the camera sensitive to any range that the sensor can "see". 

  The third thing to consider is the Anti Aliasing (AA) filter--technically a low pass filter--used in most modern digital cameras. The filter is typically optical in design, or it can be some sort of in-camera device that shakes the sensor very slightly *. Either way the effect is the same. The goal is to spread out the image across neighboring pixels to lessen artifacts that can cause undesirable effects in the image. Moire is the most common and typically the most visible. This can be seen when an image was taken of something that has a repeating pattern, such as a fine fabric weave. Most situations don't require this filter, but some do. The moire effect can be removed in post processing, so not having the filter is not a real burden. And some increase in resolution is gained. In the Sony cameras, however, the AA filter is part of the CC filter, so to remove the AA filter it is also necessary to remove the CC filter. The gain is large from removing the CC filter, but not so much from removing the AA filter. It is nearly impossible to reinstall just an AA filter, but using only a CC filter is relatively easy. SImply put a screw on CC filter over the lens when limiting the range to the visible spectrum is desired. These can be had quite easily from the web along with the IR filters. 

  The stock CC & AA filter stack is independent of the protective clear glass mounted over the sensor. The protective glass is returned to it original position over the sensor once the conversion is complete, but the AA & CC filter is left out. In this particular camera, it is easily put back into place later if desired.  

  I've converted a full frame sensor just recently, go HERE to see more about the results. 

  Go to my contact page with any questions, or if you care to comment. 

* I would love to see this implemented in the Sony line of mirror-less cameras. This would allow one to turn off the AA effect when desired. 

© Daniel Morrison 2014