Importing DNG File into Fiji

Hello,

I am trying to import a DNG file format into Fiji. I tried using Import>Raw and also Bio-Formats, but have had issues with both.

For Import>Raw I know that my image is 24-bit, width = 768, and height = 1024, but I wasn’t sure how to determine the offset? The result is a very grainy image that looks nothing like the original.

For the Bio-Formats, I just kept the default settings and am able to open the image. However, the image now shows that it is 8-bit. I was hoping to do my processing in 16-bit so that I am not losing any pixel data.

I am very new to ImageJ/Fiji, and am unsure of where I am going wrong. Any guidance would be greatly appreciated!

Thanks,
Mary Beth

It could be that Bio-Formats is not behaving ideally, we’d love to be able to check. Could you upload a small public 24-bit sample DNG image to http://qa.openmicroscopy.org.uk/qa/upload/ or otherwise make it available? Or if you have only private data, let us know and we’ll keep the file under wraps.

Hello,

Thank you for the quick response! I’ll load a test image in DNG taken and uploaded the same way.

Thanks for your help!

Hi Mary Beth,

I don’t know how Fiji’s raw import works, but I had a look at your image imported with Bioformats. And it looks fine. It is an 8-bit image. Maybe you get a bit mixed up with the terminology. As it is a three channel RGB image there are 24 bits in total per pixel. But in general - and that’s what Bioformats does - one refers to the bits per channel, which is 8 in that case.

Kind Regards,
Dominik

Hi Dominik,

Ah, ok. I think I see what you’re saying. I’m trying to make sure that I preserve as much pixel data as possible when transferring an image from a smartphone to my computer and then opening in Fiji. So you are saying that the RBG image (when imported using Bioformats) is 24-bit, but the 1/3, 8-bit info in the header is noting that each color channel is 8-bit. Correct?

So then if I wanted to process the image as a 16-bit image, I can select Image>Type>16-Bit and still have as much pixel information as possible from the original image?

Thank you for your help!

Best,
Mary Beth

Yes exactly, so both is correct, your image is 24bit / pixel, but 8 bit / channel.
I’m not sure if a conversion from 8 to 16 bit would achieve anything. It certainly won’t give you more information. Depending on the operations you’re doing afterwards it might be an advantage to have 16 instead of 8 bit per channel, but to be honest I don’t know. You could try your image processing workflow with 8 bit and 16 bit and then check the histograms if it made a difference.
Kind Regards,
Dominik

Great! Thanks Dominik, that was very helpful!

Best,
Mary Beth