You Say You Want a Resolution: How Much DPI/PPI is Too Much? (2023)

Great blog post, so good to hear back from the experts.

But this all underscores for me that digitizing is not always the best means of preserving an image – a good print can have a longevity that is difficult to match digitally because of changing technology and the frailty of digital media. I advise my photography students to print good quality prints and store them properly as another layer of ‘backup’. I have prints that have withstood 150 years that are remarkably well preserved, and digital images lost to the ages locked away in a zip drive I am no longer able to access. So the dpi/ppi doesn’t matter if you can’t read the file. High quality silver halide prints are important, and digital files, while allowing uses of the image not possible with a print, are no substitute.

Great to read about this and hear from Barry Wheeler on the interesting and little-known matter of “sampling efficiency.” While we are at it, we ought to reminder readers of Barry’s earlier postings on this general topic — here’s the link to part 3 of a three part blog: http://1.usa.gov/11UemG7. (Find parts one and two on your own!) I will also suggest taking a look at a pair of blogs I wrote even further back on determining the inherent level of picture information in photo negatives: http://1.usa.gov/15j9Pu3 and http://1.usa.gov/18jngww. But my blogs are not as clear as Barry’s regarding sampling efficiency.

Mr. Riecks and Wheeler –

I completely understand what you say about not being able to improve upon the quality of an image just by scanning it at a higher resolution (like capturing a VHS tape to DVD doesn’t improve the quality either).

However, as Mr. Middleton noted, the quality of printing, computer monitors and televisions will continue to improve. Given that, would not a much higher res scan than 300 dpi, with higher pixel dimensions, project better on really large viewing areas (like, for example, a display wall or an Ultra HD TV with an 8MP monitor)?

Just wondering, thanks – Ken Hopson

My question is: what strategy should be used when scanning old family photographs (on paper) assuming the image will be viewed on a next-gen TV.

I believe we are in the midst of an evolution in the way photos are viewed – and that this should be a factor in how we digitize our existing material.

There have always been two ways to look at a photo: as a print, or a projection. Right now, very few photos are viewed as prints, yet this conversation seems limited to discussions about the effect of dpi on subsequent same-size prints (or enlargements) with reference to scanner and printer technologies. I think we should be scanning for the way images will be viewed in the future.

And how is that?

Right now, photos are hardly seen at all. They get snapped by the dozen, glanced at via the camera’s 3 inch screen, and deleted when the memory card is full. Some are emailed – and then squinted at on the other guy’s phone, or seen on a tablet. A few are appreciated once or twice on a computer monitor – but not exceeding 20 inches – and then lost. Today, the average person’s photos are stored in so many places and are so poorly labeled they have become ephemera – enjoyed for the moment and then forgotten.

I think this is about to change. People are starting to ask – why am I taking all these photos if I’m never going to look at them again? When Grandma dies and nobody can put their hands on a decent recent photo, much less find Uncle Jimmy’s baby pics, people will re-think their habits.

I predict we are headed right back to 1865. Or 1965. Remember lantern slides and Kodachrome?

The distance between a tiff (or RAW) and a high-quality, hand-held print is so arduous a trip I don’t believe prints will recover. Their function – images to be easily seen, cheaply duplicated, and readily shared – has been usurped by digital media. The newer technology is so much better, I don’t see prints coming back.

What I see is a resurgence of the old-fashioned slide show. People will select their best images, put them in a certain order, burn them to a DVD, and view them on the TV in the living room. I say “old fashioned” because I don’t mean the kind of production with a synchronized music track, frozen voice-overs, and multiple KenBurns effects. People won’t make that effort, as easy as it can be. But, when they start losing precious keepsakes to digital clutter and failed hard drives, they will begin making the effort to get all the best of last summer’s vacation pics in story-order, in one safe place, which can be easily viewed to maximum advantage – a DVD, plopped in a player, watched on a 42” (or 72”) 1080 LED TV, in front of a comfortable sofa – just one image after another with a simple dissolve, advanced by remote control, accompanied by a spontaneous narration (probably by the photographer) geared to the time and the specific audience present.

Compared to the days when Dad had to darken the room, find the projector, check for a burned-out bulb, unfold and stretch-out the 3 foot 3 legged screen, and hope the slide would drop thru rather than jam up the tray – today’s slide show technology is a dream.

So, in deciding what resolution to use in scanning old family photographs (on paper) my question is: what strategy will be best assuming the image will be viewed on a next-gen TV.

John Black: I do not think that you should consider the display size when scanning, assuming you scan for preservation. Whatever your guess for a display size is, it it likely to change over the years. Go for the largest size that makes sense information-wise from the original material and the scanner (just as the article suggests for paper originals).

Unfortunately, the article is right about consumer level scanners being inadequate for scanning negatives. I checked my own cheap scanner (Epson V500) and even though it is advertised as 9600DPI, it really only resolves around 1200DPI, just as the article suggests. I have a bunch of negatives that I would like to scan for the first & final time and I am considering renting equipment that does them justice.

As an archivist, I think the question of digital resolution is somewhat misplaced, a bit like arguing about the best manner for holding a handful of sand. Digital imaging is not a means of preservation (in the archival sense) but rather a convenient means of simplifying distribution and advertising. I sincerely hope we have learned something from the Great Microfilmed Newspaper Debacle of bygone years–where newspapers were chopped up for photographing and then simply discarded: reformatting at any resolution is never an adequate substitution for the original object. Digital technology offers some great benefits, but for routine use electronic forms are expensive conveniences at any resolution. No digital file is ever nearly as “permanent” in the way a physical original is.

Am I arguing for no longer digitising? No, just ensuring that we don’t let the inherently fugitive format become the practical standard of preservation. An individual or institution would be well advised to pour their limited funds first into preserving original material, and only secondarily into digitisation–with the explicit understanding that maintaining and migrating digital forms is far more expensive in the long run than costs involved in preserving the negatives or prints.

Beyond that, maintaining a general-use digital image at 300/400dpi is an adequate practical step. If a higher resolution is needed, rescan the object and let whomever requests it foot the bill.

The LC’s Digital Scan Service scan specs are here: http://1.usa.gov/18qlAEQ

Richard Saunders: What is the state of negatives after 100 years? After 300? 500? Your time perspective seems limited to the near future. I would argue that digital is exactly the definition of permanent as it is unchanging (assuming using best practise of having multiple copies with checksums, migrating losslessly and all that). Opposed to that are the originals, be they negatives or paper, that are doomed to a – hopefully – slow deterioration.

At the museum where I work, we use images in two ways. The first is reproducing the image as artifact. In this case, we do little to no retouching or recolorizing, and show images at a 1:1 scale, or slightly larger if it will benefit the visitor experience. The second way is to apply some treatment, scaling, etc., to provide a different experience. This may be applying a color to images, making them a monotone image in the background, or making the image very large for impact. In the case of making images very large (e.g. 8-10 feet tall), I have found that we are often able to scan-in original images at sufficient resolution to scale them. If the image has already been scanned in at 300 dpi and the original is not available, the image is rarely big enough to print with good quality. Although this may seem obvious, it is important to note that depending on your final image use, you must allot potentially a significant amount in your budget and timeline to process images to render them usable at whatever size you are printing them at.

This analysis assumes the end result is a print roughly the same size as the original for reproduction or display, though scaling for large prints is mentioned in a comment. It fails to address the benefit of higher resolution scanning for analysis of the photographs or for use in video, both of which require higher resolution and lower resolution scans limit both potential uses.

Again the intent and possible future use are deciding factors, but given scanner capabilities and low storage costs, why not recommend higher resolution scans, particularly for content such as interior and exterior views that contain clues that can be uncovered by enlarging the image?

There is no perfect answer. However there are informed choices to be made.
Factor #1: is the resolvable or perceivable information content in the original and that varies with media and within media. Newspaper halftone vs Kodachrome 5×7 transparencies have very different levels of information.

Factor #2: The true or actual sampling efficiency of the capture equipment. 300, 400, 600 ppi are just numbers that you set your equipment to in the software they are not indicators of the end result. And no two scanners scan the same.

Factor #3: Total lack of a standard viewing interface. And most prints or display devices can not render the fine information content found in the original.

Factor 4#: Scanning approaches that aim to maximize quality and anticipate future outcomes have not been proven to be an efficient means to “preserve” anything. The historical record favors chance survival of data regardless of the substrate the data was contained within. Ubiquity and Utility greatly increase the chance of “preserving” information. Also there must be some informatic maxim that states: Expect Loss

Factor #5 Are we getting more or less visually literate to demand greater quality? History of Photography more or less has been to sacrifice quality for the sake of ease of use. I tried Wet Plate it was fun, but…

So, the 300/400/600 ppi numbers serve as an informed comprimise that factors in sample rate/image quality/and utility of the file.

A couple of comments

A common confusion with ppi: it only applies to conversion of a flat object, not to digital photo. The statement “human vision is only capable of resolving 326 ppi” is misleading, because the resolution clearly depends on a distance. For example we can see with bare eyes the spider web: 0.00012 in = 8333 ppi minimum.

In general the scan resolution should be such, that one can distinguish the smallest details visible on the picture, whatever they are. It may mean 300 or 600 or more depending on image. Good magnifying glass may help..

Future-proof means that we do know, what a future researcher will look for. Do we? If you scan a smallest newspaper text (say ruby) at 300 dpi, you will get a barely readable text, but difficult to determine the details. A researcher was looking for tiny holes made in the picture by insects – hi resolution scan allowed to see the shape of the hole. This is especially important if the image is of archival value, and it is expected, that the original will not be accessible (or accessible with difficulty)

There is little point in scanning for information that isn’t there. The resolution of most 35 mm camera lenses is about 2000 x 3000 pixels. You can increase this to 3000 x 4500 pixels if:

* Using very good lenses
* Using very CLEAN lenses
* The image lighting has minimal light on the lens to produce internal reflections.
* You are using a very fine grain film.

One way you can test the effect of increased scanning resolution:

1. Make a series of scans at various resolutions.
2. Now, using a package such as ImageMagick or NetPBM, do some processing:

a. Try expanding the 300 dpi image to 600 dpi using appropriate intermpolaton (usually bi-cubic)

b. Now, subtract one image from the other. You want to get the absolute value of the differences.

c. The resulting image will be VERY low contrast. Normalize the contrast.

d. Look at the image. If there is additional information in the 600 dpi image, you will see ghostly, shadowy detail in the the difference image. If, OTOH, all you see is speckles, then the additional scanning density was just meauring noise.

Working with prints is even worse: A print can only resolve a luminance range of about 100:1 — About 7 stops. Many films can do 1000:1 — 3 stops more. Good printers take this into account. Plus the enlarger lens multiplies the original lens’s limitations by it’s own.

From 35 mm or smaller film I would be surprised if 600 dpi reveals any information not present at 300 dpi even for high quality 8×10 prints.

If the original was shot in a larger format, however, it may be worth doing.

In either case, going back to the negative has merit. And all negatives should be scanned at 2-3 thousand dpi.

Sorry, I’m late to this party, but thought I throw in some thoughts, incase anyone has new ideas…

I am working on a one year history project for a small community in Maine. I have to scan hundreds of postcards, photos, documents and maps, some going back to the 1700s!

1. PRESERVATION: I want my stuff to be safe. (LOCKSS – Lots of Copies Keeps Stuff Safe). I return original material to owners, or store in our library. I use Cloud Backup(DropBox), off-site hardrive, and local harddrive(flash).

I remember paper prints and DVDs. Both were useful in their day. I find it amazing that anyone worries about printing an 8×10. The VAST majority of photos will be looked at and zoomed on digital devices. DVDs are quickly becoming obsolete. (Why, Apple hasn’t used them in years 🙂

2. PRESENTATION: Every person under 20 in America (and most the rest us) looks at photos on an iPhone, iPad, HD TV or other digital device – and the results are phenomenal! My 83 year old mother and my mother in law (in South Africa) LOVE their iPad PhotoAlbums.
ERGO: I want a backlit HD (‘retina’) digital display.

3. DETAILS: Most people know to pinch to zoom. And they love to zoom. They HAVE to zoom because smartphones are so small, but they make it so easy to zoom.
ERGO: I need the BEST RESOLUTION possible, because I’m going to look very closely at my mother’s face in the 1942 group photo.

And I agree fully that some photos need 2400dpi, and others (e.g. half-tone postcards) max out at 200dpi.

TIME: I want to SCAN as quickly as possible, and I want displays that will DISPLAY large files as quickly as possible. (e.g. A high quality image of a MAP can be over 100mb) I used to scan all photos to end up with about a 1meg file, because that was all most PCs and Macs could deal with in a fast slide show.

SPACE: Getting cheaper all the time – almost not an issue.

Help?: Anyone know of a great, FAST flat-bed or other scanner? (For maps I guess I still have to use a good copy-stand.

Cheers!
Phil

What is Optical DPI in mono laser printers? How it matter ?

NOTICE: Most of you, including the “experts” are just dead WRONG!!!

Scan a photo, any photo that includes
people or objects in the distance/background

Scan this photo at 300dpi TIFF (even JPG)

Then

Scan again at 1200dpi or even 600dpi

Now ZOOM in…you will see MANY details that you will NOT EVER see in the distance with the 300dpi scan

I will stick with 1200dpi or 600dpi TIFF for MY family photo archive master file.

I will never have to go back and find the original photo And scan again to enlarge for printing etc etc

And I will be able to read that license plate or view what watch or ring an ancestor is wearing etc etc

By the way I know all this to be a FACT from hard fought experience!!!

For family photo scanning/archiving go with 1200dpi, 800dpi or 600dpi TIFF format.

You’ll be glad you did later….And so will your beloved family members.

“EXPERT”

“An EX is a has been, and PERT is a drip under pressure”

Randy,

Thank you for writing. It is clear that I could have conveyed some of the information better.

I wrote this article as a response to all the various and contradictory recommendations found in user manuals, from scanner to scanner. Consumers and amateur photographers lack clear, simplified advice. I consulted digital-photography professionals who work in a highly specialized world of Test Charts, targets and multiple thousands of DPI, and often I felt that their advice was a bit too detailed and instance-oriented for a broad audience. For this article, I picked and chose through their advice for what to say and what not to say.

In the end, the correct answer to the question “How Much DPI/PPI is Too Much?” is “It depends.”

Here are responses from my colleagues. In particular please read Dan Margulis’s article, linked below.

Mike
—————

Carl Fleischhauer

For routine snapshots, 300 ppi is not bad but when there is valuable detail, clear in the original, higher resolution is wise.

It all depends on the detail in the original print. It is certainly the case that some home/snapshot types of photo prints will benefit from higher-than-300 ppi levels of resolution. Professionally made photographs are more likely to contain more detail. In many cases there will be a benefit to making higher resolution copies. Scanning the original negatives would be even better. Here at LC, we scan 35mm negatives at 3600 pixels on the long side, roughly 3000 ppi.

We had a nice back and forth in 2013 about some Lincoln-related photos that required extreme zoom ins to analyze, followed by a debate over what was actually there. (One good description here, in the Smithsonian magazine, which also references a photo at the National Archives: http://www.smithsonianmag.com/history/Interactive_Seeking_Abraham_Lincoln_at_the_Gettysburg_Address-180947919/.) I think there were some other Civil War period photos where specialists demanded extra high resolution. Of course we keep all originals so we can re-scan if needed.

The trade-off is that if you scan large numbers of pictures at very high resolution, you have a lot more data to manage over time. But there could be times when it is worth doing.

———–
David Riecks

If you scan everything at super high resolution — especially from a print — most of the time it’s going to be a waste. Most of the time, seeing that closely — especially on small prints — simply magnifies defects and increases the amount of dust-spotting you’ll have to do. That spotting work applies for both prints black and white negatives — though with color slides you may have the option to use a “digital ICE” algorithm on some scanners to automatically remove dust (though often with a slight reduction in sharpness).

The same applies to copy work using analog / film methods. Anyone who has made a copy negative of a small print and then enlarged it to 8 or 10 times the size of the original knows that the results are usually disappointing (like shooting a 4 x 6 print and then making a 16 x 20 print from it). With digital, you have a few additional tricks up your sleeve but they result in a lot more time and effort to get even mediocre results.

My usual recommendation is to start with a reasonable sampling rate and use that for the whole collection and don’t throw out your source material. If there are “super special” images then go back and rescan those at a higher sampling rate when you have time or need to for a particular project.

Scanning images at the Smithsonian is very different that for a home owner/ hobbyist who is scanning their mother’s collection of photos taken with a Brownie box camera.

Another factor has to do entirely with the quality of the originals. As Carl mentioned, you should use a higher sampling rate for negatives, as opposed to prints as they contain more: more dynamic range, more detail, more sharpness. Any print will never be as sharp, as it’s a generation away from the original and cannot contain as wide a range of tones, since they are not transparent. The quality of the optics throughout also makes a difference, as well as how the photograph was made. Even a high end Zeiss lens can yield a blurry/fuzzy photo if shot hand-held (rather than on a tripod) and/or if the aperture is wide open (because of little depth of field). A Diana (plastic camera) is never going to give you a sharp image, and many “box cameras” aren’t much better. Scanning a 3 x 5 print from one of those at 600 ppi, or higher isn’t going to yield much more usable data, just more data.

Which reminded me of this chapter in a book by Dan Margulis on “Resolution.” https://www.ledet.com/margulis/Makeready/MA25-Resolution.pdf See illustration 15.7 on page 306 for an illustration. The caption reads: Figure 15.7 Does resolution equal detail? The top version seems soft, even though it was scanned at three times the resolution of the bottom version, and takes up nine times as much disk space. In areas of one color, like the grass, the higher the resolution, the more even the color will become. The text in the article goes into much more detail.

Here’s the nub:
But with a photograph, too much resolution, in addition to the shortcomings enumerated above, actually does hurt. Which of the two images of grass in Figure 15.7 do you like best? If we want something that looks like blades of grass and not AstroTurf, the bottom version seems clearly better. But it’s the lower-resolution scan! Doesn’t high resolution equate to more detail? Of course it does. But here, we don’t want detail; we want the illusion of detail. That’s what the bottom one provides. Let me try to explain how and why.

And here’s the executive summary:
In scanning, moving to a lower resolution is a move toward action and variability. This is a fine concept, but if the resolution gets too low, the image will become harsh and jagged. A higher resolution is a move toward smoothness and consistency, which are also laudable goals, in moderation. Too much resolution will make the image look soft and de-focused. It follows that there is no one “correct” scanning resolution.

I can’t put it better than that.

David

Hello,

March 28, 2017

Thank you for posting my comments and thank all of you for your responses.

I spent quite a bit of time on your website reading and I had some questions I was hoping someone that would be considered an expert can answer for me.

I’m scanning paper colorphotographs I may want to enlarge later does this mean I have to scan them and 48 bit color or can I get by with 24-bit color?

I’m also scanning paper black-and-white photographs that I may want to enlarge later does this mean I have to scan them at 16 bit black-and-white or can I get by with 8 bit black-and-white?

Just so you know I’m scanning 2000 family photographs lots of black-and-white photographs and color photographs

I’m scanning the photographs as tiff files probably 600 dpi for color and 1200 dpi for The black-and-white photographs

I will be converting the tiff images to JPEG’s and then sending them to relatives

Knowing that I am converting tiff images to JPEG’s and possibly using 48 bit color and 16 bit for black-and-white photographs is there going to be a problem down the line with them being able to view the photographs or software compatibility or whatever?

If I photograph is scanned at 600 dpi 48 bit color, can the DPI be upscaled later to 1200 dpi or whatever or does the photo have to be scanned again? (same question for black-and-whit Photos)

Can a scan of a photograph at 24 bit color can it be upscaled to 48 bit color without re-scanning the photograph?

Can a scan of a black-and-white photograph scanned at 8 bit can it later be upscaled to 16-bit without re-scanning the photograph?

Will a 24 bit scan of a color photograph look good printed at the actual same size? My understanding is that if you enlarge it it should’ve been scanned at 48 bit to print the colors properly…

Do I have to scan black-and-white photographs at 16 bit to get good Full qualityprints and possibly enlargements or is 8-bit good enough?

My Canon canoscan 9000F Mark II Scanner is 24 bit color default and 8 bit black-and-white default… However I can scan color photographs at 48 bit color but black and white photographs can only be scanned in 8 bit….16 bit is reserved for film only.

I’ve spent 10 hours a day for six weeks trying to get a good start on my scanning project and the above questions are very important to me so I can get going here can someone help answer my questions?

Respectfully yours,

Randy

A term I use is what I call Kodak negative quality meaning the quality of the negative of a high quality photo taken with a good quality 35mm camera in relation to dpi and resolution. Today everybody is gaga over over theinstant convenience of digital cameras, What they don’t realize is that the properly taken photo from a quality Kodak negative is still the best photo quality that there is, but over time technology will over take traditional film and negatives one hundred years from now 2017 good photos will be printed and displayed in any size wanted and will come from digital negatives also in the future we will have a digital image preservation archival device unimaginable today.All we have to do is unlock the remaining laws of physics ( particle physics) and do the necessary research, but for today 2017 digital overall can’t beat a real quality Kodak negative from a properly used 35mm camera.

To the poster who apologized for coming “late” to the party- lmao as I’m a year later than y’all- hahaha…. However thought it a great forum and worth chipping in with a, er… “TV angle”. As a professional video editor who often needs to add Zooms and pans etc using motion control to still images I abeee that the highest (2000dpi – 3000 ppi etc) possible scan rates is absolutely crucial! It goes beyond the limited scope of the “experts” who wrote in the above article replying to the original query. Definitely worth capturing at as high a dpi ppi as possible for HD broadcast and editing motion control on the original image which is done when there are many stills (ie a civil war documentary or one in which the producers are cheap or simply have NO budget so they can only afford stills instead of video) in order to make it more visually stimulating for our attention deficit disordered viewers. Great forum and topic.

I read this just for the master pun in the title. The Beatles forever!

I am somewhat confused at the number of so called experts who confuse pip’s and dpi’s. According to the Epson book on printing (I assume as they are a leading printer manufacturer they know a little better than most), ppi’s and dpi’s are like comparing apples and elephants. ppi’s are reasonably described, it is the number of pixels per inch. However, according to epson the dpi’s changes with the number of inks available in a specific printer. As an example at 100 pip’s with a printer with three inks the dpi is 300 whereas with a printer with 9 inks the dpi’s are 900. So the resolution of the print is set by the pip’s but the range of the color gamut, is much wider with a printer with 9 inks when compared with a printer using three inks.

I do not agree that high resolution scanning is not worth the time and hard drive space. You should scan a regular family picture 4×6 at 1200dpi minimum to have acceptable quality details … especially for faces which are much smaller that the whole picture and the faces are the most looked portions of family pictures and people want to zoom in (once you zoom in a face if you scan anything less than 1200dpi you will remain very disappointed of what you will see on your screen). Measure the faces size on your picture first.. for example if the face is 1/4 of an inch then scanning at 1200dpi will give you 300 pixels to display on your screen for that face. At 300 pixels the face should not look pixelated on regular HD screens even when zoomed in. Anything less would be questionable.

Possibly someone can clear up some confusion I am experiencing regarding DPI/PPI. The efforts to protect photos and artwork online has always been a challenge beginning with the minimally effective watermark technology. However, understanding that DPI focuses on PRINT quality while PPI focuses on VIEWING quality, I’m wondering if posting a photo/artwork with very low DPI would somehow thwart the effort to copy/print the photos or artwork. Does this idea go anywhere? Thanks. Joel

You guys really help me out . I really need this information.

Many many moot points above.

As another user mentioned, digital technology is constantly improving. It could indeed get far beyond the point of portrayal of what’s even noticeable to the naked eye/ear etc, and people would decree that it’s so amazing and every man and his dog would want to own the technology.

Keeping things less detailed on purpose is unfathomably stupid.

As for the eyesight’s limitations, do you guys also argue that high res phones and Ultra Hi-Res screens shouldn’t exist? Get real.