This content has been marked as final. Show 12 replies
Well, the compressed lossless I believe only deals with the embedded preview
image (what programs that can't processor raw files fully shows.)
Convert to Linear is in my book a bad thing. It converts the RAW data to
bitmap data. That is lossless and is basically like coverting your raw image
to JPEG or TIF. I see no advantage to doing this for any reason.
Embed the original raw file converts your image to DNG (still raw data) and
then embeds the original file in as well. For example if you convert a CR2
raw file to DNG and choose embed original raw file you will have a DNG file
that also contains the original CR2 file. This will cause the size of the
DNG file to be very large as it is basically two raw files in one. I also
don't see a lot of use or point to this.
Thank you for your explanation Robert. So, which one is to choose? Compressed?
You are referring to saving a DNG. So effectively if you are starting from a non-DNG raw file (e.g., a Canon CR2 or Nikon NEF) then you are effectively converting the raw file from one format to another.
In general to preserve the original data you want to choose Compressed lossless. The size of the preview determines how big of a JPEG to include inside the DNG for preview purposes. Bigger preview = higher quality but also more storage required. However at this moment you can't take the DNG file and turn it back into your original non-DNG raw file (e.g., the CR2 or NEF), unless you also check the "embed original raw file". By checking that option you can extract the original non-DNG raw file, but the downside is that your created DNG file is quite a bit bigger than if you hadn't checked that box.
ok, now it's clear, thanks Eric!
Interesting choice of words - However at this moment you can't take the DNG file and turn it back into your original non-DNG raw file... So perhaps we will someday see an option to de-convert? That would be extremely useful for those of us who would rather not store/retain the originals or take the size hit for embedding.
Yes, I would use compression. It is lossless and does save hard drive space.
I wouldn't bother embedding the original and I would never convert to
I agree with you Greg that the choice of converting back to original and a XMP sidecar for the metadata would be a fantastic option and would make DNG a safer option for many users.
Hope it's possible...
Interesting sentence, indeed, but let's not read too much between the lines, and refrain to ask more info, as it is already kind of Eric et al to share so much in here!
It was just an observation. Look at it more as support for a feature request. I'm aware that Eric may not be at liberty to answer questions about future technology developments in a public forum.
I should be possible though? Saving the RAW data twice seems not neccessary in the implementation to day. If you could store the metadata of the properitary file in a container and reconstruct the file from that container and the RAW data in the DNG...
Unfortunately it is quite a bit more complicated than that.
I'm oblivious to the difficulties of doing this back-and-forth conversion, but it will greatly ease my reservations about using DNG as my one and only raw image format.
In fact, if I were sure I could go back to my NEFs if needed someday, I'd convert everything to DNG, now, and I'd be happier than before :)