Stitching Images For Larger Prints

0Comments
For this image I decided I wanted to cover all of Lower Manhattan from the Statue of Liberty (far left) to the Empire State Building (far right, under the Brooklyn Bride, colored red, white and blue. I was using a Canon EOS 5D Mark III. My EF 24-70 f/2.8L II at 70mm covered the skyline and water with some sky vertically, so I positioned the camera vertically and proceeded to take 9 shots, moving the camera by turning the tripod head on it's rotating base. I overlapped portions of each frame so Photoshop would have a point of reference when stitching. Each of the nine exposures was taken at ISO 200, 90 seconds, at f/16.

For this image I decided I wanted to cover all of Lower Manhattan from the Statue of Liberty (far left) to the Empire State Building (far right, under the Brooklyn Bride, colored red, white and blue. I was using a Canon EOS 5D Mark III. My EF 24-70 f/2.8L II at 70mm covered the skyline and water with some sky vertically, so I positioned the camera vertically and proceeded to take 9 shots, moving the camera by turning the tripod head on it’s rotating base. I overlapped portions of each frame so Photoshop would have a point of reference when stitching. Each of the nine exposures was taken at ISO 200, 90 seconds, at f/16.  If printed at it’s native resolution at 300 dpi, it would measure 18.39 inches by 88.25 inches.  My photo lab maxes out at 108″, which it says it can print this image to.

Several months ago I was asked by a potential client if I had any images that were capable of being printed very large- up to 20 feet across! It pained me to explain that, no, based on my camera’s resolution, I did not have any images capable of being printed that large.  I had never gotten into doing many stitched panoramas or other prints, and couldn’t afford a Gigapan or other panorama photography tool.  For the most part, I’d had no call for it in my daily business. Generally, when shooting landscapes, I think in terms of one frame, and fill it with my composition. This has worked well for the most part, as long as I didn’t want to print much larger than around 48″ inches across.  Suddenly, however, I had a desire to go much larger.

This past week in the United States, we commemorated the 12th anniversary of the September 11th terrorist attacks. Every year New York City remembers the victims with a tribute in light- two columns of light representing the fallen towers of the World Trade Center.  It seemed a perfect opportunity to start playing with panoramics, especially since one of the images the aforementioned client wanted was a skyline shot of New York City.

The general rule of thumb for printing on standard inkjet printers is to print at 300 dpi.  To find out how large you can print an image, simply take the pixel dimensions and divide by 300. From a camera such as the EOS 5D Mark III, that means an image of 5760 x 3840 pixels can be printed at about 12.8 inches by 19.2 inches.  It is true, using various resizing techniques you can print larger.  I have on many occasions. But to get to the extreme sizes beyond approximately 48″, you’ll need to start combining images by stitching them together.

There are currently a few automated panoramic photo options on the market, including Gigapan’s and Panogear’s. Both can be somewhat pricey.  But just because you don’t have these nice accessories does not mean you can’t make stunning panoramic images.  A tripod is helpful, but not completely necessary if you can handhold the shutter speed your camera is set to.  A tripod is helpful for locking your camera in place from shot to shot. The reason a tripod is helpful is that if your tripod head has markings for panoramics on the base, you can use these for reference when repositioning the camera for each shot. More on that in a bit.  Another helpful tool is an L bracket. This will help you position your camera vertically if desired to shoot verticals to stitch the final piece.  L brackets can be purchased from several manufacturers and are usually camera-specific.  Acratech makes a universal L bracket with a quick release that any camera can attach to using an Arca-Swiss style plate.

You’ll want to start by defining your image in your mind. Where does it start, where does it end? Then, how far up does the image go, and how far down?  can you cover the up and down with one vertical?  Or would you be better off shooting two rows of images. Or more? Keep in mind when planning that you’ll want to shoot with some extra area around the image to leave room for cropping if needed.  You’ll also want to make sure you leave some overlap in each shot so the stitching software can find a point of reference to see where the next shot goes. I used Photoshop for these, but there are other programs out there. Feel free to suggest your favorite in the comments below.

For the first image in this article, I wanted to shoot the Lower Manhattan skyline, from the Statue of Liberty to the Empire State Building.  I had a 24-70mm lens, and at 70mm I covered exactly what I wanted, from top to bottom, with a vertical shot.  I took 9 shots in that orientation, while rotating the tripod head incrementally until I got my last shot.

For the second shot in this article, there was a more prominent foreground element, the pilings from the old pier.  I decided to do this one in a horizontal orientation, using two rows of three shots.  This was again taken using the 24-70mm lens, at 70mm.  I shot this one starting at the far right, shooting in columns- upper right, lower right, lower center, upper center, upper left, lower left.  I used the playback feature on the camera to check my reference points. Again I stitched it using Photoshop’s Photomerge feature.

This shot was six horizontal shots stitched together. Because of the overlap, the 3 shots across and 2 up and down will not add up to the full resolution of the individual images simply put together. This image came together at  10,531 pixels by 5904 pixels. At 300 DPI the file can be printed at 19.68 inches by 35.1 inches. My lab, however, tells me they can print this image up to 8 feet wide.

This shot was six horizontal shots stitched together. Because of the overlap, the 3 shots across and 2 up and down will not add up to the full resolution of the individual images simply put together. This image came together at 10,531 pixels by 5904 pixels. At 300 DPI the file can be printed at 19.68 inches by 35.1 inches. My lab, however, tells me they can print this image up to 8 feet wide. They are not using inkjet printers and thus are not subject to the same parameters.  The same still holds true however.  The larger the file, the larger it can be printed.

I have not yet heard anyone say that any photo stitching program is perfect.  There will be errors in stitching.  A misplaced post, a skewed building. To correct these, I simply opened the source file and added it to the stitched file on a new layer.  Then I created a layer mask to show only the area I wanted shown, which would correct the issue.

For your exposure, you’ll need to be in manual mode. You need the exposure to be uniform across the image.  If you leave your camera in any auto mode where the camera helps set the exposure, you run the risk of your exposures varying.  For the first image in this article, my exposure was 90 seconds at f/16, ISO 200 for each image.  This is important, particularly when photographing the area around the statue of liberty which had huge dark areas.  In auto mode, the camera will try to brighten these areas, which will cause problems when the stitching if the skies or water don’t match from shot to shot. In the interest of full disclosure, I made this mistake myself with the second image, the six-shot stitch.  I shot in aperture priority and there was a variation of plus or minus 2/3 of a stop from shot to shot.  This caused all kinds of headaches in my first attempt at stitching.  I was able to correct this by reproccessing the RAW files with an exposure adjustment to match the exposures.  In addition, if your camera has a feature for vignette correction, such as Canon’s Peripheral Illumination Correction, turn it on.  This will even out the exposure so there are no dark areas in the corners, which can be difficult to correct later on.

I’ve toyed with stitching panoramics before, but never seriously.  This is one of my first attempts at a serious pano.  It’s well worth exploring more in the future.  I might even start saving for a new piece of equipment just for that purpose!

 

Read more from our Post Production category

Rick Berk

is a photographer based in Freeport, Maine, shooting a variety of subjects including landscapes, sports, weddings, and portraits. Rick leads photo tours for World Wide Photo Tours and his work can be seen at RickBerk.com and you can follow him on his Facebook page and on Instagram at @rickberkphoto.

  • Rob

    Nice article, I’ve never thought of doing pano’s with night shots as I also thought there would be too much difference in the sky (or water in this case). Great tips.

  • Very nice article and superb pics

  • Aju Jayapalan

    This is awesome….I am still learning how to shoot and stitch panoramas. This one of the Chicago skyline is the largest one I have ever made…still needs some touch-up though. Had taken 42 images in portrait mode (two rows). 42 because I was paranoid shooting hand-held and did more than 50% overlap (also I am very inexperienced). The bad part of taking so many images was when Photoshop was trying to stitch them, it took about half an hour on my laptop.

    [eimg link=’http://www.flickr.com/photos/ajujayapal/9867060515/’ title=’Chicago_Downtown_Small_72′ url=’http://farm8.staticflickr.com/7421/9867060515_312d947272.jpg’]

    The dimensions of the uncompressed final image was 29322 x 11120 px (300 dpi) and close to 2 GB in size! I have never done this before and am very excited!

  • Tom

    For hugin, an open source panoramic software, did a pretty good job. Very versatile and cheap (it’s hard getting cheaper than nothing)

  • Dan Merkel

    Good point about shooting in Manual mode to keep things consistent throughout the series of shots. But don’t forget to set your White Balance as well; oftentimes, people use Auto WB and this too can change from shot to shot.

    For a very large print as described, I’m not sure that 300dpi is necessary. A print that is 4-5 feet wide won’t be examined at close range like a 4×6 inch or 8×10 inch print will.

    For really wide prints, check with commercial printers in your area. Our local Allegra shop http://www.allegrafindlay.com/ can print pictures up to 54″ tall and as long as the roll of paper it is being printed on. This could easily be 15-20 feet or more in width. This comes with excellent results, I might add.

    dlm

  • Josh

    Nice read! I too am beginning to explore panoramic images for the purpose of being able to print large images.

    The things I have learned (the hard way) to control for each image in a stitched photo are: aperture, shutter speed, ISO, focal length, focus point (manual focus is a must), and white balance. For some of these, you can compensate a bit when shooting in RAW and make a few adjustments in the computer prior to stitching. Make sure you check the exposure throughout the entire view you want to capture. While exposure might look good facing away from the sun, you’ll end up with a burned out image the closer you pan to the sun at the other side of your image. Also make sure you include more area in your capture that you need. Depending on stitching parameters and image orientation, cropping will be necessary and may be significant. It is very frustrating to see that an important piece of your composition needs to be cropped out due to an adjacent picture not being perfectly aligned.

    Another thing to keep in mind is final image file size. In RAW on my Rebel XSi (APS-C sensor), a typical image might be 10-15MB each. Combining a number of images and saving as 16-bit tiff gives me a 330MB file. Many computers (mine included) don’t like handling images that large and post-processing becomes more challenging.

    I know every image will be somewhat unique, but if the second shot in the post was printed at 8′ wide, horizontal resolution would be in the range of 110 dpi. Again, I’m new to this but that just seems too low. While the “intended” viewing range might be significantly further away to take in a whole image, a primary reason for printing large format isn’t to let the overhead jetliner see your image. To me, it is to draw the viewer physically closer to the image and see all of the tiny details that wouldn’t necessarily be visible if taken as a single image. It’s a more immersive experience for the viewer that way.

    I’ve got a recent image from Glacier National Park’s Avalanche Lake that I’m considering trying at 200 dpi but may do a test print or two of a few locations in the image to see how it really turns out at a range of resolutions. The full image size is 11760 x 4896 so 4′ wide should be possible. 5′ would be closer to 200 dpi and may be the limit for this image.

    -Josh

  • Andy

    Don’t make the mistake of confusing your image resolution with the printing resolution – they are unconnected.
    A printshop will be using 1440 or 1200 dpi to output your photos at normal sizes. Printers used for larger work will still be using 600dpi as a minimum.
    They can all print at 300dpi but there’s little point and they probably don’t have that resolution profiled anyway unless it is a grand format printer 3m or wider.
    Think of your image resolution as making an image out of blocks – the smaller the building block the more detail you will have.
    Now think of the printer resolution as the amount of dots it will use to print the blocks.
    If the blocks are big or small (low res or hi res image), the number of dots used by the printer to print these blocks does not change.
    If you have large blocks (low res image) the lack of detail becomes quickly apparent on enlargement although the number of dots used to print these blocks may be high.
    If you are using a high res image these small blocks of info can be enlarged some way before they become of a size which becomes unacceptable but again the number of dots used to print these blocks does not change.
    The only time the actual printer resolution becomes an issue is if the ink dots themselves are visible which is only an issue on close inspection – anything more than arms length and dot size tends not to be an issue so most common wide format printers select a compromise between quality and speed and use 600 or 720 dpi for most work.
    Just don’t use printer dpi as a guide to how you handle your image file size – your printshop will decide this independently regardless of what you do with your files.
    Your job is to make your image building blocks as small as possible and the only way to do this is by the CCD size in the camera or by the zoom and stitch method outlined above.

  • josh

    You’re correct, Andy. I should have been using ppi instead of dpi. Thanks for the correction!

    Point still stands, 300 ppi is considered “photo quality.” Is it possible to have an image printed with a size of 100 ppi and still be “photo quality”? That’s the point of my comment. Do the rules change when printing large format? Perhaps it simply depends on the photo. Perhaps it depends on the intended use/audience. Something to ponder.

    Don’t want to take away from the post. Well written, and a great way to gain those extra pixels necessary for large format printing. It’s also a great way to give me extra pixels to work with using my 4 year old entry level dSLR without having to spring for a new camera (sensor)!

  • Jim

    Microsoft Research has the Image Composite Editor ( Microsoft ICE). Process your RAW files, save them as lossless jpgs, then drop them into ICE. Totally free, amazingly accurate, and surprisingly fast. I have used it to stitch 100+ images spanning over 180 degrees horizontal and 5+ lavels vertical.

  • John G

    For those without Photoshop – provided they’re using Windows, there is a good, easy & free stitching program. Image Composite Editor (ICE) appears to have been a research project by Microsoft back in 2011. It’s available here:
    http://research.microsoft.com/en-us/um/redmond/groups/ivm/ice/

    I use it for my stitching needs and find it works very well. Just a casual user. I don’t have Photoshop, I stay in Lightroom (due to layer phobia).

    Cheers, John

  • Josh, I’ve looked into printing of images at large sizes, and according to the information I’ve found (including a photographer who prints wall sized images) the larger the print you make, the lower the number of ppi you need to print in. Mostly, this is because large images are usually designed to be viewed from a distance (our eyes will see it as a solid image from a distance, even if the dots/pixels are visible up close), not neccessarily up close inspection. Obviously if you want people to be able to inspect your images closely you would need more ppi to retain that level of detail.

  • Pereira Mari

    Hi, i am just wondering. .. how you know after the first pic taken, where to turn your camera for the next pictures… yout camera has some program to helping with each image, your tripod has the scale numbers to localisation or something… im confused. Anyway beautiful pic ????

  • Baldrick

    Canon’s Peripheral Illumination Correction only works for JPEGs.

  • Rick Berk

    Not true. It will work for RAW files when processed with Canon Digital Photo Pro software. It can then be turned on or off in the software as desired. If you use anything else to process the RAW files, then yes, the Peripheral Illumination Correction setting is ignored.

  • Rick Berk

    Knowing where to look is what it’s all about. I look at the scene around me and decide what’s next. I am always composing shots in my head even without a camera. Nothing to do with software.

Some Older Comments

  • Timothy September 28, 2013 10:32 pm

    Josh, I've looked into printing of images at large sizes, and according to the information I've found (including a photographer who prints wall sized images) the larger the print you make, the lower the number of ppi you need to print in. Mostly, this is because large images are usually designed to be viewed from a distance (our eyes will see it as a solid image from a distance, even if the dots/pixels are visible up close), not neccessarily up close inspection. Obviously if you want people to be able to inspect your images closely you would need more ppi to retain that level of detail.

  • John G September 28, 2013 01:33 pm

    For those without Photoshop - provided they're using Windows, there is a good, easy & free stitching program. Image Composite Editor (ICE) appears to have been a research project by Microsoft back in 2011. It's available here:
    http://research.microsoft.com/en-us/um/redmond/groups/ivm/ice/

    I use it for my stitching needs and find it works very well. Just a casual user. I don't have Photoshop, I stay in Lightroom (due to layer phobia).

    Cheers, John

  • Jim September 28, 2013 10:04 am

    Microsoft Research has the Image Composite Editor ( Microsoft ICE). Process your RAW files, save them as lossless jpgs, then drop them into ICE. Totally free, amazingly accurate, and surprisingly fast. I have used it to stitch 100+ images spanning over 180 degrees horizontal and 5+ lavels vertical.

  • josh September 24, 2013 07:58 am

    You're correct, Andy. I should have been using ppi instead of dpi. Thanks for the correction!

    Point still stands, 300 ppi is considered "photo quality." Is it possible to have an image printed with a size of 100 ppi and still be "photo quality"? That's the point of my comment. Do the rules change when printing large format? Perhaps it simply depends on the photo. Perhaps it depends on the intended use/audience. Something to ponder.

    Don't want to take away from the post. Well written, and a great way to gain those extra pixels necessary for large format printing. It's also a great way to give me extra pixels to work with using my 4 year old entry level dSLR without having to spring for a new camera (sensor)!

  • Andy September 23, 2013 07:48 am

    Don't make the mistake of confusing your image resolution with the printing resolution - they are unconnected.
    A printshop will be using 1440 or 1200 dpi to output your photos at normal sizes. Printers used for larger work will still be using 600dpi as a minimum.
    They can all print at 300dpi but there's little point and they probably don't have that resolution profiled anyway unless it is a grand format printer 3m or wider.
    Think of your image resolution as making an image out of blocks - the smaller the building block the more detail you will have.
    Now think of the printer resolution as the amount of dots it will use to print the blocks.
    If the blocks are big or small (low res or hi res image), the number of dots used by the printer to print these blocks does not change.
    If you have large blocks (low res image) the lack of detail becomes quickly apparent on enlargement although the number of dots used to print these blocks may be high.
    If you are using a high res image these small blocks of info can be enlarged some way before they become of a size which becomes unacceptable but again the number of dots used to print these blocks does not change.
    The only time the actual printer resolution becomes an issue is if the ink dots themselves are visible which is only an issue on close inspection - anything more than arms length and dot size tends not to be an issue so most common wide format printers select a compromise between quality and speed and use 600 or 720 dpi for most work.
    Just don't use printer dpi as a guide to how you handle your image file size - your printshop will decide this independently regardless of what you do with your files.
    Your job is to make your image building blocks as small as possible and the only way to do this is by the CCD size in the camera or by the zoom and stitch method outlined above.

  • Josh September 23, 2013 01:46 am

    Nice read! I too am beginning to explore panoramic images for the purpose of being able to print large images.

    The things I have learned (the hard way) to control for each image in a stitched photo are: aperture, shutter speed, ISO, focal length, focus point (manual focus is a must), and white balance. For some of these, you can compensate a bit when shooting in RAW and make a few adjustments in the computer prior to stitching. Make sure you check the exposure throughout the entire view you want to capture. While exposure might look good facing away from the sun, you'll end up with a burned out image the closer you pan to the sun at the other side of your image. Also make sure you include more area in your capture that you need. Depending on stitching parameters and image orientation, cropping will be necessary and may be significant. It is very frustrating to see that an important piece of your composition needs to be cropped out due to an adjacent picture not being perfectly aligned.

    Another thing to keep in mind is final image file size. In RAW on my Rebel XSi (APS-C sensor), a typical image might be 10-15MB each. Combining a number of images and saving as 16-bit tiff gives me a 330MB file. Many computers (mine included) don't like handling images that large and post-processing becomes more challenging.

    I know every image will be somewhat unique, but if the second shot in the post was printed at 8' wide, horizontal resolution would be in the range of 110 dpi. Again, I'm new to this but that just seems too low. While the "intended" viewing range might be significantly further away to take in a whole image, a primary reason for printing large format isn't to let the overhead jetliner see your image. To me, it is to draw the viewer physically closer to the image and see all of the tiny details that wouldn't necessarily be visible if taken as a single image. It's a more immersive experience for the viewer that way.

    I've got a recent image from Glacier National Park's Avalanche Lake that I'm considering trying at 200 dpi but may do a test print or two of a few locations in the image to see how it really turns out at a range of resolutions. The full image size is 11760 x 4896 so 4' wide should be possible. 5' would be closer to 200 dpi and may be the limit for this image.

    -Josh

  • Dan Merkel September 23, 2013 12:40 am

    Good point about shooting in Manual mode to keep things consistent throughout the series of shots. But don't forget to set your White Balance as well; oftentimes, people use Auto WB and this too can change from shot to shot.

    For a very large print as described, I'm not sure that 300dpi is necessary. A print that is 4-5 feet wide won't be examined at close range like a 4x6 inch or 8x10 inch print will.

    For really wide prints, check with commercial printers in your area. Our local Allegra shop http://www.allegrafindlay.com/ can print pictures up to 54" tall and as long as the roll of paper it is being printed on. This could easily be 15-20 feet or more in width. This comes with excellent results, I might add.

    dlm

  • Tom September 22, 2013 11:52 pm

    For hugin, an open source panoramic software, did a pretty good job. Very versatile and cheap (it's hard getting cheaper than nothing)

  • Aju Jayapalan September 22, 2013 02:26 pm

    This is awesome....I am still learning how to shoot and stitch panoramas. This one of the Chicago skyline is the largest one I have ever made...still needs some touch-up though. Had taken 42 images in portrait mode (two rows). 42 because I was paranoid shooting hand-held and did more than 50% overlap (also I am very inexperienced). The bad part of taking so many images was when Photoshop was trying to stitch them, it took about half an hour on my laptop.

    [eimg link='http://www.flickr.com/photos/ajujayapal/9867060515/' title='Chicago_Downtown_Small_72' url='http://farm8.staticflickr.com/7421/9867060515_312d947272.jpg']

    The dimensions of the uncompressed final image was 29322 x 11120 px (300 dpi) and close to 2 GB in size! I have never done this before and am very excited!

  • Mark Zelazoski September 22, 2013 12:30 pm

    Very nice article and superb pics

  • Rob September 22, 2013 09:56 am

    Nice article, I've never thought of doing pano's with night shots as I also thought there would be too much difference in the sky (or water in this case). Great tips.

Join Our Email Newsletter

Thanks for subscribing!


DPS offers a free weekly newsletter with: 
1. new photography tutorials and tips
2. latest photography assignments
3. photo competitions and prizes

Enter your email below to subscribe.
Email:
 
 
Get DAILY free tips, news and reviews via our RSS feed