Should I merge first, then convert with NLP -- or the other way around?

I am digitizing 120 film with a crop sensor digital camera. I have been filling the full width of my sensor with part of the negative, then moving the negative so I can copy the rest of it. And finally, merging the two files in Lightroom as a panorama to create a square image.

Before merging, each RAW file is 3264 x 4896 pixels, and the uncompressed RAW file size is about 32 MB.

Apparently, Lightroom is converting the merged image to a DNG format with pretty heavy compression, because the merged file is about the expected resolution (approximately 4800 pixels, square) – but the file size is only about 6 MB. So, if I merge first, the NLP conversion will apparently being done using compressed files rather than RAW files(?)

I am wondering if there would be any advantage to first converting the RAW files with NLP, and then merging them in Lightroom?

Or should I be looking for some other software that will merge my RAW files without compressing the heck out of them?

@runswithsizzers , you could test this yourself, as you seem to have all necessary software.

As far as I checked it out, merging converted RAWs can create strange colours, unless you’re really careful to e.g. completely reset the source image settings. Results will vary anyways because of changes in colour and tonality due to both LR and NLP fiddling with colour and tonalities.

My testing with three (3) captures of ca. 30 Mbit file size produce DNG panos of about 40 Mbit file size, no matter whether I stitch and convert or convert and stitch. DNG files can be compressed quite a bit, e.g. if the image has not much detail, noise and contrast. Source images in my tests measure 6000x4000 pixels and the pano gets to be about 6000x8000 pixels.

The benefit of stitching in my case (I mostly used high iso film) is in the visibility of the grain mostly and I therefore decided to try stitching only rarely as opposed to “super resolution”.

@Digitizer - I should have mentioned, all of my 120 film will probably be black and white, so no color issues. I may shoot some 135 color film, but I don’t normally try to stitch 135.

Yes, I am trying to test this myself. So far, I am not seeing any clear differences in IQ, altough I am seeing some exposure differences. Unfortunately, this first set of 120 negatives I am working with have some sharpness issues, so I am unable to evaluate fine detail.

I was wondering if there was any sound theory why one way might be preferable? Or if I even understand what Lightroom is doing?

You say you are getting a merged image which has a file size larger than your source images? That is what I would expect, but my merged image has a much smaller file size than my source images. Are you using Adobe Lightroom Classic to merge your files? If so, what version? Apparently, somewhere around v.13, Lightroom changed to a different compression scheme for merged panoramas(?)

Here’s my test in colour


Left to right: Negatives and VCs converted presaturation set to 1, 2, 3…
Top row: One shot negatives
2nd row: Merged image with negative components selected downwards (per column)
3rd row: Top component
4th row: Middle component
5th row: Bottom component
6th row: Merged images with converted components selected upwards (per column)

We see that rows 2 and 6 show slightly different colours (and possibly tonalities)
This is indeed caused by how the pano is stitched (first to last or last to first) and NLP amplifies the differences. Here, the results have a slight touch of Hulk, sorry green.

As for B&W, presaturation does not seem to have any influence, therefore your test should be easier to execute and document. Please add your findings here.

I tested this with Lightroom Classic 13.4. Newer versions tend to have problems with NLP.

Post your captures (or a link to a share) if you want me/others to check your files and system behaviour.

I am still investigating the cause of poor image quality of my 120 negatives. I think the negatives are not sharp, out of the film camera.

Only if I can get better quality negatives will it make sense for me to compare the quality of my camera scans.

My version of Lightroom Classic is 13.0.2.

Tell me again, what is the file size of your merged DNG images in MB (megabytes)?

I am having trouble accepting the idea that my file sizes can go from 32 MB for the RAW originals to 6.5 MB for the merged file. That may be plenty of quality for my final exported JPEG, but I usually do a fair bit of editing after converting, so I’d rather preserve more data in the merged file until later in my workflow.

See file sizes of the test I documented above :
Bild vom 2024-09-09 um 13.12.47

Note how the pano names reflect the name of the first file that was selected for merging.
…1621.cr2 contains the full image of a 4.5x6 negative, other cr2 files only contain parts.

As previously mentioned, when I merge two RAW files in Lightroom Classic the resulting DNGs are much smaller than yours – only about 6 MB
Screenshot 2024-09-09 at 9.57.00 AM

But I have discovered another way. In Lightroom Classic, I select the photos I want to merge; then, from the Lightroom menu, dropdown to “Edit In” and select “Merge to Panorama in Photoshop.” Photoshop gives me the option save the merged image to a variety of different file formats, including 16-bit TIFFs, which are running about 130 MB. Probably, a lot more than is needed for b&w negatives. But if the images were from high dynamic range transparency film might help prevent clipping or banding??

After I get the images fully edited, then I will probably convert those big TIFFs to JPEGs, and archive the TIFFs somplace beside my main working SSD.

I merged the images like this:

  • Select the images
  • Press ctrl-M
  • Select “Perspective” stitching and let Lr do its work

Again, I use Lightroom Classic 13.4 on macOS 14.6.1 on Intel iMac 2019

As I have no RAF files around, I cannot test if the small dng files have to do with x-trans vs. bayer pattern CFA. Maybe only the previews are merged? Or Smart Previews?

Merging in PS and exporting as TIFF-16 gets me files of >1.5 GB, definitely not what I want.
Exporting as TIFF-16 a pano from Lightroom produces files of about 300 MB.
:thinking:

It sounds like you are confusing a losslessly compressed file with an uncompressed file.
Also, DNG is not a lossless format - at least not exclusively. It can be compressed in either a lossy or lossless way - or it can be uncompressed.
DNG files can use various compression methods to reduce the file size without compromising image quality.There is zero loss of quality with a losslessly compressed file even though the file size reduction can be considerable. This feels counter intuitive but is true. It’s the same with losslessly compressed camera files - they are much smaller than uncompressed but retain all the data.
The most common compression method used in DNG files is lossless compression, which preserves all of the original image data while reducing the file size. This is achieved using algorithms such as Linear Raw, Deflate/ZIP, or Lossless JPEG, which are all supported by the DNG format.
In addition to lossless compression, the DNG format also supports lossy compression methods, such as JPEG and lossy DNG compression. These methods can provide smaller file sizes at the expense of some image quality, which may be acceptable for certain applications or storage requirements.

Yes, but there is no way to influence Lightroom’s algorithms in the window for merging photos for a panorama.

Adobe’s DNG Converter features different compression methods and Lightroom exports to DNG with or without lossy compression…which shrinks file size to about 1/3 the size. But again, merging a panorama has no option for DNG compression - unless Lightroom takes it from what is set in the export dialog, which I’d find very odd. Nevertheless, I recreated a pano I made for the previous tests and it turned out to be 18 MB only. :thinking: :thinking: :thinking: :thinking: :thinking: :thinking:

Being a wildlife photographer I hardly ever do panos as those pesky critters won’t stand still long enough to set up a tripod, attach all the gear, check the alignments, composition, exposure, etc :slight_smile: so I did some research and came up with this… LR now saves the merged image as lossy compressed DNG format rather than lossless DNG as in previous versions.
Lightroom Classic (and ACR) recently added support for a new DNG format which enables the ability to create RAW files which are 92% smaller with no visible loss of quality! This is made possible by using a new “lossy” image format based on the new JPEG XL (aka JXL) file format in DNG v1.7.
The results are extremely good and would be indistinguishable from the original in nearly any real scenario. The loss of quality is nearly undetectable for the vast majority of RAW files.
There are no settings that can be changed to alter the degree of compression or choose a different format. The only alternative is to go into PS to make the merged file.
Hope this helps.

Must have happened in the last 4 days then, even without updating Lightroom Classic? I’m still on 13.4!

And even more confusing: Just did that pano again and file size was back at 41 MB :exploding_head:

Just quoting the blurb.
I’m on Lightroom 13.5.1

Thanks for the additional information. It may be that Lightroom’s new lossy DNG format based on JPEG XL does allow files to be “92% smaller with no visible loss of quality.” I would be OK with that if the resulting file was my final edit.

My concern is this: How well will this highly compressed JPEG XL file hold up to further editing such as exposure changes, highlight and shadow recovery, white balance and color manipulations – and whatever kind of edits NLP does?

Until I can find the answer to that question, I am sending my overlapping photos from Lightroom to Photoshop for merging, where I have more control over the format / file size.