Hi micaToolbox/QCPA folks,
First, thank you for creating such a comprehensive set of tools and guides for this kind of work! It’s amazing to be able to go through all these steps in a few minutes with accessible equipment. Much better than cobbling together inherited Matlab scripts…
I have a question about creating a 300-700nm range cone catch model for our camera (Nikon D7000/Coastal Opt 60 mm 1:4 UV-VIS-IR Apo Macro). We have access to a spectrometer & photos of an X-rite colorchecker, so it seems like we just need to buy a set of pastels. Based on the documentation, it sounds like if we’re using pastels & measuring their reflectance curves, we don’t need the colorchecker. So a couple of questions before we navigate COVID restrictions to try and get all these components together:
- Is it correct that if we’re using pastels, they essentially replace the colorchecker for creating a 300-700nm cone catch model for a camera?
- Is there any advantage to using both the pastels & the colorchecker (i.e. as extra points in the 400-700nm range), and if so, is that feasible or is it more trouble than it’s worth?
- Any particular pastel sets that you recommend? We’re assuming that flat-sided ones with a relatively square aspect ratio are best, but other than that we’ll probably just go with the cheapest one we can find.
Any advice you have is appreciated! Many of these photos were collected over several field seasons (including before micaToolbox was released), so we are trying to retrofit our calibration steps for the QCPA pipeline as well as we feasibly can.
Thanks! Hopefully you’ll find the toolbox much easier than old Matlab methods 🙂
1 – Yes, the pastels can completely replace the colour-checker chart, and allow you to create cone-catch models for a given set of photoreceptors, 300-700 or 400-700. The pastel set won’t be much good for linearisation, but hopefully you saved images as RAW anyway.
2 – The colours in the pastel set are probably sufficient to replace all the colour-checker chart values. One advantage of using only one set is that they’re arranged in a grid (so easier to measure using the chart measurement tool).
3 – All the pastel sets I’ve measured are pretty good. Flat sided (square profile) is good. You can easily clean the pastels by rubbing them on a clean piece of paper. Avoid any sets with too many obviously fluorescent colours. There are nice 64-colour sets out there, which is a great number for the models to work with, much better than the number in a standard checker chart. One thing I’d recommend if possible is using natural light (or whatever light source you used for photos) for measuring the chart values with your spectrometer. This is because all colours (even x-rite colour chart values) fluoresce to some degree, and by using the same illuminant you took the photos under you control for this difference.
You could also “mix & match” a couple of sets. Not all colours have interesting UV peaks/reflectance dropoffs, which the model needs to work well. These tend to be the yellow, orange, blue and purple colours. So you could take some of these “UV-interesting” colours from a second set to replacing any fluorescent colours or redundant grey pastels which don’t add much to the modelling.