Wednesday, February 19, 2014

Histograms

Based on the discussions in the LightZone forum and my own experiments I found that histogram ≠ histogram.

I created in DPP a slightly processed JPG of the same image I used in my previous post. I opened then the JPG image both in DPP and LZ and the RGB histograms were different even the images looked the same. I had earlier assumed that the differences in the histograms were caused by different processing of the raw image, but that is not true. The histograms don't visualize the same thing!

Canon Digital Photo Professional LightZone
I found a piece of software ImageJ that has an option to copy-paste the channel histogram data into Excel. I then generated a simple channel intensity vs. pixel count chart and that corresponds quite well with the DPP histogram. It looks like the scaling of the y-axis in DPP is not linear, e.g. the first peak in blues is about a third of the highest peak in DPP but in Excel it's about two thirds (200k vs 320k). I tried to change the x-axis to logarithmic so that I could have produced the histogram in LZ, but the result wasn't the same. I guess LZ shows some kind of luminosity vs. channel intensity histogram as the x-axis would indicate.

Linear and logarithmic histograms generated in Excel

No comments: