Cut resolution of .DNG file and generate a new .DNG file with the flexibility of a RAW image

[billa91604872]

My new 60-mp camera generates 80 megabyte DNG files. I like the resolution and ability to crop severely, but not all the files are "keepers." I would like to keep a smaller but RAW version of those images. Is there software or a Plug-in simply to cut the resolution in half, say, and generate a new .DNG file 

I was surprised that no one at Tech Support could help. In principle, this should be very, very easy:  just eliminate every other pixel, then include whatever "envelope" info is in the original capture, and put out a brand-new DNG file that can be handled in Camera Raw. I cannot see any obstacle. My camera, pre-capture, can do this with its circuitry, so I find it hard to believe that software does not exist to do the same thing post-capture.  

Any advice. I do NOT want to change the file to .TIFF, .PSD, .JPG or anything else. I want to begin and end with the flexibility of a .DNG file.

[davescm]

The whole principle behind raw is that the raw data from the camera is unchanged, so your method would be flawed. It would also render the raw useless as each pixel in a raw file contains one grey value. That either represents Red or Blue or Green (there are actually twice as many greens as the other colours) depending on the location in the pixel grid. The other colours are interpolated on demosaicing. Deleting raw pixels would ruin such demosaicing.

Dave

[D Fosse]

A raw file is never modified in ACR/Lightroom. It's strictly read only. This is an unbreakable rule.

You can do it in camera, but only prior to pressing the shutter. Be warned though, those files will not be readable by the new "Enhance"-functions like Denoise or Super Resolution.

When you buy these cameras you just have to factor in that you will need a lot of storage space with room to grow. It's just a factor in the equation.

Edit cross-post

[D Fosse]

Interesting point, Dave. The a7r v can do reduced size compressed M at 26 MP, and compressed S at 15 MP. Diving into the specs, it's clear that this is not a sensor crop, but actual downsampling. So how do they maintain the Bayer mosaic pattern? It probably works only at certain ratios.

Another possibility is of course that it outputs demosaiced data, and this may be why it doesn't work with Denoise. Only the full 60 MP is accepted there.

[billa91604872]

This appears to be the answer, but I have not checked. I found this on YouTube.

You bring the original, enormous .DNG file into Lightroom, not Photoshop.

Then choose Export, specifying a .DNG output with LOSSY compression and also enter a value for the new file size you want. The new image can then be worked on in Camera Raw, in Photoshop or Lightroom, with all the sliders you can use to edit a “virgin” DNG file. Presto — keep the benefits of .DNG but cut the resolution so storage is manageable.

This technique seems to work well in the video. And I am surprised that no one at Adobe would suggest this. You can even copy the Develop settings to duplicate the operations performed and paste them into another, bigger DNG file from the same shoot.

Here you go.

Tell me if this is workable, or whether Lightroom no longer supports Export to a DNG. I use Lightroom for Library and database functions, but use Photoshop for the heavy-lifting of Post-.

Thanks.

-BA


https://youtu.be/SGFZi63tph4
How to convert Raw files to smaller file-sized DNG Files
youtu.be

[D Fosse]

It appears to work, but by necessity this is a demosaiced dng, and with lossy compression at that.

However, it does seem to retain at least most of the dynamic range of the original, highlight recovery seems more or less intact, so apparently it is still linear data. So still some advantage over a rendered RGB file.

[D Fosse]

I remain a bit skeptical. Demosaiced means three channels instead of one, so that alone should triple file size. And yet it's about half (not resized). So the compression has to be quite aggressive.

[davescm]

The big problem with lossy compression is just that - data is thrown away. So if you decide in the future that part of one of those images is now useable, e.g. in a composite, then you are starting with a lower quality image that if you had kept the original. If already demosaiced you can't benefit from any demosaicing algorithm improvements in the future. Over the years I have reprocessed several images using more up to date versions of ACR process than were available when the raw files wer saved.

I know it is an worn expression, but disk space really is cheap these days particularly if you don't need fast access. I don't keep every image (technically sub-standard go in the bin) but those that are worth keeping at all I keep at full raw.

Dave

[D Fosse]

Another way to look at it is that raw files are "naturally" compressed to begin with, since all the information is recorded in a single channel and a predetermined Bayer filter array to "uncompress".

Squeezing full file size down to 1/3 isn't bad, right?

[davescm]

Plus of course the use of lossless data compression on that file - it is indeed an efficient storage of all that image data.

As an aside, I am surprised that 3 colours at each camera pixel site as per the old Fovean sensors has not been pursued further.

Dave