Showing posts with label digitizing. Show all posts
Showing posts with label digitizing. Show all posts

Wednesday, April 25, 2012

What is the best way to digitize old photographs for preservation?

Question

What is the best way to digitize old photographs? I scanned some of them and there was a reduction in quality due to the scanning process. I also have some photos where scanning is impossible - the photos are struck firmly in a large album; tearing out them may cause irreparable damage.

Is it better to take a snapshot of an old photo using a digital camera than scanning? Are there any other ideas for best possible digitization and restoration of vintage photographs?

PS: I saw this related question. Additions are welcome.

Asked by Shyam

Answer

A modern scanner of typical resolution should be able to do a reasonably good job on old photos if well used. Method of use can make a difference. See references at end for some tips and guidance.

A 6" x 4" print at 300 dpi corresponds to 1800 x 1200 pixels ~=.... 2.5 megapixels
A 6" x 4" print at 600 dpi corresponds to 3600 x 2400 pixels ~=..... 9 megapixels
A 6" x 4" print at 1200 dpi corresponds to 7200 x 4800 pixels ~=.. 35 megapixels
A 6" x 4" print at 2400 dpi corresponds to 14400 x 9600 pixels ~=. 140 megapixels

File size grows in approximate proportion to megapixels for the same bits per pixel, so fil size increases with about the square of the dpi resolution. 300 to 600 tends to be bearable. 300 to 1200 leads to 'rather large' files.

As will be seen in the references below (especially the first one) a good 600 dpi scanner SHOULD be able to do all you need and want when scanning photos. The implications for using a camera are discussed below.

If original resolution was no object (and in this case it is)
300 dpi is OK to good for colour. Better would be 'nice'.
600 dpi is very good.
1200 dpi is beyond excellent in practice.

eg at 600 dpi a 24 megapixel print (most top full frame DSLRs and some APSCs (A77) are around that resolution) would print at about 10" x 8" or ~ A4..

Scanner resolution differs across and along the scan as one depends on available sensor elements and the other on both sensor pitch and movement and how it is driven. Most modern scanners will happily exceed 600 dpi on both axis. Other factors are liable to make the difference.

This excellent discussion on resolution suggests scanning does not give much gains above about -

  • B&W prints 400, just maybe 400 dpi
  • Colour prints - 300 dpi
  • Film - 3000 dpi

The same article notes that an HP printing expert / HP printing website say:

  • "Since printers are at or near photo quality, scanning resolution is no longer a printer-dependent value. To get photo-quality scans---even if going to a 2540 dpi image setter---200 dpi is the max resolution you need if your original is a normal silver halide color print."

The HP scanner web site says:

  • "A Note on Resolution: The vast majority of scanning projects require resolutions lower than 300 dpi. For example, scanning a photograph at resolutions higher than 150 to 200 dpi only produces a larger file, not more detail."

Using a camera:

A camera has the advantages of quick to very quick throughput compared to a scanner, and of accommodating non-flat documents well. Whether it suits in other ways is up to the user. I have achieved results which are acceptable to me for my purposes. Others may be more discerning :-).

As can be seen from the above discussion and various references, 300 dpi is liable to be very adequate in many cases and 600 dpi will essentially always be overkill, all else being equal. This implies that for a 6" x 4" print a camera of 300 megapixles will provide over 300 dpi resolution and a 10 mp camera at 600 dpi shold be very very adequate provided all other factors are properly considered. That last sentence is the area to look at.

About 4 years ago I needed to photograph a range of photos and old paper records - some from up to 100+ years old - but most rather newer. I had to copy these in a library situation and they could not be removed. Many documents and photos were in books or mounted in some way and in many cases scanning would have been difficult or worse or not at all suitable where there was significant 3D effect.

This was for a church Jubilee. Archival results were not needed - just documents that gave a good feel for the originals and photos which were as loss-free and correct-to-the-original as was sensibly possible. I used a Minolta 7D 6 megapixel camera. The results were entirely acceptable for the purpose.

I was given considerable latitude and could have brought in extra lighting. Some experimenting showed that use of a camera using existing library fluorescent lighting produce photos that were adequate for the purpose.

Lighting needs to be diffuse and even and produce no reflections or bright spots, the existing lighting suited this well.

Colour balance was set by inspection to produce results as close to the originals as possible.

Camera was tripod mounted and focus was adjusted manually and then untouched when photographing flat photos. Readjusted as required for books.

I do not now recall with certainty (somewhat to my surprise) whether I used a piece of non reflecting glass to hold photos and books etc relatively flat. Part of my brain says that I did, and this makes sense, but another part says this is a manufactured memory. I can further check that, but it makes great sense to do so and, once sorted, reflections and similar will not be an issue. Glass will lose about 15% - 20% light level (10% per way) and can add a slight colour cast - green if regular float glass used - which is very slight and can easily be adjusted for with colour balance.

The photo at the very end of this post is one example of the results. This was 40+ years old at the time it was photographed. Taken at f/5.6, 1/2 s, 400 ISO - but I tried a range of apertures and other settings. Resolution was such as to appear as good or apparently better than the originals. Size here was probably around A4.
I chose this photo as an example mainly because the young man marked with the added red dot is a much younger me :-). Right click the image and copy or save for the 3008 x 2000 original. EXIF is destroyed by the stack exchange photo storer.

I am happy with this result. Colour is about original. Detail present is apparently equal to the original. Having dug this out I'm tempted to go back to the library and have a critical look at the original and copy 4 years on.



References:

Scanning 101 - a good guide. Note that they end up with similar advice to other sites - mainly that 600 dpi is more than adequate for photos - but they give a far better discussion and more reasoned and detailed basis for their claims.
Related The 72 di fallacy useful.

Here is an excellent guide to scanning resolution written from the viewpoint of a fine-art print designer. Her specialist application lead to recommendations for line images of up to 1200 dpi - 2 to 4+ times those met anywhere else - but only for line art. Her figures for continuous tone images are much lower and detailed enough regarding input and output devices to be worth copying here.

Luisa Simone - Scanning 101 - setting the right resolution

enter image description here

(The original table resolution was inadequate !!! :-) )



A few scanning tips - useful but do read other references too re DPI. Note their good warning re 2400 dpi scanners which mostly use interleaved sensor cells to get their resolution - and so have interpixel overlap and blurring which needs more sharpening to "correct".

Bad example Pick your scanning resolution

Typical advice. Due care. How to choose scanner resolution

More typical advice. More due care Scanner guide

More similar.


Added:

Useful even though biased scanner comparisons from Epsom


Added:

Effect of photographing a photo:

Technically the new photo can never be "better" than the original and
technically the new photo will always be worse, even if only very slightly so.

However, if you use a good quality camera and set it up well as discussed above with proper attention to lighting, white balance, exposure, focus, aperture (affecting depth of field) etc then you can get photos which approach the original so closely that the differences are not distinguishable for practical purposes
AND in many cases, with an old photo of less than perfect original quality, you can get images which appear to the eye/brain as if they are somewhat better than the original.

"Improvements" may be due to slight sharpening during capture or (about the opposite) spreading of detail across several pixels, or of the lighting and white balance combination changing the relative manner in which some visual aspects are affected. This all tends to fall into the area of subjective or personal effects and the user has to decide whether to emphasise capture effects which change the appearance or to aim as much as possible for best fidelity. For archival purposes fidelity probably wins. Where end user perception matters it may be permissible to allow some effects to occur - but these are better done post-capture where possible so that the maximum number of options are left open.

eg one may try HDR effects to try to capture dynamic range present in the original and outside the range of the camera - although this is not usually an issue with old prints. One could use in-camera HDR where available (eg Sony A77) but this then loses the images that the HDR image was created from. Taking 3 images using exposure bracketing will allow post-capture HDR and leave more options available at a later date.


40 + years old photo.
Minolta 7D, 6 megapixels.
About A4 original.
Fluorescent lighting.
Details and colour are a fair match for original.
Red dot = me.
Right click image for 3008 x 2000 pixel original.

enter image description here

Answered by Russell McMahon

Saturday, December 17, 2011

What are the major differences when digitizing slides/negatives with a DSLR vs. a scanner?

Question

A flickr discussion, pointed in an answer to another question, got me excited in scanning old slides & negatives with a tripod, lightbox, macro lens and a DSLR (or with a similar set).

I have rather minimal experience in scanning with a flatbed scanner and non-existent experience with slide or negative scanner. So: what would be missed from a dedicated scanner when digitizing with a DSLR (if one already has the sufficient gear)? What generally are the major drawbacks? Or will a DSLR excel in some aspects?

Resolution is something that my Canon 450D probably wouldn't match to the 4000+ dpi that scanners offer, but that's probably something I could cope with (or is it?).

I'm aware of the possible alternatives mentioned in older questions, but I don't consider them as interesting right now.


To summarize strengths of a scanner so far:

  • ICE & Multi-Scan
    • IR-based ICE: definitely will be missed. Can't be easily DIY-duplicated — though this just means more time will be spent on the post-processing.
    • Multi-Scan: doable with image stacking
  • Slide holder & even lighting
    • Slide holder: DIY-doable.
    • Even lighting: might be hard to set up, but doable.
  • Time
    • More human intervention is needed.
    • Setup time is longer and scan time per slide is probably shorter with a DSLR, but scanning won't be a background activity.

Is this it? Are there any other technical aspects that are commonplace among scanners, but hard or nearly impossible to MacGyverize (like IR-based ICE) for a DSLR? I would suspect there is, as the scanning techniques are drastically different — then again, I don't have much scanning experience with a scanner. Is there any impact on the image quality? Would a scanner capture some details that a DSLR couldn't? Or the other way around?

I'm offering a cake for some more details.


I found some good info on comparing the density & dynamic ranges on ScanDig and scantips.com. I have a hunch that in terms of the density range, devices would be ranked:
flatbed scanner < DSLR < negative scanner < drum scanner. Unfortunately, I don't currently have deep enough understanding to form an answer based on those articles — and my hunch may as well be incorrect.

Do the negative/drum scanners excel my DSLR, that has a dynamic range of 10.8 EVs? (And would my DSLR actually be better than a flatbed scanner when scanning negatives/slides (= transparent material)?)

Answer

I'm the originator of the flickr discussion, and I'm flattered that it's thought to be worth reviving here :-) I went through the process mainly for archiving purposes. The fear of losing these personal negatives was much more important to me than technical quality. Whatever I did, it had to be fast so that I could do every single one of my negatives. I would say that if you have the space to leave the tripod and camera set up for a day or more, it's a decent solution for large numbers of 35mm colour negatives or transparencies. I'm not sure the resolution is sufficient for medium format colour or even 35mm fine-grained black and white film. I'm also not sure it's ideal if your intention is to display or print the majority of them. While the dslr raw files are flexible and high resolution, producing a decent print from one does take a fair few minutes of post processing. But as I said in the original thread, once setup it's very quick. You can rip through a 36-exp film in a couple of minutes.

tl;dr: Good for archiving large numbers of 35mm colour negs. Not ideal for fine art purposes or ongoing digitisation of new films.

Wednesday, October 12, 2011

What automatic-feed photo scanner should I use for 6“×4” / 5“×4” / A6 / postcard size photos?

Question

Can you recommend an automatic document feeder (ADF) photo scanner for usual 6" x 4" / 5" x 4" / A6 / postcard size photos?

I want to stack up say 25 or 30 at a time and leave it running to scan those to a SD card.

I already have a Epson SX600FW all-in-one with auto-document feeder (ADF) scanner, however the guide only slides to a minimum width of A4 portrait dimension. I was thinking of using some blu-tak and cardboard to make a smaller guide for photos however.

Any suggestions?

And do you think putting the photos through a roller in an ADF would damage them and therefore the quality of the scan? Photos are not as flexible as paper.

Answer

Since you're open to a negative scanner consider the Nikon Super Coolscan 5000ED, it can scan whole rolls of negatives, individual 6-photo strips of negatives or up to 50 slides (in a slide hopper) at a time. I use the 6-photo strip adapter and slide hopper.

The 50 slide hopper and negative roll adapter are extra attachments you need to purchase but the negative strip adapter comes w/the scanner.

Its not cheap but it does what it does really well both in quality and speed.

Sunday, July 17, 2011

What are the basic workings of the Lytro light-field camera?

Question

lytro.com describes their new light field camera as being able to capture the entire light field, rather than just one plane of light, thereby allowing for a whole new set of post-processing possibilities, including focus and perspective adjustment.

What sort of sensor could "capture every beam of light in every direction at every point in time"? How would an essentially infinite amount information be encoded and manipulated? Would there be a lens up front in any traditional sense?

Here is the inventor's dissertation: http://www.lytro.com/renng-thesis.pdf

Can somebody boil this down for those of us who are familiar with traditional technologies?

Answer

The easy way to think about this is as follows:

Imagine that instead of one camera, you had a grid of 100 cameras in a 10x10 array. When you fire a shot, each of them shoots at the same time. They will each have a slightly different view of the thing that you are taking a picture of. There are some mathematical models you can use to sort of "reverse engineer" the image and rebuild it in different ways. That's what this is kinda all about, except that instead of 100 cameras, you have thousands, and they are all formed by an array of lenses just above the sensor plane. So the image that comes out of the camera sensor has a bunch of circles of partial images that each differ from the one next to them just slightly. Then they use math to reassemble a single image from those partial images.