bayer | RobsBlogs

Updates to Lightroom come along relatively regularly and they tend to include new features along with fixes and performance tweaks. The latest update, Lightroom 8.2, includes a new addition called Detail Enhancer. This is a feature that is designed to provide some better small-scale detail as part of the raw conversion process. It creates a new DNG file based on a more complex calculation of the demosaicing of the sensor data.

I saw some videos about it and figured it wasn’t going to be of much use for the type of thing I am working on. However, it did trigger one possible area of interest. The algorithms are supposed to be designed to make better calculations around the different color pixels that sensors have. Sensors are set up in a Bayer Pattern where different color sensitive sensors occupy different pixel spaces. They each record in one color and then software interpolates between them to create colors for each pixel irrespective of which color was originally recorded at that location.

In a post from a while back, I mused on the way in which the colors of the Southwest Livery and the registration clashed and seemed to provide a distorted image even when everything around them was sharp. I was pondering whether this was artifacting caused by the different colors and the way the sensor was recording the data. If this was the case, maybe this new functionality would change the way things were rendered. I dug out a few of the shots that had previously demonstrated this effect and ran the process on them. These shots show the wide shot, the original rendering of the close up and the revised rendering using Detail Enhancer.

As you can see from the comparisons, Detail Enhancer does not suddenly render a perfect registration for the aircraft. However, to my eye at least, it does appear as if the results are noticeably better then they were with the original rendering. For completeness, the original rendering is done with the latest process version of Adobe’s raw converter to make things as fair as possible. It does appear to make a difference. This makes me think my theory about whiny things looked wrong might have some merit, even if this update has not fully resolved things.

I have been pondering the way in which the method by which digital images are captured is affected by what is being photographed. As part of my workflow, I render 1:1 versions of the images and then quickly weed out the ones that are not sharp. This needs you to be able to see some detail in the shot that shows whether the sharpness is there. I have found that, if a Southwest Airlines 737 is in the new color scheme, something odd happens.

Digital image sensors actually capture one of three colors. Each pixel is sensitive to a certain color – either red, green or blue – courtesy of a filter. They colors are arranged on the sensor in a pattern called a Bayer pattern. The camera then carries out calculations based on what the pixels around each location see to calculate what the actual color should be for each location. This process is known as de-mosaicing. It can be a simple averaging but more complex calculations have been developed to avoid strange artifacts.

When I photograph the new Southwest scheme, something strange occurs around the N number on the rear fuselage. It looks very blotchy, even when every other part of the airframe looks sharp and clear. I am wondering whether the color of the airframe and the color of the registration digits are in some way confusing the de-mosaicing algorithm and resulting in some odd elements to the processed image that weren’t there in real life. If any of you have photographed this color scheme, can you see whether you had something similar and, if you did or didn’t, let me know what camera you were shooting with so we can see if it is manufacturer specific or not.

RobsBlogs

Tag Archives: bayer

Lightroom 8.2 Detail Enhancer

Sensor De-mosaicing and Southwest Colors