How does one photograph a large painting in order to make prints?

Question

I have a large painting (30x40 in) that I need to photograph in order to make prints. I have a Canon Rebel T5 camera.

The list of requirements I was given is as follows:

• Dimensions: 6500 x 6500, up to 16000 x 16000
• PPI: 300ppi (150ppi minimum)
• Colorspace: RGB
• File Format: JPG or PNG
• File Size Limit: 150mb

I don't think this can be done in one shot.

I believe I might have to photograph the artwork in sections and then somehow stitch it together in GIMP or Photoshop or whatever software would do the job.

In order to photograph a painting in sections:

Do I move the camera?

Or do I leave the camera alone and move the painting around?

All and any tips would be much appreciated.

Answer 1

As often, the requirements you've been given are internally inconsistent.

• 150 pixels per inch for a 30″×40″ photo would be 4500×6000 pixels, smaller than their stated dimensions.
• 300 PPI would be 9000×12000 pixels, smaller than their largest dimension.
• You can't take a square picture (6500×6500 or 16000×16000 i.e. 1:1 aspect ratio) of a 3:4 aspect ratio painting. Well, you can but you'll have dead space on the sides which is a waste of everybody's time.
• Nobody who has a clue what they're talking about would ever specify PNG for photographic works.

As your Rebel T5 has a 5184×3456 pixel sensor, you can get a photo at a bit over 150 PPI by taking two shots side-by-side, making a 5184×6912 pixel image, larger than the 4500×6000 requirement. You'd probably want to mount either the camera or the painting on a rack so you can get precise alignment and make stitching the two images together easier (and you'd need the right lens to do that).

But really, you should be going back to whoever gave you these requirements and telling them to clarify.

Answer 2

I have had some experience with photographing paintings to make prints and I can say that, unless you have good skill with photography, photo editing, color-correction and lighting you are in for a long, expensive, frustrating and, most likely disappointing experience.

There are some obvious issues, such as producing an excellent single digital file in the requisite size but there are the unmentioned issues of color fidelity, color balance, lack of distortion, proper lighting color temperature and illumination of the print, etc.

Trying to learn while doing and then having expensive prints is, in my opinion and in my experience, a recipe for disaster.

There are numerous printers who will also photograph art work and they are easily found with a web search. If the painting and the prints are important to you, find a place that will scan the painting and make the prints.

The results will be much better.

Answer 3

I believe I might have to photograph the artwork in sections and then somehow stitch it together in GIMP or Photoshop or whatever software would do the job.

Don't waste your time doing that by hand in a generic image editor. Use a tool designed for the job, such as Hugin.

Hugin is most commonly used for semi-automatically stitching together large panoramic images from multiple shots taken from the same spot, but it can also handle stitching together a flat painting or mural from pictures taken from several different camera positions.

There's a very nice tutorial on the Hugin website titled Stitching murals using mosaic mode by Terry Duell which describes how to do that. Rather than try to replicate the tutorial here, I'll just briefly summarize the basic workflow and highlight the main ways in which the process differs from ordinary panorama stitching (for which there are also plenty of introductory tutorials):

1. Take a bunch of photos that cover the whole painting. You'll need a lot of overlap between the photos for Hugin to align and stitch them together properly, so it's better to take too many than too few. A good rule of thumb is that each photo should overlap the previous one by at least 50%.

   (Also try to keep your camera settings like zoom, aperture, exposure, white balance, etc. as constant as possible. Most cameras or phone camera apps have a manual or "pro" mode that allows you to set all these to fixed values. While Hugin can correct for things like variable white balance and exposure, it's easier if you don't have to. Also try to use neutral diffuse lighting that doesn't reflect any glare off the surface you're trying to photograph and make sure not to accidentally cast any shadows on the painting.)

2. Import the photos into Hugin and let it generate a bunch of control points for aligning them.

   (If your photos contain any incidental background or foreground objects that aren't part of the painting / mural you want to stitch together, try to find and delete any control points on those since they may confuse the optimizer.)

3. Once you have some control points, run the optimizer with the "Positions and Translation (y,p,r,x,y,z)" mode. This tells Hugin that your photos aren't all taken from the same spot.

   (You may also need to optimize the lens parameters, especially if your photos are missing metadata about their field of view or if it's inaccurate. Don't be afraid of trying different optimization modes, just check the results for sanity and remember that the more parameters you optimize at once, the more image overlap and control points you need for good results.)

4. For the preview and final image, select rectilinear projection (that's the projection used in normal non-fisheye / non-panorama photos) and use the mosaic drag mode in the preview mode to drag the painting where you want it in the final image.

   (If the preview looks bad and the photos clearly aren't aligned properly, go back and try to remove any bad control points and/or manually add good ones. Once you have a rough alignment, you can also try different optimizer settings to improve it. For example, if your images are always badly aligned at the corners, that may be a sign that your lens has some barrel / pincushion distortion that you'll need to let the optimizer fix.)

5. If your photos don't all have the same exposure or white balance (which will be very obvious in the preview), optimize those in the "Exposure" tab. Once your stitched picture looks good in the preview, you're ready to export it.

Answer 4

Studio Art major here who was required to do this exact task several times for my portfolio in order to graduate.

Basically everyone here is correct and you should just pay to have it done by a professional. You have been given the wrong information and also lack the proper camera, lens, software, and lighting to take the picture.

• You did not even list the lens you intend to use and that is a crucial part as its focal length and chosen aperture are going to determine how much of the painting are in crisp focus. You're going to want to use a field lens and not a compound lens. It will need to be fast although your aperture will likely be between 8-11. The most sharp, focused aperture of a lens is its third aperture ascending from its largest aperture/ smallest number.

• A professional would use a medium format camera to shoot this or full frame at least. Color correction will be the hardest part with biggest learning curve plus you likely don't have a Adobe RGB monitor or probably one that doesn't even produce 100% of the sRGB profile so there's no point in going thru that. Unless you want to buy a $3000 monitor just have professional.

• If the painting has texture, as in an impasto impressionistic style, then the brushwork will create peaks and valleys in the paint's surface. You will need studio lighting with diffused lighting from all sides of the painting and front so that you do not cast shadows from the paint all over the image. You'll have to bounce the light off other surfaces - don't point the light source at the painting.

• Also the TIFF file format is non negotiable. A PDF would even do if created from Photoshop.

• It the painting has lots of white or white details then you're in way over your head again.

• Also the prints you buy of famous paintings are not made using photographs or scans. They are entirely recreated as intaglio prints by printmakers who are so incredibly skilled at printmaking they can make a acid etched Bavarian limestone or copper plate color separation of a painting look like it is a painting and not a print. Printmaking is a dying art form and the majority of people who choose it as a concentration end up getting jobs doing this.

• Also 300dpi is the minimum. You want to have a higher version that you can always reprocess if you need lower. Same with bit depth. So 600 dpi 16 bit color adobe rgb or P3 color profile tiff or psd no compression or pdf with a High quality print profile setting.

• Printed on acid free matte or satin finish cotton paper inkjet only using archival quality ink. They will shorten all that to just calling it a giclee print. If the printer doesn't know what a giclee print is, use someone else.

• A laser print is not color-safe for archival. Unless they use a offset or web printer, which I doubt, they'll need to use Epson. I know people will think Canon, but for high quality High dpi prints, Epson invented and owns the patent on piezo electric print heads and so only they can produce and control the finest sprays of ink from the printer heads.

Answer 5

Call up your local copy shop and see if they have a large format scanner. Found a random old blog post (not mine) explaining the process:

Both Kinkos locations in Anchorage have high quality rolltop scanners that you can use yourself. These scanners will accept originals up to 36" wide and basically an unlimited length. They will scan at 200, 300, or 600 dpi and output to TIFF, JPEG, or PDF formats. I scanned an original 15" x 22" watercolor on 300# Arches paper, which is essentially like stiff cardboard. It took me about 3 minutes to read, understand, and execute the self-guided instructions. The clerk then took another 2 minutes to write my image (I selected a JPEG at 300 dpi) to the pocket USB storage device I provided.

The blog post also lists prices as 4$ per square foot, but I'm sure that's both out-of-date and location dependent. Anyway, it should still be a lot less than 300$ you cite for having someone else scan it.

If you digitize with a camera you need to worry about things like how perpendicular the camera is, how even the lighting is, etc. Scanner has its own built-in environment which just does all of that for you.

Answer 6

At 150DPI (*), the 3:2 format that contains all your picture is 30Mpix (30²×1.5×150²=30375000), so within what modern DSLR/Mirrorless cameras can do.

So you can spare yourself a lot of grief and post-production work by renting a camera and taking a single shot. And while you rent, you can also rent a moderately long prime lens (100mm or more) (but since you are photographing a flat object there is no need for something very long).

And if you think it's expensive, you haven't considered yet all the extra gear you need: stand for the painting, tripod for the camera, several lamps/flashes and their stands, because getting a uniform lighting than won't create nasty reflections and keep some of the texture is probably the hardest part of the whole process.

(*) which is likely sufficient, see this

Answer 7

You have not defined your lens focal length. The longest the better, the better quality the better.

Take a look at this setup: What focal length is best for photography of rugs?

To squeeze more Megapixels on the final shot, yes, you can take different ones and stitch them together. You preferably will need a panoramic head.

You can not get an exact pixel dimension by stitching files. The reason is the "analogic" part. If your camera points 2 mm to the left or right, you will get more pixels to the left or the right.

So I would aim to get a much, much bigger file and resample after.

To stitch it, use small angular adjustments, so you have a good overlap, and use the center part of each image. Even on the photos of the edges, try to leave a good extra space, again to use the center part.

Keep in mind the inconsistencies Philip Kendall is pointing in his answer.

---

Sidenote.

Probably 50mm is too short. You will be too close to the painting so you do not have parallel lines and you will cover the painting in just a couple of shots. It can be corrected tho.

Hugin is useful to stitch and straighten the images.

See if you can get a 200mm or 300mm lens. See if you can rent one.

Answer 8

What is missing in your post is the required quality you need. You will have :

• light issues : what color temperature you want and how you achieve it (2 normalized projectors would be needed)
• colour fidelity issues : if you don't know what this is about, don't even try. You need to calibrate your camera with a custom profile, and this is not a job for an newbie. Also the printing fidelity is an issue in itself and given your requirements, it's not obvious the ordering party knows anything about it
• focus issues : the focal plane of your camera is likely not a plane and you'll end with some parts slightly out of focus
• image distorsions in the corners (color and shape)

On top of that, depending on the country you live, you may face copyrights issues if you don't have the painter authorisation to release public prints, associated with heavy penalties.

Now, you can photograph any painting with any camera and print it without the Earth falling apart. Defining acceptable quality is pretty tricky, and you will not find the recipe without discussing it precisely with an expert. If you only want to do a favor for a friend, go on. If you want to produce professional quality with guaranteed fidelity, you 're in for big trouble.

Answer 9

If you know someone with a good medium format film scanner the minimums should be achievable with one 6x6 or larger exposure. So the requirements might be old standards originally written towards medium or large-format film photography.

Answer 10

Just a short answer with an alternative suggestion. I had great success with Image Composite Editor from Microsoft. It's completely free, although you need Windows. It tries to auto detect how did you move the camera. It supports 10bit PNG files.

My workflow is generally:

1. Camera in manual, you need the same exposure on all photos
   • Long lenses are better, less distortion on the edges but depending on light conditions, this might not be available option
2. Take test photos to ensure the light is correct
3. Shoot all parts in RAW with good overlap
4. If possible, include a clear white area in the photos for fixing stuff later
5. Postprocess colors in Darktable, copy the settings from the corrected photo to all others
6. Export, optionally in 10bit if you're worried the colors will still not be great
7. Stitch, crop in ICE mentioned above
8. If needed, additional postprocess in GIMP which also supports 10bits, so you still have 2 bits wiggle room for under/over exposed pixels
9. Export as 8 bit for screens. I never did this for print, so not sure if 10bit is any benefit there.