Converting VueScan DNGs into Adobe lossless/lossy DNGs?

Hi everybody,

I’ve been using NLP with Lightroom Classic, VueScan, and my Epson V600 for some months now and couldn’t be happier. I love the results and think the workflow (I follow the NLP VueScan tutorial) is pretty easy and straight-forward.

The first few rolls I scanned with an (as I now know unnecessary) 6400dpi, resulting in files of around 400MB each. I recently looked into compressing these huge files and found that Lightroom can convert VueScan DNGs into its own lossless DNGs or lossy DNGs. I tried both options with some individual photos and found that the Adobe lossless conversion decreases the size by around 25-40%, and the lossy DNG conversation by 95+% (a 400MB picture then has less than 10MB, and according to the metadata the same bit depth and resolution).

This makes me wonder what the differences in quality are between these 3 DNG types**: a) VueScan RAW DNG, b) same converted into Adobe lossless DNG, and c) same converted into lossy Adobe DNG?**

I found different answers and opinions online, saying that the difference in size between a) and b) could be because of

  • Adobe’s better DNG compression compared to VueScan’s, or
  • Lower-resolution (or, when selected, no) jpeg previews embedded in Adobe’s DNGs as compared to full-res jpeg previews embedded in VueScan DNGs
  • the infrared dust&scratch removal layer only being embedded in VueScan DNGs

Does anyone have any experience or opinions on that?

If in the end the exact same information is stored in Adobe’s DNGs as in VueScan’s, I would just convert all of them into Adobe DNGs and save 25-40% of disc space without compromising any quality.

And maybe Nate has any experience on whether it’s better converting the VueScan DNGs into Adobe DNGs before or after running the ‘Update Vuescan/Silverfast DNGs’ script?

Regarding converting a) or b) into c), the Adobe lossy DNG format: Any experiences on how much such conversion decreases the quality and processing abilities of the DNGs? Will I notice a difference when lightning up the darks or turning down the highlights in NLP?

The reason I’m wondering about all of this is that I’m also using JPEG mini to reduce the file size of all my JPEG files – and it reduces them by a lot, without any visible differences for me. I could see that the scans from a ‘cheap’ consumer-scanner like the V600 could also just be blown up and contain lots of unnecessary information (looking at the fact that even at a 6400dpi scan setting, the Epson can only physically do around 1600dpi and the rest is just upscaling and enlarging the file size without an increase in sharpness or anything else). So maybe the conversion from VueScan DNGs into Adobe lossless or even lossy DNGs just compresses better or cuts out some artificially blown up / unnecessary information – and converting VueScan DNGs into lossless/lossy DNGs would not decrease the quality and editing abilities but only safe space.

But if the Adobe DNG conversion deletes the ICE layer or any other necessary information, the conversion would of course be a no-go.

Any insights you might have would be greatly appreciated!

DNG is a standard developed by adobe, but open to anyone who wishes to implement it. Particularly camera manufacturers. It’s become the standard for archival to the point that VueScan uses it as well. Under the hood DNG files are basically just TIF files on steroids. If you’re curious about the standard per se and want to dig deeper, here’s the link to the spec.

That being said, the way that this standard is used to suit the requirements of anyone implementing it varies and in the case of Vuescan for instance, it was used such, that reopening a file you scanned and output as DNG is practically equivalent to scanning it again. You can modify every single development setting as you would expect from a raw file. Further Vuescan can also use fields to encode extra data such as the IR record from a Digital ICE enabled device.

You should think twice whether it is really worth converting your files, because you will likely lose the ability to use these features. I personally also don’t see the benefit in saving 20-40% on storage, when you can buy terabytes of HDD storage for <100$ nowadays.

If anything, try scanning at a reasonable resolution from now on, if you feel like your files are filled with redundant data. Using Vuescan to downsample these files to a lower resolution might also be an option

Hi, and thanks a lot for your response. I think I’m still unclear about some points, it would be great if you could provide more background there:

  • If in LightRoom, I convert my VueScan-created DNGs to (what I called ‘Adobe’ lossless) DNGs, I would lose that IR-channel information, even though it’s the same fileformat and included in the original DNGs? So Adobe deletes information when converting an ‘externally created’ DNG to DNG via Lightroom, even though it’s called lossless? I’m wondering if there’s a way to check this, like can I deactivate the IR channel in Lightroom somehow? For now, my Adobe-DNGs look the same, with dust and scratches removed, like my VueScan DNGs.

  • If I scan with my V600 at 3600 dpi which yields an effective resolution of around 1500/1600 or so dpi, but creates file with sizes of 3600dpi files, how do I downsample in VueScan? If I set it to 1600dpi, I understand I would again get even less effective DPI than the 1600dpi selection. Is there a setting to scan my negatives with the (max) effective resolution without interpolating/upscaling by the scanner?

Thanks a lot!

1.) I can’t comment on this, as I don’t really use Lightroom all that much anymore, let alone in conjunction with Vuescan. What I do know is that NLP doesn’t work with 4 channel DNG files(IR included in the DNG) which leads me to believe, as you are posting on a NLP forum and hence presumably using this software that you are not actually exporting the IR channel in the DNG but probably “baking” the IR cleaning into the raw data for RGB. There are tools to investigate this, I reccomend you peruse this thread and you should be able to find some info.
2.) If you’re scanning at 3600, your scanner/vuescan almost certainly can also do 1800 as the density settings usually increase/decrease by factors of 2. If I were you I’d set it to 1800 and call it a day. My personal opinion is that persuing “ultimate”/archival quality with an epson flatbed is a bit of a fool’s errand anyway. Don’t get me wrong I use one as well for MF images I wish to post online, share with people etc., usually set around 2000 DPI I believe, but I wouldn’t lose any sleep over whether 1800 DPI setting is leaving a little bit of resolving power unused. I highly doubt that’s the case anyway.

Think of your initial scans as the thing closest to the film negative or transparency. Unless you want to re-scan, I’d keep these files. Everything you want is there: RGB pixels plus a dust removal layer. And you say that conversions are good.

If drive space is your main concern, move the files to an external drive or two and continue with scans set to 2400, which will print to about 8x12 inches or larger, if seen at bigger distances.