Let's discuss photo stitching!

Taking multiple partial shots and stitching them together can significantly increase the amount of information extracted from a negative. There are two questions (for me, at least):

  1. Common wisdom suggests a 30% overlap between partial shots. If one is stitching a MF negative with (say) a cropped sensor camera, that might be as many as 6 shots or more. Do you go top, from left to right, and then bottom from left to right? Or what is the most effective way? Or, can one shoot only the upper (or left) half and stitch that, and then the lower (or right) half and stitch that, and then stitch both halves?

  2. For the purposes of just stitching the shots together, is any stand alone software (such as PTGui) inherently better than the stitching programs within LR or PS? What makes a particular stitching program “better,” if all one is trying to do is stitch the photos and afterwards process them in LR or PS?


Hi @bags,

A few thoughts to answer your questions.

  1. 30% overlap should be good. Just know that the more shots you try to stitch together, the more likely it is that you’ll run into issues getting it to stitch together. This is especially true in scenes that have lots of “empty” space, like sky. With 6 shots, you will be quite close to the image and may end up not having enough context in each frame for the software to find the right place to stitch. For medium format, my approach is typically to do just 2-shot stitches, which is an easy way to double resolution, and it will stitch very easily in Lightroom or anywhere else you try.
  2. PTGui will offer more options than Lightroom or Photoshop. And Photoshop will offer more options than Lightroom. Lightroom’s options are limited, but the biggest advantage of Lightroom is that the resulting stitch remains fully RAW, so you are not losing any of the underlying raw camera data and can just continue to work off it non-destructively. You can see more of my tips on pano merges in Lightroom here. If you plan on stitching together more than 2 frames, you might be better off doing it in Photoshop or PTGui.
  3. The most crucial part of getting this right is your camera scanning setup. First, you want to make sure that your camera is perfectly in plane with your film. You can use the mirror alignment trick to do this. Second, you want any movement of your film between shots to keep it perfectly in line with the previous shot… having a film carrier that lets you mechanically advance the frame is ideal… if you have to try to move your film by hand, it’s likely it won’t be perfectly parallel with the previous shot.

Hope that helps!


One thing can make stitching difficult: Lens distortions.

If a lens that has a lot of distortion, stitching will be quite difficult. If you can use a software that eliminates distortions before stitching, results should be better.

For stitching (which I do rarely) I preprocess shots in DxO PhotoLab, which takes into account focal length and distance settings of the lens model used for a shot, if the lens is supported. Corrections include vignetting, distortions, chromatic aberrations, differences of sharpness in the frame and more, all without leaving a RAW workflow. That last capability is quite valuable.

Workflow: Undistort (DxO) - Stitch (Lr) - Convert (NLP)

Thanks so much Nate and Digitizer. A lot of very helpful suggestions here. One of the most important takeaways for me is that less is more. That is, trying to squeeze every last datum from a negative by stitching many photos is probably counter-productive. Systematically improve process to a realistic outcome. Thanks!

Stitching can do this and the difference is that the stitched image will show much clearer film grain… :wink:

I’m curious how this would be different than the lens corrections already offered inside Lightroom Classic?

You can select pre-built lens profiles…

… or manually adjust controls…

… or even create your own custom lens profiles from scratch using the free Adobe Lens Profile Creator.

DxO lens/camera modules contain all data on specific camera and lens combinations. If you check out dxomark.com, you can see that lenses perform to different levels, depending on aperture, distance, zoom settings and on the camera which it is attached to. Combinations are measured and correction data is provided and applied automatically in DxO PhotoLab based on recorded metadata. Because of the sheer number of lens/camera combinations, not every possible combination can be measured though. Supported gear can be checked on the respective pages.

There are other products or possibilities to correct lens distortions of course. I simply happen to have PhotoLab (had it for years) and use it as needed, the new noise reduction tool is the best I know of, and provided lens/camera modules save me from having to do it myself. No matter if a lens has barrel, pincushion or moustache style distortions etc., all those defects simply go away.

CAVEAT: If your gear is not supported, you miss out on a few of PhotoLabs best features.

Oh very cool, thanks for sharing. Looks like my Fuji X-T2 is not supported. Would love to test and compare.

yes, Fuji X-Trans is not yet supported. As of now, DxO does Bayer sensors only :face_with_raised_eyebrow:

Get a GFX 100 instead? :innocent: Or the new GFX 100S?

1 Like