How to scan film negatives with an ordinary scanner. The world of pc peripherals

In general, the idea to scan and organize old photographs, of course, was hatched for a long time, it is not easy to decide on such a volume of work on scanning old photographic films (more than a hundred) and photographs (thousands). In general, since childhood, I wanted to have digitized old photographs of my great-great-grandparents, and finally, after 20 years, I decided to move on to this business.

Scanner

The first thing that was the question - of course the scanner. At one time, about 7 years ago, I tried to digitize negatives and decided to stock up on a film scanner. There was not much money, I chose what was cheaper, it turned out to be Miktotek Filmscan 35.


Compared to scanning monsters, it cost a penny, but the result was awesome. I used Silverfast for it as the most advanced software at the time (maybe now). I don’t know why, but sometimes, with different passes, this miracle gave me either a blue or a green photo, then everything hung up, it was unpredictable and very sad, I had to pore over each frame for 10-15 minutes, straightening the histograms and performing other dances with tambourine. In general, this process discouraged me from scanning films for several years, the scanner is lying around somewhere.

Now, having considered all the pros and cons, it was decided the following.
There were a few things to consider:

• for the most part, it will not be me who will scan, but the parents, since they have time now
• you need to scan not only films, but also photos
• need to scan a lot
• no fabulous budget

In addition to all of the above, I understood that now the film is no longer an actual carrier, and therefore most likely it will be necessary to scan only once, although it may take a long time.

So, film scanners have disappeared for two reasons:
firstly, previous experience has shown that you can’t buy such a normal unit for cheap, but what’s cheap - oh, I can’t stand such a hell for the second time.
Secondly, buying a separate scanner for photos and separately for film is also somehow expensive and impractical.
Moreover, I said to myself, if something good comes across, I'll take it to a professional laboratory, you can go broke for a dozen shots.

Having looked at what is on sale from what can scan film in addition to paper, it turned out that the choice is small: either sky-high prices, or just a couple of options. Breaking all the shops operating immediately after the holiday turned out that there are the following acceptable options:

• Epson Perfection V330 Photo (A4, 4800 x 9600 dpi, USB 2.0, CCD, Film Adapter)
• Epson Perfection V370, Photo (A4, 4800x9600 dpi, CCD, USB 2.0)
• Canon CanoScan LiDE 700F (A4 9600x9600dpi 48bit CIS USB2.0 Slide Adapter)
• Canon CanoScan 5600F (A4 4800x9600dpi 48bit USB2.0 Slide Adapter)

The rest was either too expensive, from 10,000, or, conversely, nothing was skillful. Unfortunately, the CanoScan 5600F is no longer available for sale, although the description is very good. The rest turned out to be, according to reviews, about the same, but the decisive role was played by the fact that there were drivers for Linux for Epson, and since I would like to work not only under Windows, I eventually won the Epson Perfection V330 Photo. I couldn’t find out anywhere , how does the 330 model differ from the 370, but since the Linux drivers were mentioned only for the 330, I settled on it, so to speak, "to avoid".

Unfortunately, I have not had time to try it under Linux yet, but in the Windows software I liked the function of removing defects - it works with a bang on black and white old photographs. But you also have to be careful with it - sometimes it can count something worthwhile for a defect.

In the reviews about the scanner, in places there is a problem with the appearance of stripes when scanning films - but I have not observed this yet. Nevertheless, in my opinion, here is something useful about this, found in one of the reviews on the Yandex market: “Two years later, I can report on the result of the investigation: there is a calibration window in the scanner frame where the white balance is set. If dust particles get there, “broken pixels” are obtained, which, when the carriage is run, give stripes. This is most likely a design defect of the new LED backlight (but who will admit it...). So gentlemen, if you have such a scanner,
remove the dust."

With what resolution to scan - this question was not the last. The scanner gives out a maximum of 4800x9600, but when I tried to set this when scanning a 9x13cm photo, the system began to swear at the scale, I had to reduce it.

The criterion for choosing a resolution is simple: if we assume that you can print with a standard resolution of 300dpi, then to get the same image, you need to have at least 300dpi. Considering that the photos are old, there is no point in overestimating this figure - all the same, physical resolution will not allow you to get quality out of nothing. Again, it is unlikely that anyone will ever want to print a poster with the image of a great-grandfather on A1 or even A4 format. If someone writes a book, it is unlikely that there will be a picture larger than a sheet. In general, I decided that for the very old, a two-fold excess will do, for better and later ones - a three-fold excess, i.e. 600dpi and 900dpi respectively. Then I chose what was closest to what the software produced, which came with the scanner.

For negatives, I decided to use the maximum - it was not in vain that I bought with such a resolution ... Most likely, this is a bust of 4800x4800dpi, but you can always cut it down later, but the main thing is that later you don’t have to rescan with other parameters and you can sleep peacefully.

Scans are saved, of course, by no means in jpeg, in order to avoid compression losses. Everything is just tiff. It seems, of course, that the place eats more, but then scan it once - and then you don’t know the problems: I do what I want. I also did not come to this right away, but practice shows that if I save now, I will regret it later and return to this issue, but if everything is to the maximum, then there is nothing to regret later.

Cataloging

Naturally, after digitization, the whole thing must be sorted out somehow. The main task was to sign great-great-relatives, because I wanted to preserve the history of the family for the future, and without competent comments, no one will ever figure it out.

The option to immediately process the photos and upload them to the site was not suitable for two reasons: firstly, you need to process everything at once, and this is the time, and the parents do not understand anything about it; secondly, technologies are changing, and who would know how a site will look like in a couple of decades, if at all it will somehow exist.

The use of a smart cataloging program was not suitable for the same essential reason - there is no guarantee that in a few decades this software will be alive and, accordingly, no one will understand what, where and how is stored in its smart unique format.

The decision came to mind to store the description in a regular text file with the same name as the photo - it is text and text in Africa, for sure anyone can read after decades, even if they come up with some other super-unicode, it is still much more reliable than special software. But as a programmer, I looked at this option with horror - well, it's ugly and that's it. Yes, and uncomfortable in the process.

Parents said that they generally want it as in a Word - here is a photo, here is a signature - and everything is clear. From such a proposal, the hair stood on end, because again - today there is a Word - tomorrow it is not.

Another option is to store signatures in EXIF. Here it was embarrassing that when processing pictures, many EXIF ​​softwares are simply ignored, as a result, losing precious signatures can be irreplaceable.

In general, after analyzing the whole situation, I made a decision: we scan the photo, sign it as EXIF, and then we make all these pictures with signatures read-only, so that there is no temptation to change something, and thus we guarantee the safety of information. If you want to change - make a copy - and go ahead. Well, backups of course. And in general, in the end, we are programmers, in order to sketch out a small script so that the entire EXIF ​​can be exported to a text file just in case, “to avoid” :)

There are a bunch of command line tools for working with EXIF ​​in Linux, but this is unacceptable for convenient work with a large number of pictures. However, here's what's out there: exif , exiftool , exiv2 , googling for more details. Next, I used exiftool for batch processing, but more on that later.

We look at what is from the GUI. Having studied what the OpenSource community offers us, I somehow settled on DigiKam - "digiKam is an advanced digital photo management application for Linux, Windows, and Mac-OSX", as it is written on their website.
I decided to edit in GIMP, the GNU Image Manipulation Program, similar to Photoshop, but open source. Therefore, the ability to edit photos for the cataloging software was not required separately, but in the cataloging itself, several things were bribed.

First, DigiKam edits EXIF, which is exactly what I need.

Secondly, all the photos are immediately on the screen, we sign in the window next to it and immediately move on to the next one - quickly, simply and conveniently.

Thirdly, it was noticed that in EXIF ​​itself there are several similar tags for commenting: Comment, UserComment, ImageComment, and so, DigiKam writes to everything at once, so the probability that other software will read this information is quite high.

In addition, reading the reviews, I was pleased with the idea that, in addition to just EXIF, the softinka can keep a directory, and without copying anything anywhere, unlike many others, but simply processing everything on the spot. This was a huge plus - I did not look for this opportunity initially, but it turned out to be very useful. And what I liked - in addition to entering information in EXIF, she writes it to her database and then it is convenient to sort and search for photos by tags, tags, descriptions, etc. And even if at some point the software disappears and the database also disappears, then a copy of the data will remain in EXIF, which, in fact, is what I need.

Some interesting thoughts on cataloging are described in the already mentioned article “The experience of creating a catalog and indexing a family photo archive. Indexing and digitizing photographic films". So, all or almost all of this data can also be stored in EXIF ​​and, if necessary, exported to any format, as it will be convenient for us.
An additional advantage of DigiKam is that you can choose any photo as the album cover, and I liked the idea of ​​having a photo of the paper album itself as the cover, for which I thank the author.

Another non-obvious moment that I encountered when working with DigiKam: if there are no rights to write to a photo file, then the software silently writes only to its database, without making it clear that there are problems. For a long time I tried to figure out why there is a signature in the program, but not in the file, especially since the “save in file” option is set in the settings. So, keep this in mind - check the access rights, otherwise you can swear for a long time.

Posting on the site

So, the main tasks - scanning and cataloging - have been solved. Now it's time to brag to relatives, show friends a photo. Naturally, by posting photos on the site. Not so long ago, I already made a soft for this case: I put the necessary photos in
catalog, launched - and you're done, the album has become. I wrote about this on Habré last time, “Simple automation: photo album”. Now, using DigiKam, I decided that right in the EXIF ​​tags you can mark a photo, whether it should be placed in a photo album or not, because when scanning there were all sorts of pictures that should not be posted on the site. Yes, and comments can now be taken from EXIF.

Everything seems to be good, but not very good.

Everything on the site is processed in PHP, and it seemed to me that there is a wonderful function for working with EXIF, read_exif_data() , but as practice has shown, this underfunction shows only part of the data, absolutely silent about the rest. I rummaged through everything I could - and the dream of an easy life sank into oblivion, I had to pull out EXIF ​​from files at the stage of generating an album, since command line tools have a place to be.

As a result, I rewrote the script, remembering the sarcastic comment to my previous article “Perl php-file generator ... Monsieur knows a lot ...”, laughed to himself that he was still right that he did not completely rely on PHP - so she would set me up now a leg, and so a couple of minutes - and the problem is solved.

So, when processing a photo in DigiKam, we mark the photo with a flag (it is called PickLabel there). The flag is written to the file in EXIF. When we process all the files from the directory, we pull out the flag using exiftool:

$flagPickLabel = `exiftool -b -PickLabel "$fname_in"`;

Well, then, depending on the flag - if it is, then we process it, if not, we skip it. Everything is set on the command line, so that it is convenient. In fact, you can process a lot of things here, it’s already the taste and color that anyone needs.

Link to the source, if suddenly someone needs to carefully look or even apply: photo_album-r143.tar.gz . How to use - mentioned in the previous article, I will not repeat.

On this, thanks for your attention, and if it was useful to someone, then I am immensely glad.
Criticism is welcome.

UPD: I accidentally found it on Habré about scanning negatives - I'm surprised I didn't notice it before. Let it be here to the heap.

Tags:

Reading notes on the present and future of amateur photography in computer-based periodicals, you involuntarily catch yourself thinking that the public is being systematically prepared for the solemn funeral of the traditional "film" process. Needless to say, the successes of the digital camera industry are impressive, the speed of transferring footage to a computer for further processing, the convenience and "eternity" of storing frames, the possibility of saving time and money on buying and developing films are more than weighty arguments. Cameras with a matrix of more than 2 megapixels allow you to get shots that, viewing on a monitor screen or printing on limited-size formats, cause strong positive emotions. But...
For those who prefer to operate in absolute terms, we recommend comparing three important indicators of digital and analog photography.

real resolution standard (24x36 mm) frame of an amateur color negative film ISO 100 is at the level of 100-110 dots/mm (2550-2800 dpi) and thus there is an average of about 8.6-10.5 megapixels per frame (with the "correct" exposure and "correct" development). Compare with 2-3.5 or more typical 1-1.5 megapixels available on the consumer digital camera market.
Without going into the subtleties of the chemical reactions of color fixation in an emulsion, we note that in the general case the image on the film has color depth, exceeding 36 bits (> 68.7 billion shades). Digital cameras in the vast majority provide a maximum color depth of up to 24 bits (> 16.7 million shades). The human eye is practically unable to see the difference between a 24 and 36-bit image, but any more or less professional processing with subsequent printing requires exactly 36 bits for correct conversions, in addition, a 24-bit image has limitations in displaying low-contrast details.

Important point - comparison cost a high-quality amateur digital model (a reasonable minimum with the appropriate resolution - from $550-600) and a film camera (from $250).
Thus, digital amateur photography is still fatally behind analog in terms of clarity, color accuracy and availability of cameras, although it is ahead of it in terms of operational convenience, low overhead costs and the ability to edit using a computer.
The "computerized" amateur faces a dilemma - the quality of prints plus the low cost of the camera itself, or convenience plus low costs.
Fortunately, there is an alternative that combines some of the main advantages of both solutions. Speech is about film scanners(they are also "film scanners", slide scanners, etc.), which allow the owner of a film camera to transfer an image from a conventional negative film or slide to a computer hard drive in digital form, suitable for further processing or storage as an "eternal" copy .

Film scanners - good and... different

It is clear that not every slide scanner will be a good purchase, models differ not only in workmanship (and price), but also in specific characteristics.

Format of negatives and slides , which can be digitized using a specific model (35mm, APS, etc.) is the first characteristic that you should pay attention to. Regardless of the other advantages of the chosen scanner, it will be useless if it does not support the format of the available films.

Optical resolution - one of the most important characteristics of a film scanner. As mentioned above, the resolution limit of amateur film is about 2800dpi(professional - from 3150 and higher), so the closer the optical resolution of the scanner to this value, the less loss of detail during scanning. At the same time, higher values ​​will not give a noticeable advantage when processing amateur materials.
If digitization is performed for subsequent output on a printer (with an optimal minimum print resolution of 300 dpi), then for printing on A4 format (with an increase of more than 8 times) it is required to scan the original with a resolution of about 2400 dpi, A6 (or 10x15 cm) - 1200 dpi and so on.
Please note that for each format the minimum values ​​\u200b\u200bare indicated - for output site A frame on a full A4 page at the same 300 dpi would require a higher resolution.
Scanning for other purposes has its own requirements. For example, web page design does not require resolutions above 75 dpi, so for a frame that is supposed to be enlarged by 4 times, it will be enough to scan at only 300 dpi (with a corresponding reduction in file size).

In addition to optical resolution, the characteristics of scanners often indicate much more interpolation- obtained due to mathematical processing of the scanned image (sometimes also due to a smaller step of movement of the scanning head). There are practically no serious improvements when using it with full-color originals, since the resolution of the sensitive ruler that perceives light remains the same, but the scanning time often increases many times over.

Optical density range (dynamic range) is an extremely important parameter for the full scanning of negatives and slides.
The very definition optical density refers to the scanned original, it characterizes the ratio of the original light flux to the light transmitted through the film (calculated as the decimal logarithm of this ratio). The minimum value of optical density is 0 (absolutely transparent area, the incident light is equal to the transmitted light), the maximum theoretically possible is 4 (a very black area that practically does not transmit light).

Optical density range is defined as the difference between the minimum (always not 0 - usually 0.1 and above) and maximum optical density (always not 4, usually less than 3.9-3.8) that the scanner can handle. In practice, the width of the optical density range for a slide scanner is its ability to capture low-contrast details in shadows / penumbra and in bright areas (the larger the range width, the more density gradations the scanner is able to separate and the closer areas will be distinguishable). Using a model with a narrow dynamic range, you can only get an overly contrasting image, with "flat" shadows and brightly lit areas devoid of detail.

Let's explain with examples. If the range is specified for the scanner 3.0D, then the maximum density of scanned areas that differ from black exceeds the minimum by 1000 times (with the corresponding number of intermediate gradations). Everything that lies beyond the upper limit for the scanner is equivalent to black. Even if the illumination is increased, losses cannot be avoided - the "shadow will recede", but the details of the areas with the lowest density will disappear.

The 3.6D scanner is capable of more - the maximum density exceeds the minimum by 3980 times, which is almost four times more gradations than the previous example. The scanned image becomes more voluminous, and the transitions of colors and penumbra become softer and more natural.
Currently the minimum allowable indicator for a slide scanner is considered 3.0D, good - 3.2D-3.4D, excellent - from 3.6D and higher.
The range of optical density is strongly associated with another characteristic of the scanner - color depth (bit depth) .
As mentioned above, a 24-bit color representation could well be enough to view a photo, but for its subsequent high-quality processing and obtaining a wide working range of optical density, 36-bit is required (12 bits for each RGB primary color or 12 bits per channel in Adobe photoshop).

The dependence of the maximum achievable width of the optical density range on the color bit depth in a simplified form looks like this:
The 24-bit color representation (16.7 million colors) provides only 8 bits per color and 256 grayscale levels, which is approximately 2.4D optical density bandwidth (256=10 to the power of 2.4).
30-bit (1.07 billion colors) - 10 bits per color, 1024 grayscale and about 3.0D.
36-bit (68.7 billion colors) - 12 bits per color, 4096 grayscale and about 3.6D.

Such maxima are not always achieved, since limitations are imposed by other factors (to achieve 3.6D, the entire chain from a high-quality reading matrix and ADC unit to the interface must support the processing and transmission of 36-bit USEFUL color information, free from noise and interference) .

The name of the characteristic often refers to "external" or "internal". Inside the scanner, a much higher bit depth (for example, 40 bits) can be used, which is required to compensate for matrix noise and other lossy operations. For the user, the output characteristics of the scanner are important - what he will receive in an explicit form. At the same time, the increased internal bit depth in most cases expands the range of optical density processed by the scanner.

Intrinsic matrix noise - a characteristic that is almost never indicated in the passport data of amateur scanners, but can be approximately estimated in practice (in a showroom, etc.) or found out from those who have already dealt with the selected model. In practice, the inherent noise of the slide scanner matrix appears when scanning areas with the highest density in the form of color "garbage", which degrades the overall image quality (natural shadows on slides and purity of bright areas on negatives). The best (and most often expensive) scanners use high-quality matrices, analog-to-digital converters, and special noise suppression and filtering algorithms (unfortunately, amateur models have not yet reached the point of cooling (lowering the temperature) of the matrix used in astronomy). In addition, special noise reduction techniques can be applied.

Focus Depth Range - another parameter that is almost never explicitly indicated in the output of an amateur scanner, but is very important when scanning slides. If the distance to the emulsion of the negative can be set quite clearly by the feeding mechanism, then in the case of a slide the situation is more complicated - the thickness of the frame is rarely exactly equal to the standard one, and noticeable deformation of the film is possible due to the stresses that have arisen during fixing in the frame. The result is that a scanner with a narrow depth of focus range cannot provide sharpness throughout the frame, or even fails to digitize the slide with acceptable clarity.
The narrowness of the depth of focus range can be compensated by the presence of an adjustment (semi-automatic or manual) or special devices for scanning slides taken out of frames.

Scan speed - a parameter that is of little importance when scanning individual frames, but very important if several films are to be processed at once. Fast scanners are able to process one frame in 20-30 seconds, but as a rule only in the "Normal" or "Standard" mode (at the same time, it takes from 25 to 40 minutes to scan one film with 36 frames, including the steps to change negative segments and possible selecting settings for individual frames). The use of special modes can increase the scanning time of one frame many times - up to 3-8 minutes (1.5-5 hours for a 36-frame film, including actions to change negative segments and a possible choice of settings for frames). From the point of view of time costs, the sequence of processing frames becomes especially important, the ability to process several frames at the same time, etc.

Interface - a characteristic that largely determines the speed of downloading the resulting image to a computer and the convenience of connecting a slide scanner. The fastest interface used in slide scanners was and remains SCSI (requires a SCSI controller and a special cable included or in a PC), the newer USB is next in speed (requires a controller and USB ports, in addition to a relatively high transfer rate, it provides also "hot" connection - without rebooting the PC), the parallel port interface closes the list. In the latter case, both a connection to a standard LPT port and a separate board can be provided.

Software features can both significantly improve the overall characteristics of the scanner, and reduce its advantages to advertising phrases. For example, a "competent" automatic converter of masked negatives allows you to get a positive image with reliable color reproduction even without "finishing" adjustments on a good scanner. The opposite is possible (although rare) - a disgusting conversion function will make the model practically useless for scanning negatives, requiring a huge amount of time to adjust the colors of the resulting image in the editor. The user-friendly interface of the utilities significantly reduces the time for scanning (an ill-conceived one increases it many times). The software package with scanners usually comes with the so-called. TWAIN-driver - a special driver that allows you to access and control the scanner from various TWAIN-compatible graphics programs (for example, Adobe Photoshop). In this case, one should not confuse the TWAIN driver with the driver for the operating system - they have completely different purposes.

Completeness - equipping the scanner with the necessary accessories and devices, cables, software, etc. An important characteristic in terms of readiness for work right out of the box (everything is there for connection, calibration, work in the operating system installed on the PC, loading slides and negatives). Completeness also determines the additional capabilities of the model when scanning in non-standard situations (for example, the presence of a special frame allows you to scan slides taken from thick frames, etc.).

It is clear that in addition to all the listed characteristics and sides of slide scanners, the buyer, as a rule, is concerned about the cost of the model. The price range of options on the market that can be classified as amateur is extremely wide - from $ 125 to $ 2800 (in the case of the upper limit, it would be more correct to speak of the semi-professional category), while a higher price does not necessarily correspond to more attractive characteristics.

Optical resolution - measured in dots per inch (dpi). A characteristic showing that the higher the resolution, the more information about the original can be entered into the computer and subjected to further processing. Often given such a characteristic as "interpolated resolution" (interpolation resolution). The value of this indicator is doubtful - this is a conditional resolution, up to which the scanner program "undertakes to count" the missing points. This parameter has nothing to do with the scanner mechanism and, if interpolation is still needed, then it is better to do it after scanning with a good graphics package.

Color depth

Color depth is a measure of the number of colors a scanner can recognize. Most computer applications, with the exception of professional graphics packages such as Photoshop, work with 24-bit color (16.77 million colors per dot). For scanners, this characteristic is usually higher - 30 bits, and, for the highest quality of flatbed scanners, - 36 bits or more. Of course, the question may arise - why would a scanner recognize more bits than it can transmit to a computer. However, not all received bits are equal. In scanners with CCD sensors, the top two bits of the theoretical color depth are usually “noise” and do not carry accurate color information. The most obvious consequence of "noisy" bits is not continuous, smooth transitions between adjacent gradations in digitized images. Accordingly, in a 36-bit scanner, the “noise” bits can be shifted far enough, and in the final digitized image there will be more pure tones per color channel.

Dynamic range (density range)

Optical density is a characteristic of the original, equal to the decimal logarithm of the ratio of light incident on the original to the light reflected (or transmitted - for transparent originals). The minimum possible value of 0.0 D is a perfectly white (transparent) original. A value of 4.0 D is a completely black (opaque) original. The dynamic range of the scanner characterizes what range of optical densities of the original the scanner can recognize without losing shades either in the highlights or in the shadows of the original. The maximum optical density of the scanner is the optical density of the original, which the scanner still distinguishes from complete darkness. All shades of the original darker than this border the scanner will not be able to distinguish. This value separates very well simple office scanners, which can lose detail in both dark and light areas of the slide, and especially the negative, from more professional models. As a rule, for most flatbed scanners, this value ranges from 1.7D (office models) to 3.4 D (semi-professional models). Most paper originals, whether photographs or magazine clippings, have an optical density of no more than 2.5D. Slides generally require a dynamic range of more than 2.7 D (Typically 3.0 - 3.8) for high-quality scanning. And only negatives and X-rays have higher densities (3.3D - 4.0D), and buying a scanner with a higher dynamic range makes sense if you will work mainly with them, otherwise you will simply overpay.

To convert negatives or slides into digital format, a special device is used - a film scanner. It differs from a conventional scanner in that it is designed to process small transparent images that have a high resolution. Although many are supplied with special modules that allow you to scan slides, the resulting product is of poor quality.

Only CCD (CCD) scanning elements can provide the required high resolution images. Therefore, all film scanners are built using them. Some models have one line of CCDs. In this case, digitization requires a three-fold pass prolongs the scan, but does not affect its result. Basically, a film scanner has a CCD, and the image is digitized in one pass. Some models use a multiple pass to reduce errors in the final image.

An important parameter that you should pay attention to when choosing a scanner is the optical resolution. The width of the most common film is 35 mm, and the image itself is even smaller. Therefore, the optical resolution must be at least 2400 dpi (dots per inch). There are scanners that provide 4800 and 5400 dpi. And although the current level of technology makes it possible to achieve even greater values, this is impractical - the grain size of even a fine-grained film will be much larger than a pixel.

Particular attention should be paid to the dynamic range or optical density. The higher the value of this setting, the better the negative scanner can render halftones and smooth color transitions. For high-quality film processing, the optical density value should be in the range from 3.2 D to 3.6 D. It makes no sense to purchase models with c, since the vast majority of films have just such values.

The quality of digitization is also affected by the bit depth of the representation of light, which characterizes the color rendering. A modern film scanner may have a 42 or 48 bit color representation, but processing in this format is only used internally by the scanner and serves to reduce the "noise" of the conversion. The final image has a 24-bit color encoding standard for computer technology.

The slide scanner in most cases is connected to a computer via a USB interface. More expensive models can connect via SCSI-2 and (FireWire). In this case, quite often there is a board with this controller in the kit.

A film scanner almost always has to improve the image. These are Digital ICE, which allows you to remove dust and scratches from the image without affecting the main image, and Digital GEM, which allows you to eliminate graininess, and Digital ROC, which allows you to restore colors in faded photographs, etc. Quite often, all these tools are combined in one package Digital ICE4 Advanced. The use of these technologies significantly prolongs the scan, but the result is excellent. For similar transformations in Photoshop, it will take much more time, and the result is by no means guaranteed.

Even the highest resolution will not be able to produce a quality image if the digital values ​​obtained during scanning do not adequately reflect the colors of the original image. Two characteristics of the scanner play an important role in correct color reproduction.

Firstly, this is color depth, i.e. the number of bits used to encode the color of each digitized pixel.

Secondly, this is dynamic range, i.e. the range of tones in the original that the scanner can discern, from completely transparent to completely opaque.

About color depth

Much of the modern software that comes with a scanner creates a 24-bit color file. However, the scanner's internal analog-to-digital conversion can specify color values ​​with 30, 36, or even more bits. This implementation is adopted because the 16 million colors available at 24 bits per pixel (8 bits for each of the primary colors red, green, and blue) can be unevenly distributed in an image. Most often, shades are lost in the shadows and in the brightest areas.

We must not forget that the presence of noise is characteristic of any semiconductor device - photosensitive elements (CCD and CDI) are no exception. A certain error in the analog signal is also introduced by the circuits of the analog-to-digital converter.

With a very high bit depth, and hence the accuracy, of analog-to-digital conversion, it is quite easy to “catch” signals that are very similar to noise. Hardware circuits and software modules can simply discard (filter) this noise-like information. This leaves a fairly wide range of values ​​for processing and saving in the final 24-bit file. Scanner software determines those 24 bits out of, for example, 30, which correspond to the best reproduction of light and shadows. Thus, increasing the bit depth of the analog-to-digital conversion leads to "stretching" the color depth at the output of the scanner to full-fledged 24 bits.

Unfortunately, the color depth characteristic cannot be used to judge whether all these bits actually contain visually important information. A significant role in the quality of the final image is played by the sensitivity of the sensors and the quality of the analog-to-digital circuit, as well as some other factors. However, on average, we can assume that the greater the bit depth of the scanned image, the higher the quality of the picture, although according to many assurances, the human eye is not "designed" for a color depth of more than 24 bits.

About dynamic range

This characteristic is rarely listed for lower-end scanners, but it is very important for professional imaging, and, first of all, when working with films. Optical density is inextricably linked with the dynamic range characteristic.

Optical density is a characteristic of the original. It is calculated as the decimal logarithm of the ratio of the light incident on the original to the light reflected from the original (for opaque originals) or transmitted through the original (for slides and negatives). Minimum possible value of optical density 0.0D is a perfectly white (transparent) original. Meaning 4.0D corresponds to an extremely black (opaque) original. In relation to the scanner, its optical density range characterizes the scanner's ability to distinguish nearby shades (this is especially critical in the shadows of the original). The maximum optical density of the scanner is the optical density of the original, which the scanner still distinguishes from "complete darkness". All shades of the original "darker" than this border, the scanner will not be able to distinguish. In practice, this means that an "office" scanner can lose all the details, both in dark and light areas, even in a regular photo, not to mention scanning a slide, and even more so a negative.

So, for example, if the scanner has a dynamic range equal to 2.5D, then it will be able to adequately digitize photographs, but will not be able to work with negatives that have an optical density of more than 3.0D, i.e. that the scanner will not perceive the darkest areas of the image and will produce an inferior scan.

A typical film has a minimum density of about 0.3 (50% opacity) and a maximum density of up to 3.3 (99.5% opacity): the range is around 3.0, although some slides range as high as 3.6. If the slide has a maximum density ( Dmax ) 3,3 , and the scanner operates with values ​​only up to 3,0, then the details of the colors are denser 3,0 are likely to be black.

Regular color photography and printed matter have a dynamic range of up to 2.5D. Negatives and x-rays - 3.0-3.6D.

Inexpensive flatbed scanners have dynamic range 2.0-2.7D, good 36-bit 3.0-3.3D, latest models - 3.6D. The range of optical densities of the scanner is determined, first of all, by the quality, type and bit depth of the ADC, CCD matrix and the algorithm of the scanner controller, i.e. built-in scanner software. Mathematical dynamic range limit for scanner with 30-bit ADC - 3.0D, and for a 36-bit scanner - 3.6D(decimal logarithm of the number of possible gradations for each color, which is equal to 2 to the power of the number of bits per color).

It should be understood that it is impossible to scan a negative with acceptable quality using a conventional 30-bit flatbed scanner, even if a slide module is sold for it. Even with best-in-class real dynamic range, the 30-bit scanner can tolerably scan color slides - but don't count on acceptable results with artistic black-and-white negatives shot by a professional photographer. Negatives require a different class of scanner.

Comparison of characteristics of density ranges should be done with caution. There are no standard procedures for measuring and recording a range of densities. Some manufacturers may perform tests to measure the actual, practical range. Others give only theoretical limits for their scanners. It is impossible to make a decision on the choice of one AI or another model only on the basis of the declared characteristics - it is better to perform several trial scans.

It should be noted that slide scanners with a density range above 3.4 cost over $10,000. It's expensive, of course, but flatbed scanners with a comparable density range, such as Agfa's SelectScan Plus, Linotype-Hell's Topaz, and Scitex's Smart 340 cost over $30,000.

There is always a price to pay for quality.