Terms from the field of scanning and digital photography.

Since in this article we will talk exclusively about scanning transparent originals - slides and negatives - I will omit all discussions about opaque samples. The article was written for a reader trained in the field of photography and computer image processing, as well as familiar with the basic concepts: optical density range, useful optical density range, photographic material latitude, contrast, average gradient, etc.

What do we have?

D First, let's look at the parameters of the Epson Perfection 1650 photo scanner. It's the only one I have, and it would be strange if I described something else. So, according to some data, this scanner in the mode of scanning a transparent original can perceive a density difference ΔD scanner = 3.2, according to other data its dynamic range is ΔD scanner = 3.0. The research I have conducted indicates much more modest characteristics for this parameter, therefore, manufacturers are disingenuous (although they do not indicate the dynamic range at all, at least for scanners of this level), saying that we can “painlessly” scan a color negative. I claim that in the form in which the scanner is supplied, it is impossible to scan a color negative without loss. So let's get started.

What do all these letters and numbers mean?

D- density, or the decimal logarithm of opacity. It is known that the human eye perceives a scale uniformly increasing in brightness, the fields of which in terms of reflection (or transmittance) do not follow an arithmetic progression (10%, 20%, 30%...), but differ from each other in a geometric progression (1% , 2%, 4%, 8%...) - and this is nothing more than a logarithmic dependence. You probably know that the musical series and its frequencies (string vibrations) also differ from each other in geometric progression. The same can be said about the strength of sound, which is measured in the decibels you know.

So, the human eye perceives the ratio of shades according to the logarithmic law, therefore in scanning techniques, etc. This is the scale used. A change of D=0.3 upward indicates that the eye sees the object 2 times darker. Density is measured in bels.

How the research was conducted

D In order to have a large range of densities, I used a sensitogram black and white film, I know all the absolute densities of its fields (taking into account the minimum density D min, or, more simply, taking into account the “veil density”), measured under the “M” status of the densitometer. Scanning a b/w negative usually occurs in a “mixed” channel, so that’s what I’ll be scanning. I took the lamp’s glow itself as a field with density D=0.0, i.e. scanned area of ​​an image without film. The sensitogram had a maximum blackening of D max =2.3, in order to obtain blackening with a density of D max =2.6, I used a neutral gray filter with a density of D = 0.3, pressed to the area of ​​maximum blackening of the sensitogram. Scanning was carried out by the Xsane program (Linux platform) at a resolution of 300 dpi in black and white mode without any adjustments (brightness, contrast, gray level), the ability provided by Xsane to set the brightness with hardware was not used. The resulting 16-bit file was measured with a 5x5 pixel eyedropper in Photoshop.

Results:

Where: D test - density in the tested negative;

D scan - the value converted from Photoshop blackening percentages to white;

% - percentage of blackening measured by Photoshop.

It is quite difficult to analyze the obtained values ​​without preparation, and it is not necessary. Based on these data, a graph (characteristic curve) was constructed, the values ​​of D test were plotted along the X axis, and the values ​​of D scanner were plotted along the Y axis.

Analysis of the received data

T Now it’s much easier to analyze the graph :-) So, what we see: the graph curve up to D test = 1.6 is quite even and smooth (indicated green), which means the scanner transmits values ​​up to this density almost proportionally, without distortion.

Between D test =1.6 and D test =2.35 the curve looks like a broken line (indicated in yellow), so I dare to assume that in this section of the characteristic curve the scanner produces “invented values”. Those. the matrix perceives them, but produces something unintelligible; in order to “digest” them into a “normal” form, the scanner has to adjust these values. This can be compared to the “decibel level” in professional video cameras. When the illumination level of the object is not enough, the operator turns on the “decibel”, the camera begins to increase the level of the signal received from the matrix, in fact, amplification occurs electrical signal. Both what is needed and what is not needed increases. Thus, simultaneously with the image, noise increases. Something similar happens in the scanner: noise appears in this section of D test, which is why the curve looks like a broken line.

Now comes the fun part. Who wrote about ΔD scanner =3.0 for this scanner? Well, well... Beyond the value D test = 2.35, this scanner does not perceive anything at all! So ΔD epson_perfection_1650_photo =2.4!, and even then, only because D test =2.35 is the last field that has a value returned by the scanner that is different from the previous one. You understand, I couldn’t highlight it except in red :-)

Results:

• The scanner is capable of normally, almost without distortion, perceiving densities of a transparent original up to 1.6;
• The scanner, while introducing distortions and “noise,” is still capable of perceiving densities from 1.6 to 2.35;
• The scanner is blind beyond density 2.4; it perceives any density above this value as black.

What to do?

D Let's see what the scanner manufacturer offers us. In Xsane (to be precise, in the backend of Sane) it is possible to adjust the brightness using hardware. That is, the scanner seems to increase the brightness of the lamp in order to “break through” D max = 2.4. In fact In fact, there is no increase in the brightness of the lamp, the scanner (or rather its firmware) processes the received values, as a result we should get a higher maximum density value, which the scanner interprets as black. So, we will use the capabilities provided by the manufacturer. Set the Brightness value in Xsane. to the maximum that the hardware allows. In our case it is 3.

As in the previous test, we build a graph based on the results obtained (in order not to overload the reader with information, I do not present them).

For comparison, the first characteristic curve (test 1) was left, the new curve (Brightness=3) is indicated in red (test 2). Let's get started comparative analysis: the scanner had ΔD scanner =2.4 and still has, on the basis of which we can judge that the “decibel level” (signal amplification mode) is always on and works in the area D test =1.6 D test =2.4 , since the scanner cannot distinguish any new, higher values ​​of D max_test.

The characteristic broken line in the area D test = 1.6-2.4 has become smooth, which indicates that the scanner firmware, when the brightness enhancement option is enabled, converts the values ​​received from the matrix more correctly in terms of tone rendering. But judging by the images, this does not make the “noise” less, it only becomes more, as they become stronger, or perhaps the “noise” becomes more even. Most likely, the latter is true.

Now let's look at the section from D test =0.0 to D test =0.5, the curve in this section has a low gamma value. That is, the lights will be transmitted softly, and lighter than they actually are.

Let's evaluate the result as a whole: the increase in brightness occurs not due to the effective use of densities, but due to changes in the level of all densities (note how the “black” value is conveyed in tone, if in test 1 it is at the value D scanner = 1.4, then in test 2 at the value D scanner =1,2). There is no point in using this option. We won't get any useful increase in brightness. The “gray field” will become lighter; the “white field” will remain the same as it was; The “black field” will also become lighter, but no new details will appear there. The scanner both “saw” D scanner =2.4 and “sees”. But the level of “noise” will increase.

To be honest, when I did this test, I thought that Epson would still “shift” the curve to the right, i.e. we will lose details in the highlights, but will gain in the shadows, i.e. D scanner will not change, but will work on a different section D test = (D max -D min). Perhaps the manufacturer was trying to implement this feature. This is indicated by the characteristic curve in the range D test 0.0-0.5. I assume that this was done in order not to lose detail in the highlights if the curve shifts to the right. In practice, only the average gradient decreased.

Scanning black and white negatives.

P We try to prove the results obtained in practice. For the “purity” of the experiment, I will always use one single black and white negative. I note that the negative used has normal densities, and is also developed to an average gradient of 0.62, which is the de facto standard. In the film laboratory it is printed at 11 light, which is the norm.

As we have already found out, one of the problems with scanning both negatives and slides is the presence of “noise” in the image. This phenomenon is especially noticeable when scanning fairly dense (dark) originals. This is due to the limited range of optical densities ΔD scannner =D max -D min.

For example: the Nikon Coolscan 4000 scanner is capable of reproducing the optical density range of 4.2 (I don’t want to upset anyone... about the Epson 1650, I already found out its ΔD = 3.0 :-)). Simpler scanners have more modest performance.

The maximum range of optical densities of a b/w negative is 2.5, ΔD max of a slide = 3.0, a color masked negative is about 2.5, but due to the presence of a mask, this type of negative has a large D min.

I am convinced that ΔD scanner =3.0 is quite enough for scanning anything, except, perhaps, X-rays. The problem is where in the negative (slide) this ΔD scanner =3.0 is located. I'll try to explain why.

Let's discard knowledge about photo paper, it can be high-contrast, high-contrast, normal, semi-soft, soft. We will use normal paper in the example, because adjusting the contrast with positive material is a “crime”. The positive should be standard (these are the rules in cinema, and in darkrooms too), you need to reduce/increase the contrast - work with the negative (change the development time, do remote sensing, use filters, countertypes, etc.). So, we use the standard positive.

Do you know what range of densities a positive can reproduce? ΔD=1.0! Total!

That's it! Thus, photographic paper does not reproduce the entire range of densities of the negative, this is not necessary, it is harmful! The result will be a wildly “soft”, low-contrast, “not juicy” picture, even if the print contains both a white and a black field! If you don’t believe me, find a negative with such an interval (ΔD=2.5) and scan it! Finding it is still a problem... Here I used a sensitometric wedge (the same one), I know its densities: black field (veil) - 0.3; white field (maximum blackening) - 2.3, thus ΔD neg = 2.0. I assigned “black” to the point with a density of 0.3, assigned “white” to the point with a density of 2.3, then scanned a sample of our negative in the same mode. “Beautiful”, right? I must admit that I raised the gray level a little, the negative turned out to be completely dark. But the critical points of black and white remained in place. So the average gradient has not changed.

Next, in accordance with the sensitogram, I assigned a “black point” to a field with a density of 0.1 (above the veil), a “white point” to a field with a density of 1.1, and assigned a “gray point” to a field with a density of 0.6, those. I imitated normal photo paper. Here's what happened:

What conclusion can be drawn from all of the above - yes, that the negative contains a huge amount of densities that will not be printed in the positive. At the beginning of the 20th century, there was a story that the average gradient (contrast coefficient) of a negative, when multiplied by the average gradient of a positive, should give 1.0, then, supposedly, the gradations would be conveyed in the correct tones. What's the result? - sluggish images! The product should be 1.7~2.2.

Thus, even ΔD scanner = 1.7 is enough to scan a negative, in case we want to imitate “especially soft” paper.

For clarity, I have marked a useful range of negative densities on the characteristic curve graph. A test object with such densities (a pretty girl and a number of gray densities) is supplied by photographic film manufacturers for fine-tuning the work of minilabs.

As you can see, the useful range of densities of the negative fits without any difficulties into the “safe” range of densities perceived by the scanner. If we exposed the film correctly, then we can even afford D min = 0.5, but for a b/w negative (not masked) this is a very high minimum density.

What conclusion can be drawn? To scan a normal b/w negative, ΔD scanner = 1.6~1.7 is more than enough.

Scanning color masked negatives

TO As mentioned above, a color masked negative has ΔD max =2.5, while having high values ​​of the minimum density D min. For example, the Fuji color negative I measured had the following D min values:

Roughly speaking, this is almost the norm (there is no GOST at hand). Now let’s add the values ​​of the useful density range of the color negative (they are the same as for b/w film) with the D min values ​​for each channel.

For clarity, let's note this on our characteristic curve graph (the characteristic curves of all three channels are similar; it is quite acceptable to depict one)

It's not difficult to notice that red the channel can be placed in the “safe” zone without any problems, there is even a small reserve; green the channel enters the “dangerous” zone with dark areas of the negative (in the positive they will become lights); blue the channel enters the “dangerous” zone halfway, from the gray to the white area in the positive.

Therefore, in red there will be no “noise” in the channel; V green channel “noise” will appear in the bright areas of the positive; V blue channel “noise” will be from gray to white. Let's try to confirm this.

As I said, I will be using the same b/w negative. To simulate color masked film, an unexposed piece of Fuji color negative film was superimposed onto the negative. I will also show histograms of the results obtained. So, let's scan the “color” negative!

By having an orange mask that turns around and turns blue, the positive appears blue. We don’t want to see him blue, what should we do? Use the software to increase the gamma of the blue layer so that the “white” field becomes not blue, but white. Well, let's try. Let's move the sliders on the histogram so that the image becomes neutral gray in all densities, from black to white.

And, lo and behold! The picture is normal in color, well, almost :-). Now let's open it in graphic editor, and look at the image sorted by channels:

Red

Green

Blue

There is almost no noise in the red channel, not much noise in the green channel and quite acceptable, but in the blue channel there is a lot of noise. This is not scanner noise, this is a problem of scanning masked films, or rather “stretching” the blue channel. To prove this, I scanned the same b/w negative, but without a mask in RGB mode and will also demonstrate it broken down into channels:

Red

Green

Blue

As you can see, there is no noise in any of the channels. So, our “enemy number 1” is the yellow-orange mask! Or rather, the high minimum density behind the blue filter. And you have to fight it.

Of course, with photo printing these problems do not arise; the photo paper (not Soviet :-)) is already balanced in terms of the photosensitivity of the layers to match the orange color of the mask. Modern color photographic papers have a photosensitivity to blue rays that is approximately 20-30 times higher than to red rays. The fact is that photographic paper (in photo enlargers, in photo printers) is exposed not to white light, but to the yellowish light of an incandescent lamp, and even passed through an orange mask. In scanners that are not specifically designed to scan negatives, the matrices are balanced to digitize slides and unmasked negatives.

Scanner manufacturers are trying to solve this problem in different ways. My Epson, for example, allows you to scan a 48-bit image, 16 bits per channel, so that there is something to “stretch”. Of course there is an effect. Compared to an 8-bit picture, the difference is colossal. Nikon uses an expensive matrix in its scanners, capable of “seeing” ΔD = 4.2, but there are other problems, precisely because of this :-)

By the way, Epson scans poorly not only color negatives, but also dense (acceptably dense, of course) b/w negatives, as well as dense slides. See the reasons above.

Thus, what is preferable for photo printing (overexposure of the negative by ½ aperture) becomes a disaster when scanning. How to deal with this? What to do?

What to do? Take two!

T oh, what about photo printing: increase the exposure!

If during photo printing we can increase the shutter speed or open the aperture slightly, then when scanning we can only increase the brightness of the light source (i.e., lamp). Although, in the “from the manufacturer” version we won’t even be able to do this. At least, I have not heard about the implementation of this feature in “budget” models. This is all great, of course, but only applies to scanning b/w negatives. In the color version, it is necessary to use exposure adjustment in three channels (in fact, two are enough - in the blue and green channels; I have never seen a blue mask). There are different ways to implement this feature:

1. Use a color mixing head from a color enlarger, or color filters that are opposite to the color of the mask (for example, a compensating blue filter for incandescent lamps) to “neutralize,” so to speak, the mask - make it neutral gray. And increase the brightness of the lamp in order to “break through” the resulting equal channel D min_negative.
2. Use three passes (one per channel) with different exposures for each channel.
3. Solutions for manufacturers:
   • use different types of lamps to scan color negatives (with a higher color temperature) and slides;
   • use lamps of higher brightness (with a margin), and the ability to reduce this brightness (it seems a good idea to use a gray filter introduced in front of the lamp, no changes in color temperature!).
   • Use two matrices. One balanced for slides, the second for masked negatives (expensive way).

What should an ordinary user do? I think that the solutions described in the first and second paragraphs can be implemented at home. The first option seems more realistic to me. At least, you can make a Preview without using specific “software” (no one wants to write? :-)). For example, make a “light box” with the ability to insert filters and thus adjust the color and brightness of the light flux. Or use a color head from an enlarger. And leave the original lamp for b/w negatives with a density normal for the scanner, as well as normal slides.

Still, why is ΔD scanner =3.0 sufficient?

D but because if there is a high density on the slide, then most likely it is not needed, you need to be able to use at least ΔD scanner = 3.0, but in that place of the original density range where it is really required. The problem is where in the negative (slide) this ΔD scanner =3.0 is located. Making the ΔD scanner larger simply makes no sense, and in the case of Coolscan it’s even harmful. Because in the end the negative produces a rather soft (or low-contrast) picture. Any increase in contrast or gamma using “software” increases the level "noise". True, you can scan an image with a resolution of 4000 dpi, make all the adjustments, and reduce the resolution. But then it turns out that 4000 dpi is needed only in order to then reduce the noise. It’s confusing... Sorry anyway. In any case, this is a very good scanner, for the money it costs. In short, you don’t need to increase the ΔD scanner, but add the ability to adjust the exposure!

Take your negativity back! I need a slide!

TO Once upon a time, I didn’t have a very good idea why printers hate scanning negatives; there were a lot of assumptions: they didn’t want to bother with color rendering, raise the contrast - and all that kind of stuff. The main reason is different. In principle, there are always “noises”, either they are visible or not. So, from all of the above it follows that “noise” tends to appear in the darkest areas of the original. When scanning a slide, the “noise” appears in the shadows, and making out the “noise” in the shadows is quite problematic. When scanning a negative, “noise” also appears in its darkest areas. And everything would be fine if there was no need to reverse the negative. Have you already guessed? When turning a negative into a positive, the “noises” appear in the highlights, and it is not difficult to see them, but not noticing them is indeed a problem. In addition, with modern implementations of scanners, even professional ones, it is almost impossible to scan a negative with high quality! To do this you need to control the exposure. Do you know such scanners? If so, please email me the titles and, if possible, links.

What can you say about the new b/w masked films?

WITH I think that Leonid Vasilyevich Konovalov made this “new” film on “Svema” back in 1989 (I could lie, but those are the times), in order to “painlessly” use b/w film frames in color printing. Well, oh well... The main color of the mask is “orange”, therefore, red rays pass through it best. As a consequence, the mask has the lowest minimum density in the red channel. Just scan the red channel. If this option is not in your driver, scan RGB and take the red channel; “the rest” can be thrown out :-).

What does a housewife need?

D In order to scan a standard negative with high quality, a housewife needs a scanner with ΔD scanner >=1.7 and three “handles”. Two for adjusting the amount of blue and green light, and a “knob” for adjusting the overall brightness of the light source. To scan a standard slide, you need a scanner with a ΔD scanner >=2.5 and a “knob” for adjusting the brightness of the lamp.

Conclusions:

1. The Epson Perfection 1650 photo scanner has ΔD scanner =2.4, useful density range ΔD scanner =1.6.
2. In the form in which the scanner is supplied by the manufacturer, it is suitable for scanning:
   • b/w negatives, including masked ones ( red channel);
   • slides of normal density with a small number of dark areas;
   • unmasked color negatives (remember the Soviet film DS-4?);
   • the scanner is conditionally suitable for scanning color masked negatives (the practical use of these “scans” is a big question; they are only suitable for “previews”).
3. The denser the original we scan, the more “noise” we have.
4. The scanner can be adapted to scan color masked negatives by attaching a higher-power lamp to it and using color (blue-blue) filters to correct the color of the light flux.

Lyrical digression (cynical)

IN in general, this is a normal situation when scanners are made by people who have not scanned anything other than their wife’s photograph and have scant knowledge about negatives, positives, and other “nonsense.” Movie cameras (and not only cameras) are made by people who don’t work in cinema. These same guys (Kodak and Fuji) came up with a mask for color film (if anyone doesn’t know, it’s of little use, practically none) and a fourth violet-sensitive layer, instead of changing the spectral sensitivity of the red layer. It is because of these “friends” that in our country, instead of our own, normal one, an incorrect standard for measuring densities was introduced (but it corresponds to the world standard!), and the fact that the curves on an ideal film have a different gamut because of this - no one cares . So, a lyrical digression...

How do you scan?

F Epson's branded "firewood" is only suitable for checking the functionality of the scanner upon purchase, and scanning texts (in 48-bit mode :-)). I use Linux with the Xsane program because there is a "cart and a little cart" manual settings, including hardware settings. And most importantly - Xsane costs nothing! Why don’t I use SilverFast?, because I don’t have it :-), and my demo version is “dead.” If someone gives it, I won’t be offended :-). One of these days I’ll try VueScan, they say it’s a good scanning program, and there’s a version for Linux. I plan to attach a color head from Krokus GFA to my scanner. I think I will do this soon.

The photo shows a landscape near the village of Kazanskaya, Rostov region.

Gratitude.

IN I express great gratitude Leonid Vasilyevich Konovalov for his help in correcting, in his own words, “spelling” errors.

Materials used in writing the article:

• L.V. Konovalov, “How to understand films,” VGIK, 1997.
• V.A. Yashtold-Govorko “Printing photographs”, “Art”, 1967.
• Materials from the site bog.pp.ru

Responsibility?

A what it is? :-)

The author's opinion on the above issue is not the “ultimate truth”. I am only stating what I have checked, tried, “felt”... The author’s opinions, conclusions, results and statements may not coincide with yours or anyone else’s. The recommendations given in this article should not be taken as a guide to action. Any suggestions that you may implement in your equipment after reading this article are done at your own peril and risk. The author does not accept responsibility for any damage that may be caused directly or indirectly by using the recommendations contained in this article.

Copyright

E This article, as well as translations thereof, may be reproduced and distributed in whole or in part in any medium, physical or electronic, provided that this copyright notice is retained on all copies. Commercial distribution is permitted and encouraged; but the author of the article would like to know about such use.

All translations and derivative works made from this article must be accompanied by this copyright notice. This is to prevent restrictions on the free dissemination of this article. Exceptions may be made if special permission has been obtained from the author, who can be contacted at the address below.

The author would like to disseminate this information throughout different channels, but at the same time retain copyright and be notified of all plans for distributing the article. If you have any questions, please contact the author of this article at e-mail: [email protected].

Vasily Gladky, 2003

2.6 Technical data

1) Resolution

Resolution tells us how many pixels or dots per inch can be captured and is expressed in ppi (pixels per inch) or dpi (dots per inch). The more pixels or dots are captured, the higher the detail in the scanned image. A resolution of 300 x 300 dpi corresponds to a total of 90,000 dots in an area of ​​one square inch.

Optical resolution

Optical resolution depends on the number of photocells on the photosensitive element (horizontal optical resolution) and on the step size of the carriage motor that moves the photosensitive element across the document (vertical optical resolution).

2.7 Interpolated resolution

Whereas optical resolution can be achieved by hardware, interpolated resolution is achieved by scanner software. Through algorithms, the software creates additional (virtual) pixels between the real pixels captured by the photosensitive element, thus achieving the highest possible resolution. These additional pixels are the average color and brightness values ​​obtained from adjacent pixels. Because these extra pixels don't actually represent the document being scanned, they are less accurate and don't enhance image quality. Therefore, in terms of image quality for a scanner, the optical resolution value is more important.

Sometimes, however, interpolation is important when the horizontal optical resolution, which depends on the number of photocells on the photosensitive element, is limited. For example, if the scanner were operating at an optical resolution of 300 x 600 dpi, the scanned image would be distorted because the horizontal optical resolution is lower than the vertical optical resolution. In this case, the optical resolution must be interpolated to achieve 600 x 600 dpi.

2) Color depth

Color depth, also called bit depth, indicates how many colors can be represented in a pixel. It depends on the sensitivity of the AD converter. An AD converter that uses 8 bit signals can represent 2(8) = 256 brightness levels for each color (red, green, blue) for a total of 2(24) = 16.7 million colors. In this case we have a color depth of 24 bits.

Internal and external color depth

Some scanners vary in internal and external color depth. The internal color depth indicates how many colors can be represented by the AD converter. External color depth indicates how many colors the scanner can actually render to the computer. The external color depth may be lower than the internal depth. In this case, the scanner selects the most appropriate colors and transmits them to the computer.

Color depth and quality

For scanning black and white documents, a color depth of 1 bit (0 or 1) is sufficient. Scanning color documents requires a much larger number of bits. Scanning a document at 24-bit color depth (16.7 million colors) results in near photographic quality, referred to as true color. Although at the moment most scanners on the market work with an internal and external color depth of 48 bits.

3) Optical density

Optical density is a measure of the opacity of an image area. It indicates the degree of light reflection of this zone. The darker area is a weaker reflection. The range from the brightest area (white) to the darkest area (black) in an image is the density range or dynamic range.

Optical density is measured with optical densitometers, and ranges from 0 to 4, where 0 is pure white (Dmin) and 4 is very black (Dmax).

With a narrow dynamic range, the scanner may not capture some of the image details and lose information. The brightest value that can be recorded is called Dmin, and the darkest value is Dmax. To get the best results, the scanner's dynamic range should include the dynamic range of the document that will be scanned.

In this case, the dynamic range of the scanner includes the dynamic range of the document so that numerous details in the white and black areas can be captured by the device.

The dynamic range of scanned originals varies from document to document.

As you can see from the table above, the scanner must have a particularly wide dynamic range to work with negatives or slides - these are the main properties inherent in photo scanners. The possible dynamic range of a scanner depends on several factors, such as the color depth of the AD converter, the purity of the lamp light and filters, and system interference (noise).

1. CCD or CIS: scanner technologies

There are two technologies of photosensitive elements:

3.1 CCD– a photosensitive element based on CCD (charge coupled devices). Typically, it is a strip of photosensitive elements.

As the carriage moves, light from the lamp is reflected from the scanned media and passing through a system of lenses and mirrors, hitting the light-sensitive elements that form a fragment of the image.

While moving, the carriage passes under the entire media, and the scanner compiles an overall picture from sequentially “photographed” fragments - an image of the media...

CCD scanner technology is quite old and, I must say, leading at the moment. It has the following positive aspects:

1) The CCD scanner provides greater depth of field. This means that even if you are scanning, say, a thick book, the binding area, which is usually difficult to press completely against the glass, will still be scanned with acceptable quality.

2) The CCD scanner provides greater sensitivity to color shades. Although many people call this argument “FOR” CCDs controversial, often CCD scanners actually recognize more colors than scanners of other competing technologies, which we will look at below.

3) CCD scanners have a long service life. Typically 10,000 hours.

Main disadvantages:

1. Greater sensitivity to mechanical influences (shocks, etc.).

2. The complexity of the optical system may require calibration and/or cleaning of dust particles, through certain time operation.

3.2 CIS (ContactImageSensor) – the photosensitive element is a line of identical photosensors, equal in width to the working scanning field, which directly perceive the light flux from the original. The optical system - mirrors, refractive prism, lens - is completely absent.

This is a fairly young technology that Canon is actively developing and promoting.

Main advantages:

1) The scanner turns out to be quite thin. Due to the lack of an optical system. The final product has a stylish design.

2) The scanner turns out to be cheap, because... CIS elements are cheap to produce.

3) Because in the CIS scanner, the mercury lamp is replaced by LEDs, we get several advantages: the absence of a separate power supply (the scanner receives power via a USB cable), constant readiness for work (no time is required to warm up the lamp - you can immediately start scanning after the user gives the command ); and a fairly high scanning speed (which again comes from the fact that the scanner does not need to heat the lamp).

4) The absence of the need for additional power from an outlet makes the scanner mobile: it is light in weight and compact in size, it can be carried with you along with a laptop; You can scan anytime, anywhere, even when your laptop is running on battery power.

5) CIS scanners are usually much quieter than CCD scanners.

6) It is believed that the absence of optics makes the CIS scanner less sensitive to external mechanical influences, i.e. it is more difficult to spoil it with careless handling. But you should also take into account that the tablet glass of such a scanner is often thinner than that of its competitor with optics.

Main disadvantages: CIS elements:

1) Due to the lack of an optical system, the photosensitive element has a shallow depth of field. Up to 10 times smaller than a CCD scanner. This means that scanning thick books is difficult because... The media should be pressed as tightly as possible against the glass.

2) The CIS scanner loses approximately 30% of its brightness after 500-700 hours of operation. Of course, usually for home use this is often not critical, but for those who scan often and a lot, this can be a decisive factor in their choice.

3) A CIS scanner, as a rule, has a smaller color gamut than a CCD, however, Lately The gap between these technologies in terms of color gamut is either insignificant or non-existent.

3D scanning

Currently, tacheometric surveying is widely used to solve construction and architectural problems, which makes it possible to obtain the coordinates of objects and then present them in graphical form. Tacheometric surveying allows measurements to be made with an accuracy of several millimeters, while the measurement speed of the tacheometer is no more than 2 measurements per second. This method is effective when shooting a sparse area unloaded with objects. The obvious disadvantages of this technology are the low speed of measurements and the ineffectiveness of surveying busy areas, such as the facades of buildings, factories with an area exceeding 2 hectares, as well as the low density of points per 1 m2.

One of the possible ways to solve these problems is the use of new modern research technologies, namely laser scanning.

Laser scanning is a technology that allows you to create a digital three-dimensional model of an object, representing it as a set of points with spatial coordinates. The technology is based on the use of new geodetic instruments - laser scanners that measure the coordinates of points on the surface of an object at a high speed of the order of several tens of thousands of points per second. The resulting set of points is called a “point cloud” and can subsequently be represented as a three-dimensional model of an object, a flat drawing, a set of sections, a surface, etc.

A more complete digital picture cannot be provided by any other known method. The shooting process is fully automated, and the operator’s participation is limited to preparing the scanner for work.

Hardware and software

With the advent of digital cameras, this task has become indecently simplified. It is no longer necessary to develop, print and even scan, even the most budget models always write the shooting date in EXIF, and non-budget models also write the location coordinates - all that remains is to copy the files from the memory card and use any viewer program you like. What if you had several generations of photographers in your family, even amateurs?

This article will discuss what to do with old negatives, slides and prints. I note that I did not open America and any more or less qualified user can easily do all this himself.

1. Equipment

Purchasing a professional film scanner was not part of the author’s plans: in addition to negatives and slides, the archive contained about 4,000 photographic prints, which require a flatbed scanner, ideally with automatic feed. Of course, it is better to scan the original negative than the positive printed from it, but it was impossible to figure out for which photographs the negatives were preserved. Toad and common sense did not allow me to buy two scanners for what was essentially a one-time job.

As a result, for 5990 rubles. I purchased a mid-range Epson Perfection V350 Photo flatbed scanner equipped with an AFL (Auto Film Loader). Optical resolution of 4800 DPI allows you to scan negatives and slides. Of course, the dynamic range for this money is not the same as that of professional film scanners, and the speed leaves much to be desired, but...

In addition to the scanner, you will need a photo tank for washing old 35 mm films and a couple of clothespins for subsequent drying. You also need disk space: ~9000 photos scanned in adequate resolution (JPG of maximum quality) took 45 GB from the author. If someone decides to store data in a loseless format (TIFF/PSD/etc.), then even more.

2. Software

4. Background correction. In meaning, this is analogous to Levels correction in Adobe Photoshop. It works well, some frames can be “extracted” immediately at the scanning stage. The “high” level is almost never used: if the frame is initially dark, trying to apply a filter will reduce the contrast to unacceptable levels.

5. Removal of defects. The most controversial filter. In pictures with a large number of uniformly filled areas (sky, calm water, furniture) it really allows you to remove a large number of defects. In photographs with a large number of faces of a small size relative to the frame area (group portraits, demonstrations), parts of the face may be mistaken for a defect, with all that it implies. He especially doesn't like the eyes :) The filter is resource-intensive and increases scanning time.

Sync Picasa Web Albums and Disk Catalog

After the first files from the scanner appear in the directory, you need to set up synchronization with Picasa web albums. In the album properties, select “Enable synchronization”:

After turning on the synchronization mode, do not forget to specify the size of the photos. For Reserve copy need to install " Images in original size" This will not affect the viewing speed, but it will greatly affect the synchronization speed (depending on your Internet connection speed). You can also turn on the “ private"if you don't want (I, for example, don't want:) for your photos to be publicly available. In “private” mode, you can grant viewing and editing access rights to the ones you select. Google users(Google account required).

That's all. Now, if you have the desire and time, you can digitize everything that was filmed in the pre-digital era. The scanner scans, Picasa automatically uploads photos to the web, and you don't forget to do it from time to time. backups to other media.

Don't forget about backup!

Additional Information:

- : A wonderful resource with articles on film scanning.
- in the same place: “Why you shouldn’t scan films on a tablet” (I completely agree, but...)

» Minilab scanning devices

We continue to familiarize ourselves with the principles and features of the work of mini-photo laboratories. Let's try to understand how the density and color characteristics of a negative are measured and exposure parameters are calculated.

To see and analyze what you see (in our case, a negative image on film), you must, at a minimum, have “eyes and brains.” The functions of these organs in the minilab printer are performed by the scanner. Features of the image reading method and the algorithm for processing the received data determine the degree of reliability of calculating the exposure time to obtain a high-quality print.

As for the “eyes” of the scanner, the more detailed information they report the negative to the computer (the greater the resolution and dynamic range of the measuring system), the better. However, in reality, the amount of information processed is limited by the capabilities of the computer hardware and algorithm and the processing time, which must be consistent with the performance of the rest of the printer systems. Moreover, the task that the scanner is designed to solve is not only and not so much in compensating for the previously described factors associated with the negative, paper, optical and chemical paths of the printer. The scanner's algorithm should, ideally, classify the shooting conditions of the object and calculate the correction for its optimal reproduction on the print. It should be borne in mind that the task of determining the subject of photography often cannot be unambiguously solved not only by powerful software and hardware, but also by the operator himself, since ideal density correction for one area of ​​the image can lead to loss of detail in another area. For example, a face in the foreground “embossed” by a flash has a much higher density on the negative than background objects, which may be of no less interest to the shooter. In this case more acceptable solution there may be a compromise where the foreground object is printed slightly denser to reproduce background detail. The problem of reproducing details simultaneously from areas of the negative of high and low density is solved by adaptive masking used in the printer newest generation Agfa MSP DIMAX. A liquid crystal matrix is ​​introduced into the optical path, on which a masking image is automatically formed, compensating for the high contrast of the original negative.

Let's try to figure out how scanners of different printer models ( Noritsu QSS1401/1501/1201(2)/1701(2), Gretag MasterOne/MasterLab(+), Agfa MSC) cope with such a complex task, and to what extent their functioning can be optimized by tuning.

Through the eyes of a scanner Noritsu is a CCD matrix of 128x128 elements onto which a frame is projected through a lens corresponding to the film format. The image is read three times per filters R, G, B. Lenses and filters are located on coaxial turrets. After pre-amplification, the information in the form of an analog video signal is supplied to the scanner’s processor board, where it is digitized and analyzed. Despite the fairly high resolution of the CCD matrix and the solid computing power of the processor, this scanner often makes mistakes when calculating exposure. This is due to both the imperfection of the algorithm and the properties of the measuring system: the characteristics of the filters are not adapted to the spectral sensitivity of photographic paper and are unstable over time (filters quickly burn out). The dynamic range of the measurement system is not sufficiently adapted to the full range of image densities on film. Setting up the printer when working with a scanner involves calibrating the signal amplification (with potentiometers on the board preamplifiers), determining the area of ​​the CCD onto which the frame is projected (for each film format), and storing the values ​​for the unexposed film frame. Practice shows that, to reduce the percentage of defects, operators Noritsu prefer to work in semi-automatic mode, when the scanner corrects only color shifts, and the operator enters density corrections. The color correction function deteriorates as the filters burn out, and often the role of the scanner is reduced to positioning the frame.

Scanner of mentioned models Gretag works much more effectively when determining correction for both density and color. Its measuring system is a line of photodiodes that scans the frame in 12 positions behind each of the R, G, B filters (for a full 135 format frame, an 8x12 data array of points is scanned for each color) ( Fig.1). Such a small resolution imposes certain limitations on the efficiency of recognizing small objects, but the processing algorithm does a good job of classifying typical scenes. The line of photodiodes is the only organ of vision of the printer (printers Noritsu, in addition to the scanner matrix, have three photosensitive sensor R,G,B, performing integral measurement of frame density). Therefore, working without a scanner is only possible in fixed exposure mode. Signals from the photodiodes, after adaptive amplification, are digitized by a 12-bit ADC, which provides a sufficient dynamic range of the measuring system. The algorithm classifies the image, trying to classify it into one of the groups according to shooting conditions (Flash-1, Flash-2, Back Light, Green, Snow). For each group, the probability of a subject being assigned to it is estimated, and the resulting values ​​are involved in the process of calculating the exposure time, along with the parameters in the printer’s memory that determine the degree of correction for each of the groups. The Flash-1 group includes scenes with a pronounced high-density object in the center of the frame (it is assumed that the foreground object was shot with flash and a plus density correction is required for its normal reproduction). Typical example- a face in the foreground, shot with flash. If one or more dense areas of the negative are off-center, the scanner analyzes them color balance and in case of closeness to the balance of human skin, it takes them as the subject of shooting, classifying the plot as the Flash-2 group, and, just as in the previous case, carries out a positive density correction. The scanner assigns a scene to the Back Light group (bright background) if it detects a sufficiently large area of ​​high-density negative, limited to the edges of the frame. This area is classified as a bright background and a minus density correction is applied. A typical example is a bright sky in the background. Scenes with objects against a brightly lit green background are classified as the Green group and require minus correction. It should be noted that although the scanner takes into account color balance when assigning scenes to the Flash-2 and Green groups, the corresponding correction is made only for density. The scanner classifies low-contrast objects on a uniform light background (snowy landscape, sky) into the Snow group. Such stories require negative correction. Special buttons on the keyboard allow you to “tell” the scanner which case it is dealing with.

When calculating color correction, the color shift limits set in memory are used on each of the color axes (Y-B, M-G, C-R plus additional axes for incandescent and fluorescent lamps), above which the correction is not applied (the presence of a natural color dominant is assumed). The degree of correction is determined by the maximum value specified in memory (Color Correction Factor) and the amount of deviation from the “gray center”. It is maximum at small deviations and decreases linearly to zero as it approaches the established limits. The balance of the “gray center” is individual for each film. The memory stores the average density of the normal negative and the mask for each configured film channel according to the DX code. Statistics are kept on these values, and the specified values ​​can be refined over time using statistical data. When calculating the density and color deviation of each frame, the measured integral density is compared with the density of a normal negative, taking into account the mask deviation.

The scanner shows acceptable results when operating in automatic mode. Errors in density average 5-10%. Here are typical cases of errors. When offset from the center until it touches the edge of the frame of a foreground subject shot with flash, the scanner can assign the scene to the Back Light group, instead of Flash-1, and apply correction with the opposite sign. Human faces in a group photo may be too small for the scanner to detect. He will not apply the correction provided for the Flash-2 plot, and they will appear too light on the print. A scene containing white objects shot in evening or yellow-red artificial lighting (ship, building) can be classified by the scanner as Flash-2. In this case, the printer will print too dense, bringing white objects to the normal density of a human face. Often the scanner will try to normalize a light-colored shirt, mistaking it for the main foreground object (Flash-1). It is clear that the portrait turns out to be too dark. Significant color shifts caused by improper processing and storage of the film are almost not corrected. It is impossible to avoid some color distortion if there are small color dominants in the plot. In manual printing, an experienced operator can anticipate some of the situations mentioned and try to correct them. Optimizing the operation of the scanner algorithm is the process of finding a compromise by adjusting the parameters of the same name in memory, which are responsible for the degree of correction of each of the subject groups. Also, a compromise between the print quality of scenes with color dominants and the correction of unwanted color shifts is the adjustment of the correction limit values ​​and CCF.

Shows the best results with automatic printing TFS scanner printer families Agfa MSC. “Total Film Scanning” technology allows you to print all products in a channel common to all films with minimal operator intervention (only film loading). Even films with serious deviations due to violations of the processing and storage process are corrected quite satisfactorily. The procedure for setting up the printer is extremely simple. Let's try to figure out how this simplicity is achieved. The “eyes” of the scanner consist of three lines of 16 photosensitive elements, each of which is exposed to one of the main spectral components of light, as well as an additional line for analyzing the density of the negative ( Fig.2). The scanner filter block has characteristics adapted to the spectral sensitivity of the emulsion of the type of photographic paper used, and is made in the form of a replaceable clip. This allows the scanner to see the negative through the “eyes” of the photo paper. There are no moving parts - scanning occurs as the film is fed. When scanning a full frame of 135-format film, the computer obtains a 16x31-dot array of data for each of the three primary colors. When you load film, it is completely scanned. The data collected from the entire film is analyzed by the scanner's algorithm, and the identified features are taken into account, along with information about each frame. The information obtained is sufficient for the algorithm to correctly calculate not only the correction associated with the characteristics of the films different types and manufacturers, but also compensated for color shifts of films with various deviations from the norm. Classification of individual frames into subject groups is carried out similar to what happens in a scanner Gretag, but with a more reliable result, due to both the higher resolution and information about other frames of the film. The algorithm’s performance with scenes containing a dominant color is noteworthy. When calculating the color correction of an individual frame, the algorithm ignores areas with increased color shift, which makes it possible to obtain an undistorted color rendition of an object in a scene with a dominant color.

Setting the scanner parameters DL1, DL2, DL3, stored in the printer’s memory, allows you to optimize the scanner’s recognition and correction of specific shooting conditions. For example, if you notice that prints from high-contrast negatives containing a foreground subject taken with flash are underexposed, you should increase the DL1 setting slightly. The DL2 parameter is responsible for recognizing and correcting contrasting scenes with a bright background. As is the case with Gretag optimizing these parameters is a search for some compromise. Correction of negatives with low contrast, as well as scenes against the backdrop of large water surfaces, snowy landscapes, etc., is done by adjusting the DL3 parameter.

By correctly setting these parameters and adjusting the threshold for recognizing color dominants, the operator’s work in automatic printing mode becomes extremely simple and convenient, even if the film contains frames with significant deviations from normal exposure conditions.

Concluding a comparative review of the principles of operation of ML scanners and their ability to correct the density and color of photo prints, I would like to note that even the best scanner, equipped with a good algorithm, is not able to compensate for serious deviations in the technological parameters of film and paper processing processes from normal ones. In other words, you should always remember that the corrective work of the scanner is most effective provided that both the film processor and the paper processor are operating normally, from a chemical point of view.

Igor GORYUNOV, Pavel ZAKHAROV

Links to related topics:

Descriptions of mini-photo laboratories
A periodically updated section of the site dedicated to descriptions, first of all, of new, and also, whenever possible, old models of mini-photo laboratories.