7
\$\begingroup\$

I have a set of three images imported in Lightroom 3.6 which I want to stitch into one panorama using Photoshop's CS5.5 Merge to Panorama tool. As I see it I have two options in doing so.

  1. Doing all modifications (colour, brightness, contrast, etc...) first, syncing the changes between the three photos and then stitch them together.

  2. Stitching the photos first and then use Lightroom to apply modification to the whole image.

Which of these two methods is considered to be best practice?

Improve this question
\$\endgroup\$
1

4 Answers 4

Reset to default
10
\$\begingroup\$

As a general rule, the stitching program should get the highest possible detail of the images, allowing it to make the best out of it. Therefore, I would not pre-process the images as it might degrade the information.

However, if you have a large amount of input images, it might get difficult to work with the resulting large picture afterwards, due to your PC's memory and CPU constraints. Only then I'd revert to pre-processing.

There's only one situation where I was forced to pre-process: When I had images of varying brightness, due to having taken them close to sunset, and the light getting darker during the panoramic 360 session, so that last image was darker than the first, making it hard to get them properly stitched. In this case, I had to gradually adjust the brightness of all images to make the last one match the first one more closely before stitching them.

Improve this answer
\$\endgroup\$
4
  • \$\begingroup\$ I was going to say the same thing, good answer. I would add, this is all assuming that the images were taken "correctly" with the same camera settings and exposure. \$\endgroup\$
    – dpollitt
    Commented Apr 22, 2013 at 0:21
  • 1
    \$\begingroup\$ @dpollitt All good stitching programs can handle images of different brightness and smooth the variations between images. Provided you use the "correct" technique and shoot with a lot of overlap between images, you don't have to keep the same settings for each shot, allowing them to vary will maximise the dynamic range you can capture. \$\endgroup\$
    – Matt Grum
    Commented Apr 22, 2013 at 7:34
  • \$\begingroup\$ @MattGrum - Interesting, I did not know that. It has been a few years since I did much stitching at all. I used to use PTAssembler by Max Lyons but that is very old at this point. \$\endgroup\$
    – dpollitt
    Commented Apr 22, 2013 at 14:42
  • 2
    \$\begingroup\$ @MattGrum - I had tried a few stitching apps and they failed at the gradual decrease of brightness in my 360 pano sequence - all they did was to "smooth" the first and last picture by adjusting the area where they overlapped, which was not creating a good result at all. I don't know of any app that would understand that it has to change the brightness gradually of ALL involved pictures in the panorama, unless there's a way one can direct them manually to do that, but surely not automatically. \$\endgroup\$
    – Thomas Tempelmann
    Commented Apr 22, 2013 at 15:18
7
\$\begingroup\$

Since I prefer to do as much editing of things like brightness, WB, contrast, etc. as possible before converting from RAW, I tend to edit each before stitching. I save final sharpening adjustments until after the merge. The nice thing about using Canon's Digital Photo Professional, which I normally use for RAW conversion, is that you can copy/paste a "recipe" from one photo and apply it to the others. Assuming you exposed all of the frames at the same settings this works out quite well.

There are programs that will allow you to import RAW files for stitching, but I am not aware of any that will export the combined files as a RAW file. The best you get at that point is a 16-bit TIFF. Once converted to TIFF the advantage of RAW for things like WB correction, recovering blown highlights, and changing exposure more that +/-1 stop without degrading the image are lost, so to fully leverage these tools they need to be done before the stitching.

The workflow I use is:

  1. Import RAW files into DPP, apply edits (including initial sharpening using the unsharp mask), and then export as 16-bit TIFF files.
  2. Use a stitching program to merge the images and export as a 16-bit TIFF (these files can be rather large).
  3. Use PS or DPP to touch up any areas around the seams if needed, do final sharpening adjustments, and export as a jpeg. If more than one display size is intended: repeat sharpening and export at the appropriate resolution for each size.
Improve this answer
\$\endgroup\$
6
  • \$\begingroup\$ Interesting point about losing headroom with 16-bit TIFF. I sort of assumed that 16-bit TIFF would have the same capabilities as RAW Canon .CR2 files. \$\endgroup\$
    – Saaru Lindestøkke
    Commented Apr 22, 2013 at 5:20
  • \$\begingroup\$ Nope. There is still more latitude for adjustment than an 8-bit image, but demosaicing, which essentially sets WB, exposure, and contrast has already been "baked in". \$\endgroup\$
    – Michael C
    Commented Apr 22, 2013 at 5:56
  • \$\begingroup\$ You can copy/paste "Develop settings" in Lightroom too..... (similar to your "recipe" in DPP). I'm not a fan of DPP, have never gotten on with it. \$\endgroup\$
    – Mike
    Commented Apr 22, 2013 at 10:32
  • 1
    \$\begingroup\$ Another nice thing about the "recipe" is you can apply it to all selected images without having to open and paste it to each one. I've found on a few discussion boards that the reason a lot of people don't like DPP is because it "can't" do something it actually can - they just haven't bothered to find out how to do it in DPP. \$\endgroup\$
    – Michael C
    Commented Apr 22, 2013 at 10:52
  • 1
    \$\begingroup\$ Sony's Image Data Converter works the same way with "recipes". Process one image of the set; save recipe; apply this recipe to all remaining images at one go without a need to open each one by one. \$\endgroup\$
    – Esa Paulasto
    Commented Sep 23, 2013 at 3:52
2
\$\begingroup\$

After playing with stitching for a while now, I am with Michael on this one.

I tried it both ways and got much better results by doing most of the adjustments inside Lightroom first (sync what needs to be synced) and then do the stitching. After stitching and refining the layers where CS5 did not quite get the transitions right, dodging and burning and such, I do Cmd-Option-Shift-E and take the result back into Lightroom to do my final (local) adjustments like sharpening.

As a side-note: I found the tutorials on the subject by Serge Ramelli on Youtube quite helpful.

Improve this answer
\$\endgroup\$
1
\$\begingroup\$

I wrestled with this decision just last night myself! Having never done a panorama stitch, I decided the following:-

  1. In Lightroom, I simply ticked the boxes one one image in the set for Lens Correction and Chromatic Abberation correction ON.
  2. Copied the develop settings to all the other images in the set.
  3. Highlighted them all to send to Photoshop Elements
  4. Stitched them together using the "Auto" method. It did really well!!!
  5. Cloned in to fill the edges of the panorama as there was a bit of 'transparent' void around the edges (note: Elements did offer to try to do this for me but when I said yes it just said I didn't have enough RAM!!! -- I have 8Gb!)
  6. Flattened and saved to .tif.
  7. Continued the entire rest of my processing in Lightroom...

I really like how it came out. Will post a link later this evening to the result.

Hope that helps?

PS, should also add that when I took the pictures, I made sure there was a good amount of overlap in each one as I read somewhere that really helps when auto-stitching pics...

Improve this answer
\$\endgroup\$
2
  • \$\begingroup\$ For step 5: cloning in the void around the edges takes a lot of RAM if you use the "content-aware" option. Just clone in smaller patches of void one by one. \$\endgroup\$
    – Saaru Lindestøkke
    Commented Apr 22, 2013 at 12:20
  • \$\begingroup\$ That's what I ended up doing :) Using the clone tool bit by bit, constantly resampling, constantly re-cloning until an area of transparency was painted over reasonably well, then kept using a combination of smaller more detailed cloning from one place to another so it wasn't obvious, as well as using the content-aware 'healing' tool to remove odd replicated rocks and grass areas and so forth... Time consuming but I like the result and it doesn't look cloned at all I don't think :) \$\endgroup\$
    – Mike
    Commented Apr 22, 2013 at 12:24

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.