A few weeks ago I asked a question about creating multispectral images using ImageJ and the micatoolbox. I was having issues opening my .RAW images, getting these three errors:
- “Timeout- DCRAW doesn’t seem to have processed this file”
- “There are no images open in line 63 run (“32-bit<)>;
- “There are no images open in line 363 photoID = getImageID(<)>;
Since then I have uninstalled all the software and did a clean downloaded the following: ImageJ for Java 8, MicaToolbox V.2.2.1 and LibRaw V.0.20.2 on my Windows. I downloaded ImageJ to my desk top and extracted the files for both micatoolbox and LibRaw into the plugins folder of ImageJ. I have the four folders required for micatoolbox as well as the folder \”LibRaw-0.20.2\” in the plugins. I tried opening a .RAW file again, but I still get the three above errors. I\’m not sure what\’s going wrong.
Thanks for sending this through, I can’t open the file either. Ok, this is an important piece of information: DCRAW is a tool that uses the metadata in your image file to apply the camera model that took the image (rather, the sensor design/mosaic) to linearise the pixel response. Given that you have artificially created a .raw file (btw. I am not aware of any camera that produces .raw files, they all have they company-specific RAW image formats such as .orf etc.), I am not surprised that DCRAW can’t deal with it (i.e. your file is called .raw but it isn’t actually a RAW file?). The idea of RAW is that it is the unmanipulated sensor data. If you say you’ve “created the file in photoshop”, then that doesn’t seem to be the case and as such correct image linearisation doesn’t seem possible. Briefly, by virtue of the very concept that makes calibrated photography possible, DCRAW (and therefore the toolbox) can’t deal with RAW images which aren’t out of a camera, i.e. where there is no link between the raw image information and the architecture of the sensor that took the image. I don’t know enough about this in detail, but that’s my guess?
I would assume that you have created your file from a .jpg or another compressed non-raw file? If so, the best chance at achieving successful cone mapping would be to use the jpg linearisation tools in the toolbox as explained on the website:
Else, please use the RAW files straight out of your camera. I hope this helps!
Yes, what Cedric said^
I would add that you can actually use images from other software, but you need to perform the relevant linearisation and cone-mapping modelling to get away from whatever non-linear colour space the image was in (e.g. some people have used 32-bit linear images from MATLAB). Using non-linear images requires e a lot of modelling and checking. That’s why the RAW formats are so “safe” to use for image analysis, because they aren’t messed with by unknown software and processes.