tips & techniques - haywood & sullivan

!

published by: HAYWOOD + SULLIVAN

Low-Res OriginalsLow-Res Line ArtFaded OriginalsColorizingFine Line ArtBorders & FramesSignaturesPolaroidsScreensCracks & ScratchesB&W from ColorOptimizing GIFs

Tips & TechniquesMoirésSweet SpotMax Res3-DBackgroundsHigh-keyLow-keyHigh ContrastPoor ColorColor from B&WColor NegsSharpening

http://www.hsdesign.com

! Ways to Eliminate Moires

All text excerpted from “Make Your Scanner a Great Design and Production Tool” © 1995 Michael J. Sullivan

moiré is a wavelike pattern that occurs whenever two or moreregular patterns are superimposed on one another. Moirés

occur when you attempt to scan something that has alreadybeen halftoned — an image that has already been printed in a

magazine, newspaper, book, etc. (see Legal and Ethical Issues ).

One way to eliminate a moiré is to use hardware and/or software thatautomatically rescreens an image during the scanning process. LightSource’s Ofoto is one software solution that works especially well withhalftones above 100-lpi. Another example is FotoLook, a scanning pro-gram with automatic sharpening and descreening that comes with theAgfa Arcus Plus scanner. It works particularly well with color halftoneimages. If you work with a lot of preprinted images, the “push-button”convenience of these packages may be well worth the investment.

For those of you who don’t have either Ofoto or an Agfa scanner, youcan eliminate the moiré in an image-editing program such as Photoshopusing a variety of filters and sharpening. Halftone images all responddifferently to scanning and sometimes only certain combinations offilters produce optimum results. Rest assured, the results are equal to andpossibly better than the automatic solutions.

Don’t expect to get more resolution or detail out of a halftone image thanthe lpi of the original. The best you can expect from your scan is toequal, but never improve upon, the original halftone.

A different kind of moiré occurs when you print black-and-white com-puter screen shots, particularly where a patterned background occurs.You can correct this by scanning the image at an spi number that is afactor of the final printed resolution. For example, if you’re printing animage at a resolution of 1,200-dpi you should scan it at 75, 100, 120,150, 200, 300, 400 or 600-spi. The trick is that these numbers all divideevenly into 1,200, and, thus, no moiré is introduced by the halftoningprocess. Manipulate the image in an image-editing program as describedin this demo for eliminating moirés in traditional halftones.

A

Step 1When the photo is first scanned, asignificant moiré appears in the dotpattern.

What Not to Do

Starting with a 1:1 scan (where a150-lpi halftone was scanned in at150-spi) yields less than satisfac-tory results.

OriginalThis black-and-white photo of a computer chip was clippedfrom a hightech brochure. It needs to be scanned, re-screened and reprinted at its original size.

Prescreened images should be scanned at an spi that istwice the halftone frequency or lpi of the halftone image.The optimum scan for an image clipped from a magazineis 266-spi, twice the typical 133-lpi of magazine halftonescreens. A newspaper image should be scanned at 130spi, or twice the typical 65-lpi of a newspaper halftone.Don’t scan at a resolution higher than what’s necessaryto do the job, because the descreening process is opti-mized for a 2-to-1 sample rate. Besides, scanning animage at a resolution of more than 300-spi creates anunnecessarily large file and requires additional RAMto handle.

Because this image is a 150-lpi halftone, it should bescanned at 300-spi (a 2:1 ratio) and scaled at twice the size

(200 percent) of the original to maximize detail and provide the bestsampling rate for the descreening procedures depicted.

Step 2Apply the Gaussian Blur filter andSharpening tool. Gaussian Blurworks best when it is assigned aradius of 1.0 to 2.0 pixels.

Step 3Reduce by 50 percent and applythe Sharpening tool again. Thefinal image, output at the same sizeas the original clipping, has aresolution of 150-dpi.

Using Photoshop’s Gaussian Blur and Sharpening

Step 2, continued

A setting too high, such as 3.0,destroys the detail in the image.Use 1.5 for most scanning applica-tions.

Using Photoshop’s Despeckle Filter

Step 2Apply a 50 percent scale reductionto the despeckled image and theresults aren’t bad. Although thefinal image seems as though itcould benefit from some sharpen-ing, don’t try it. Because theDespeckle filter didn’t actually getrid of the moiré, the Sharpeningtool would only accentuate thenoise.

Step 1The Despeckle filter is applied tothe same 2:1 scan used above. Thisresults in a slight improvement.You can also try applying the Me-dian filter, especially useful withcolor images. Halftone imagesrespond differently to scanning.Sometimes only certain combina-tions of filters produce optimalresults.

Using Agfa Arcus Plus and Fotolook

Step 2A 50 percent reduction to 150-spiplus sharpening in Photoshopyielded this halftone.

Step 1This 300-spi image is the result ofselecting automatic sharpening anddescreening before scanning.

Tip

It’s important to remember that thin,nonopaque paper allows whatever inkthere is on the back to show through.To prevent this, use black paperbehind your image when scanning.You may need to compensate formuddy or darkened whites created bythe black paper showing through byadjusting the gamma or contrastlevels of the image in your scanningsoftware or later in an image-editingprogram such as Photoshop.

Descreening was set at 150-lpi andSharpness was set at low.

! Finding Your Scanner’s “Sweet Spot”

Most Scanners have minor inconsistencies. Given the tendency, it shouldthen come as no surprise to realize there are some portions of the imag-ing area of your scanner that are better than others. If possible, youshould place your image within this “sweet spot” to obtain the best andmost consistent scans.

Step 1To find your own sweet spot, simply make a scan of the entire image area(my scanner’s live image area is 8.5" by 14") using a clean white surface,such as a sheet of opaque paper, to scan. Set the resolution low, say 72 and100-spi. (For this demo, there’s no need to make a high-memory scan.)

Step 2Bring the image into an image-editing program such as Photoshop anduse the Equalize command to exaggerate any minor differences withinthe image area. As you can see, my own scanner has dark spots on theedges as well as a bright streak on the left. The botchy area about eleveninches down also is something I need to keep aware of when scanningsubtle images.

Once you’ve determined where your scanner’s sweet spot is, you maywant to make a cardboard template to aid you in future scanning.


TheSweet Spot

Too Dark

“Normal” scan of entire live imagearea (8.5" x 14")

Scan after Equalization to enhance minor differences.

! Scanning at Higher Than Optical Resolution


When rotated 90°, with its longest dimension onthe vertical axis, and scanned at 1,200-spi, thesame image is comprised of more “real” databecause most of it was captured at the scanner’strue optical resolution of 1,200-spi.

When you choose a resolution that goes beyond thestated optical resolution of your scanner, it’s importantto consider the orientation of your image before scan-ning.

Most flatbed scanners deliver asymmetric opticalscanning. (Slide scanners rarely work this way.) Forsuch flatbeds, typical “true” optical resolutions are300-spi horizontal and 600-spi vertical. Higher pricedflatbeds may offer “true” optical resolutions of 600-spi horizontal and 1,200-spi vertical. Because yourimage-editing program needs square samples to workwith, flatbed engineers resort to a sleight of hand:They interpolate the horizontal data up to match thevertical resolution, in effect “doubling” the horizontaldata. As a result, when you scan beyond yourscanner’s maximum optical resolution, you’re notgetting “real” data but “best guess” or “faked” data forthe horizontal data.

When scanned with a horizontal orientation atl,200-spi, most of this image’s data is capturedon its longest dimension at the scanner’s maxi-mum optical resolution of 600-spi. The resulting1,200-spi visual is made from a 50/50 split of“real” and “faked” data.

There is a way you can minimize this effect. Whenscanning at a higher than true optical resolution, besure the longest dimension of your rectangular imagematches the higher optical number. If you have a4" x 6" photo, orient it on the scanner by aligning thelongest dimension of the photo parallel to the dimen-sion that has the higher optical resolution. For mostscanners this is the vertical (sometimes referred to as“slower”) dimension. Note: Square images generallyaren’t affected by orientation because there is no netgain to be made by rotating them.

The net result for a rectangular image with a length-to-width ratio of 2:1 is up to 25 percent more “real”data — even though the file sizes are the same!

600-spi(optical)

25%moredata

1200

-spi

! Scanning 3-D Objects

Keep in mind that if you scan at1,200-spi or more, you can enlargeany object for high resolution out-put. This 1,200-spi scan of a dime,output at 600 percent and 133-lpi,yielded this highly detailed image.When you need a close-up of a tinyobject, the scanner works muchbetter than a camera with a macrolens — much faster, too.

Color scans of metallic objects will inevitablyproduce rainbow patterns in some portions of yourimage, similar to what’s shown above.

To remove the rainbow, bring your image intoPhotoshop and select the rainbow regions. Openthe Hue/Saturation controls and individually selectthe red, green and blue, reducing the saturation ofeach as shown.

Before you begin to scan, prevent the object from rolling around bytaping the back of it to a large piece of paper—in this case, I used whitebristol board. The paper serves as the background of the object while it’sbeing scanned.

Use Your Scanner as a Camera!

Scanning Cylindrical Objects

Scanning Metallic Objects

Instant “Zoom” Lens

You can use your flatbed scanner to capture 3-D objects much as a cam-era would. (Assuming, of course, they fit on the glass platen!) Shownhere are just a few of the many things I have scanned over the years.What is particularly nice is how fast and easy it is to obtain “photos” thisway. Plus, the images are yours — no rights need to be negotiated forany of your scanned “photos.”

The obvious problem with flatbed scanners is lack of lighting control.Most flatbed scanners capture some of the dimensionality ofan object by casting a slight shadow. (See the diagram ofHow a Scanner Works) The direction of theshadow depends on the optics of your particu-lar scanner and where the object isplaced on the glass platen. Youhave to experiment with yourscanner to see whereshadows fall each timean object is repostioned.

Scanning FoodThe most important thing to remember is to clean up after-ward. Other than that, you’ll be impressed at how good foodlooks when scanned. The soft, natural shadows created by thescanner are very close to the effect of studio lighting.

Step 1Determine where you want the shadowto fall on your image by making apreliminary scan. Reposition the objectand its backing on the scanning bed ifthe shadowed side of your image isn’texactly where you want it to be. Tocreate a lighter shadow, apply additionallighting by positioning a light source(such as a table lamp) underneath the lid,close to the object, resting on the platen.Be careful when doing this—table lampscan get very hot!

Step 2Use channel masks in an image-editingprogram such as Photoshop to changethe background. First mask the object byusing the selection marquee and theoption key to create a continuous selec-tion path. When the object is selected,save the selection to a channel, invert it(to select the background region), anduse the gradient fill tool to fill the back-ground with a smooth blend.

Step 3The final result is equivalent to theeffect of studio lighting at a fraction ofthe cost.


!


ne of themost

versatile usesof your scanner is inproducing interesting back-ground effects by scanningobjects and textures. Youcan create your own libraryof ready-to-use, copyright-freeimages that can add depth andvisual interest to projects that have no photo budget. Often the qualityof these images is superior to what you would achieve using a studiocamera and lights.

An endless variety of backgrounds is possible — everything from leavesto coffee beans to wood siding can make an appealing backdrop for acatchy headline or block of text. You can also create unique abstracteffects by bringing scanned images into an image-editing program whereyou can posterize them or add a filter.

Even black-and-white-scanners can be used to generate color back-grounds. To make our holiday greeting card, I scanned some pine swagson my black-and-white scanner and then converted the scan to an RGBimage in Photoshop. After I changed the overall color to green, I addedother holiday images.

! Scanning Great Background Effects

Fabric and paper textures yield great results. My favorite trick is to scanrecycled papers for use in four-color print projects where I want a softer,more textural look on a coated paper.

The key to successful scans is keeping your scanner clean. You don’twant kiwi fruit juice getting into your scanner’s electronic mechanisms!After scanning the great outdoors, take the time to clean your equipment.

In general, backgrounds are best scanned at 100 percent. You should alsosave them at the same frequency as the intended output. I typically scanat 300-spi, but this level of resolution usually results in a file of at least20MB for an 8.5" x 11" scan! For less detailed imagery, you may be ableto make do with a lower resolution, but for highly detailed material,nothing less will do.

Aluminum Foil Wood Granite

Coffee BeansLeaves Wool Blanket

Rattan Marbled Paper Silk Scarf

Tip

If you’re planning to over-print text on a four-colorbackground scan, the scanshould be light enough tomaintain text legibility. Forthis reason; it’s wise toeliminate the black screenwhen you convert the scanto CMYK. This increasesthe contrast of black textoverprinting the back-ground and facilitates copyrevisions on the blacktext's film without affect-ing the background image.In Photoshop you caneasily create a customcolor separation that elimi-nates the black screen byselecting Preferences/Separation Setup (underthe File menu). ChooseGCR and set the blackreplacement to None. Thenconvert your RGB file toCMY (empty K) and voila,a color separation with noblack dot!

O

! Scanning High-Key Images


OriginalThis 35mm slide of an ultrafairmodel with lemons is almost ashigh-key as the classic “polarbear in a blizzard.” A histogramof the image verifies this, show-ing a concentration of data inthe highlight regions on the farright portion of the scale.

Step 1After bringing the image into Photoshop, you mayneed to adjust the overall tone of the image. Do thisby choosing the Curves command and adjust the RGBchannel curve to emphasize the highlights (gamma =0.75). The curve shown for this particular imageincreases the contrast in the highlight region, makingthe midtones slightly darker — just what you want.

Step 2In this case, I also selected the red channel of thecurve and adjusted it slightly to make the flesh toneswarmer. Save the curve to disk and label it High KeyCurve or a similar identifier.

igh-key images are those that have most of the data in the high-light to midtone regions. It’s important to note that high-keyimages are meant to be light, as opposed to an image that is over-

exposed or faded (see Salvaging a Faded Original for more informa-tion on restoring a faded photograph). You don’t want to “rescue” a high-key image — you just want to make sure it will reproduce at its best.

To achieve optimum results when scanning a high-key original such asthis, the objective is to end up with clean whites while still retainingdetail in other portions of the image. You may find an initial scan yields asatisfactory high-key image and you may not need to go through thesteps outlined in this demo. However, if you need to make adjustments,and your scanning software supports it, you can create a new tone curveand use it to achieve a better image when scanning.

I scanned the image in this demo with no tone corrections and thenbrought it into Photoshop to determine the optimal tone curve. Aftersaving the revised tone curve and importing it into my scanning software,I was able to rescan this image for optimal data capture.

H

Use the saved tone curve in your scanning application by choosing Im-port Tone Curve. After this, rescan the image to achieve optimum results.Notice that after the rescan the histogram shows more data in themidtones as well as cleaner whites, indicated by the expanded highlightdata in the histogram.

After the rescan, touch-ups can be made in an image-editing program. Inthis case, I removed dust with the Rubber Stamp tool and the model’seyes were brightened with the Dodge tool.

Step 3

When transparencies arescanned on flatbed scanners,light may interplay between

the film and the glass platen, pro-ducing concentric rainbow-tingedrings called Newton rings in certainareas of your image. (This is whytransparency scanners have noglass.) Clean up the rings on yourimage by bringing your scan into animage-editing program, or try arescan after repositioning the filmon a different portion of the platen.

Tip

! Scanning Low-Key Images


ow-key images are those with most of their data in the shadowto midtone regions. They present a challenge to flatbed scanners,which have a hard time “seeing” the dark regions of an image.

The trick is to move some of the data into the midtone regions, whilemaintaining high contrast in the shadow regions.

The automatic “exposure” feature of most flatbed scanners has a ten-dency to compensate for the dark regions, resulting in washed-out scans.To get around this problem, I generally scan low-key images with no tonecorrections and set Black Point to the darkest region of the image.

L

OriginalAn initial scan of this 2 1/4" transparency shows aconcentration of data in the far left portion of its histo-gram. Notice that very little data was picked up in themidtones and highlights of the image.

Scanning low-key images also tends to produce noisein the shadow regions. To prevent this, I suggest youscan without sharpening your image. Later on, youcan select the dark regions of the image and blurthem to eliminate noise, then invert the selection andsharpen the remaining midtone and highlight data.

Step 1After bringing your initial scan intoan image-editing program, adjustthe overall tone by using the Curvecommand and adjusting the RGBchannel so the image’s data risesdramatically in its shadow regions.This gives you the detail you needin the dark areas of the photo.

Step 2In this case, I also selected the redcomponent of the curve and ad-justed it to emphasize the fleshtones in the photo. Save the curveby naming it “Low Key Curve” ora similar identifier.

Step 3Load the revised tone curve intoyour scanning application bychoosing Import Tone Curve, andrescan. As you can see from thehistogram, the new scan of thisimage retained far more detail andcontrast in the shadow regions thanthe original scan.

! Getting the Most From a High-Contrast Original


h, the hardest thing to scan is a high-contrast image, unless, ofcourse, the original image is meant to be high contrast. In that

case, nothing could be easier!

But I’ll assume you’ve turned to this demo because you have a high-contrast image on your hands that you want to make less contrasty, andthat can be tricky.

The problem with a high-contrast original is that all of the data has beensquished into the highlight and shadow regions with very little data left inthe midtones. The job is to extract as much of the data in the shadow andhighlight regions as feasible, and move that data toward the middle whilemaintaining detail and contrast.

OriginalThis image has highlights that are too light and shad-ows that are too dark. This photo was printed on high-contrast (#4) photo paper, which contributed to theproblem.

A histogram of this high-contrast photo shows that theimage’s data is bunched at the ends of the tonal spec-trum. Adjusting the Levels command won’t helpbecause it doesn’t allow you to pull the data at bothends toward the middle of the spectrum.

You also might assume that the classic gamma curveis the way to go with a high-contrast original. How-ever, using this curve results in midtones and high-lights that lack detail. The only curve that works is aninverted S-curve. With this curve both the shadowsand highlights are concentrated toward the mid” tones— just what you want.

A

INPUT

OU

TPU

T

SI HI

HO

SO

Areas of emphasis (Both

highlights and shadows)

Do This:Scanning with an inverted S-curve produces an image that looks less“hot.” Pulling up the histogram of this less-contrasty image confirmswhat the eye can see — a broader tonal range in the midtone regionof the image.

! Correcting Poor Color


hooting a picture in poor lighting conditions often results inan image that has an unnatural color cast. But fear not, it’s easier to

color correct an image in the computer than it is in the darkroom. Thesurefire way to get outstanding scans every time is to scan first with notone controls (e.g., gamma = 1). Then use the Curves command in aphotoediting program to adjust the curves of the image to where the colorshould be. Save the curve and rescan using Import Curves. You’ll get agreat image when you scan this time.

Step 1I increased contrast in the shadowsdramatically by bringing the imageinto Photoshop and adjusting it’scurve as shown above.

Step 2I also added warmth to the pictureby separately adjusting the redcurve to give it more dominance.

Step 3After saving the curve adjustments, I rescanned using the imported curve.A second scan produced this much improved image.

OriginalThis particular image is not onlytoo dark, it’s also too blue becauseit was shot at dusk. The length ofthe exposure also resulted inreciprocity failure — a conse-quence of using film not designedto work in low light condition.

S

!

Note:

The red gel is likely to be the

darkest of the three colored gels,

resulting in a noisy scan. You

should be able to achieve a reason-

able scan in spite of this if your

green and blue scans are relatively

noise-free.

R

Color Scans From Black-and-White Scanner


s unreal as it may seem, you can make color images from ablack-and-white scanner. Although it’s somewhat time-con-

suming and the color isn’t quite up to that of a color scanner,the quality will often do in a pinch. Here’s how to do it.

You’ll need to use three colored filters or gels: pure red, pure green andpure blue (not cyan). These can be purchased at office supply stores(where they’re sold for overhead presentations), theatrical supply houses(where they’re sold for coloring stage lights), or directly from EastmanKodak (where they’re sold for lots of money).

A

OriginalThe color 4" x 5" transparencyused for this demo was scanned at300-spi for best output to four-color film at a scale of 100 percent.I taped my original to the scanningbed to prevent it from shifting eachtime I slipped a colored gel be-tween it and the glass. By taping iton just one side, I could easily slipeach gel between the original andthe bed.

Step 1Slide the red gel between the original and the scanning bed. Before scan-ning, choose Crop and Set White Point/Set Black Point to determine thebest exposure. (For this image, I chose the lightest area, the collar labelon the second sweater from the right, as set white point and the inside ofthe shoe as set black point.) Scan the image in grayscale mode and save itas Red Scan, or give it a similar label.

G

Step 4Combine all three grayscale scans into one RGB file by choosing MergeChannels on the Channels dialog box (under the Mode menu). SelectThree-file RGB method, choose the file names and click OK. Save as

Repeat this procedure for the greenand blue gels, saving each file undera new name.

B

Step 2 Step 3

“Myscan RGB” or a similar name,clean up, and adjust as you wouldany other.

The color image on the left, createdfrom a black-and-white scanner,pales in comparison with its cousinshown below, a color scan producedfrom a color flatbed scanner. But asa last resort, it will do.

! Scanning Color Negatives


here’s no easy way to create a color positive from a color nega-tive with the scanning software that ships with most of today’s

flatbed scanners (although slide scanners are another matter).Flatbed scanner manufacturers usually (but not always) provide some

tools and support for capturing transparent positives but have sorelyneglected the need for information on what adjustments to make whentranslating a color negative into an accurate color positive.

You’ll be in for a shock if you try to create a color positive by scanning acolor negative as a transparency and then simply inverting it. The color ofthe film translates into a color positive that has a sick, bluish cast. This isbecause when the Invert command is applied to the scanned negative, itconverts every color in the image to its complement. Because color nega-tives are made with an orange substrate and orange maps to blue wheninverted, the results are not very appealing, to say the least.

The trick is to bring your negative image into an image-editing program,neutralize the orange cast, “invert” each of the color channels, and savethe final curve to use in rescanning, as shown in the following demo.

OriginalThe original is a 4" x 5" colornegative that was scanned as atransparency at 100 percent and300-spi. I brought it into Photoshopand adjusted each of the colorchannels as shown.

What Not to DoDon’t use the invert command on acolor negative. You’ll end up look-ing awfully blue.

Step 1To neutralize the orange color castbefore inverting, pull up the Adjust/Curves/Auto feature that can befound under the Adjust Curvesdialog box. Auto forces the overallcolor in the image to become neu-tral, removing the orange cast.

T

The image left on your screen after applying Auto is still a negativeimage, but you need to see what you’re going to get before you rescan.To see what the positive image will be, and make any necessary adjust-ments, “invert” each of the color channels. Separately choose each colorchannel (hidden under the RGB channel menu) and invert each one by

Step 2

dragging the begin/end points tothe exact opposite side, and voila,a positive image!

Step 3Go back to the RGB channel andmake any overall adjustments tothe tone of the image. In this case,I wanted darker midtones, so Iadjusted the curve downward. Ifyou need to make any individualcolor adjustments, it’s best to goback to the channel in question andmake minor adjustments to thatparticular curve (such as adjustingthe red curve to the left for awarmer image).

Step 4Save the curve as Color Neg Curveand click OK to exit the Adjust/Curves dialog box. Rescan theimage using the imported curveyou just saved. You’ll get a great“positive” RGB scan that you canthen save, clean up and manipulateas you would any scanned image.

!

Tip: Going too far with thistechnique yields an imagethat looks obviously fake. Itcan also produce ghostly“halos” around sharpenedareas — a sure sign of toomuch of a good thing. Applythis technique in moderation.

Sharpening Scanned Images


he human eye has a natural tendency to view a scanned image as“soft” or out-of-focus. You’d think that a higher resolution scan

would help, but that’s not the case. All scanned images need somesharpening, even those scanned on high-end drum scanners. Masterprinters and color separators will attest to this fact — they’ve been deal-ing with this problem for years.

“Unsharp masking” is the trade term for a standard technique that print-ers and color separators use to sharpen images by accentuating the differ-ences between adjoining areas of significantly different hue or tone. Thetraditional technique uses a mask that’s a slightly out-of-focus duplicateof the original image. When the original is rescanned with this mask,

there is an increase in the degree of contrast at the boundaries oftone shifts, however, subtle gradations in tone and hue re-

main untouched. The result is increased sharpnesswhere you would normally want it — in the mosthighly detailed areas of the image.

You can apply this same sharpening technique toyour images with the Unsharp Mask filter.

AmountRefers to the intensity of the Unsharp Mask effect. A setting between100% and 200% will do, depending on the Radius. The bigger the Ra-dius, the less Amount needed. My “standard” Amount setting is 120%;however, some images need more than this, and others less.

RadiusRefers to the dimension, in width, of every sample that will be affectedby the Unsharp Mask algorithm. I generally choose a setting between 1and 1.5, depending on the resolution of the file and what I’ve designatedfor the Amount and resolution of the file. The higher the resolution of theof the image, the greater the numerical setting for Radius. The formulato use is: Output resolution divided by 200. For example, designate aRadius of 1 for a 200-spi image. Designate a radius of 1.5 for a 300-spiimage. The more Amount you use, the less Radius is necessary.

ThresholdSpecifies how many numbers of samples in an image will be sharpened.A setting of 0 will affect every sample, whereas a setting of 50 will affectalmost none of the samples. Highly detailed images such as line art,require a setting of 3, whereas portraits look best with a setting between5 to 9. (We want to keep minor wrinkles down to a minimum, don’t we?)

It can take a while to come up with the right combination of settings foran image. That’s why many manufacturers of scanning software nowoffer sharpening as an option during image capture.

OriginalThis 35mm slide with no sharpeningwas scanned at 1,200-spi. The overallimage looks slightly (and predictably)fuzzy. The results of several differentcombinations of Unsharp Masksettings are shown.

T

Amount=25Radius=1

Threshold=5

Amount=120Radius=1

Threshold=5

Amount=200Radius=2

Threshold=1

Best Bet

! High-Res Backgrounds From Low-Res Data


Let’s say you have a low-res scan that you’d like to use for a background.You need to enlarge it to fit your overall image area, but you don’t wantthe pixelated look that results from enlarging low-res images. Well,bunky, the only way to achieve this is to fool the eye. Here’s how to do it.

This original is a video capture —a 640 x 480 grayscale image shotwith a B&W video camera. Itneeded to be remade into a slick,print-worthy, four-color image,suitable for display on the cover ofa Boston-area magazine.

I captured the video signal usingthe software that came with myvideo digitizer. I then saved theimage to disk in TIFF format(PICT would have worked just aswell).

I opened the saved video-capturefile in Photoshop. Because Iwanted to add more area to the topof the image, I chose the Image/Canvas command and increased thecanvas size from 640 x 480 to 640x 960, keeping the original dataconfined to the bottom half of theimage area.

I converted the image fromgrayscale to RGB by choosingthe Mode/RGB command and thenenlarged it using the Image/Sizecommand. When you enlarge,choose a resolution that is at leastthe minimum lpi output (for ex-ample, 150-spi), and select theproper size width of your finalprinted image plus allocatedbleed for each dimension of atleast 1/8 inch. This increases thefile size of the image tremendouslywhile the data (and apparent reso-lution) remains the same. Thatmakes sense — you can’t createmore data where there wasn’t anyto begin with!

To fool the eye into perceivingmore data than there really is,perform some painterly “magic” onthe file by choosing a special ef-fects filter. Adobe’s Gallery Effectsoffers quite a few as does Kai’sPower Tools. Some of the specialeffects filters supplied withPhotoshop, such as the motion bluror mosaic, could work as well. Forthis image I chose a combination ofPhotoshop’s radial blur and noisefilters in addition to Fractal DesignPainter’s oil brush tool.

When the effect is complete, con-vert the image to a CMYK file,adjust for press conditions and dotgain, and save as a five-file EPS(also known as DCS).

Step 4The final touch is to place something in the fore-ground that the eye will see first and focus on. In thiscase, the magazine’s cover lines and logo served thispurpose. The overall effect fools the observer intothinking the background contains more data than therereally is. It just goes to prove the old adage: Whengiven a lemon, make lemonade!

Original

Step 1 Step 2 Step 3

!

Using Photoshop’s Gaussian Blur and Threshold Mode

Getting High-Res Line Art From Low-Res Data


ow do you achieve high-resolution images when you have alow-resolution scanner? You cheat! There are several ways of“cheating” or getting around the low-resolution problem. The

best way depends on the type of scanner being used, the scale ofyour final image, and how detailed your final image needs to be.

The easiest solution, by far, is to scan your image as a grayscale imageinstead of as line art. The 8-bit depth of a grayscale image gives theillusion of higher resolution, as compared to the 1-bit depth of a line artimage.

If you own a low-res scanner, or one that won’t support an 8-bitgrayscale scan, you can bring the image into an image-editing programsuch as Photoshop and use the program’s editing tools to adjust its reso-lution, as demonstrated in the following procedure.

You can also use a program specifically designed to remove thebitmapped look or “jaggies” that are seen in images produced from low-res data.

OriginalThis 1-bit scan at300-spi was made froma turn-of-the-centurybook of printers’ andengravers’ decorations.

Step 1Convert your 1-bit line art imageinto an 8-bit grayscale image bybringing it into Photoshop andselecting Gray Scale from theMode menu.

The final 300-spi image usingJAG II.

For comparison purposes, scanningour low-res original in grayscalemode produced this 300-spi image.

Step 2Apply the Gaussian Blur filter,setting the radius at 1.5 to 2.0pixels.

Step 3Open the Levels dialogue box andthreshold the data by moving the Hi

and Si controls toward the middleas shown. You have to experimentwith each image to find out wherethe settings should be for optimumresults.

This is a one-step solution where the scanned image isopened in JAG II and the Anti-Alias command isselected. JAG II does the rest.

Scanning the Original as a Grayscale Image

Hi Si

Using Ray Dream’s JAGII

H

!

Note: Apple’s PhotoFlashsoftware offers automatedtools that make quick workof clean-up jobs such as thisby eliminating scratches anddust marks .

Salvaging a Faded Original


I suspect this section will be one of the most read of the entirebook. And the reason is obvious — there are tons of poor originals

out there!

It is my firm conviction that you can obtain better images from poornegatives, prints, and slides with your scanner than you can in the bestphoto darkroom! This is because the computer contains tools that dark-room people can only dream about. I should know — I was a profes-sional studio photographer many years ago. I will demonstrate how I“rescued” an old picture of my great-grandfather that I found in myparents’ attic by adjusting the tonal settings on my Agfa Arcus Plusscanner before scanning, making further tonal adjustments and retouch-ing in Photoshop. Because the final image was reproduced in four-colorfor the print version of my scanning tips, it was output directly fromQuarkXPress as a set of three-color (Black, Magenta, Yellow) films.

OriginalNotice how the original black-and-white photo has faded — a classic“sepia” look. This image is a good candidate for a desktop scanner sinceall the important data is in the midtones. Because this photo is going tobe reprinted in a way that replicates its sepia-tone coloring, I will scale itat 100% of its original size before scanning. Scanning it at 300-spi givesme the resolution I need for output to a 150-lpi halftone screen at itsoriginal size.

It’s important to make tonal corrections in a faded original by makingadjustments in the lights and darks before you scan, if your scanningsoftware allows this, rather than afterward in an image-editing programsuch as Photoshop. Your goal is to enhance the tonal range of the originalby deepening the shadows and brightening the highlights.

Step 1Preview the scan within your scanning software. Set the output levels toadd depth to shadows and brighten highlights. Because Agfa’s FotoLookdrives my scanner, and it doesn’t offer a histogram view to show thetonal range of the original, I chose to scan the original with no tonesettings and then look within Photoshop at the histogram (using theLevels command) to see where the data lay. You can see how this fadedimage has a severely compressed tonal range.

Step 2I selected the Dmin/Dmax command on my scanning controls to expandthe tonal range of the image. The histogram of the resulting scan showsthe results. I also chose to crop in on the original image since the back-ground contained many stains and smudges and added nothing to thepicture.

Step 3After bringing the scanned image into Photoshop, I made further adjust-ments in the photo’s tonal range by carefully adjusting the Curves controlas shown. This adjustment added punch to the shadows and lightened themidtones without affecting the highlights.

Step 4The only way to clean up an image like this is to get your hands dirty!I used the Rubber Stamp tool to carefully eliminate unwanted scratches,dirt and stains. Use the Burn and Dodge tool as necessary for addedemphasis. In this instance, the chair, suspenders, hat and eyes were selec-tively darkened and/or lightened.

! Colorizing an Old Photo


Converting an old black-and-white photo of a family member into acolor version of the same photo is something your family will likelycherish (Ted Turner not withstanding). The trick is to colorize the entirephoto with warm undertones to add a vintage look, and then individuallycolor different components in the photo.

OriginalThis black-and-white 8" x 10"photo of my dad was scanned at200-spi as a grayscale image.I brought it into Photoshop todo the colorization.

Step1Bring the grayscale image into an image-editing program and convert toDuotone mode. In this case, I chose a warm reddish-brown as the secondcolor. I adjusted the curves for this color and black as shown. Whendone, convert the file to RGB mode and save it under a new name.

When colorizingan old photo,choose subtlecolors for arealistic, time-worn look. Use abrighter, morevibrant palette fora postcardlikeeffect.

Step 2By isolating selected regions(defined by the marquee tool)and using the Paintbrush in ColorMode (instead of Normal), youcan add whatever color you wishto your photo wherever youwant it. Isolated regions shouldbe feathered with a radius of 2.0or more when they’re selected tosoften their edges.

Step 3When using the Paintbrush inColor mode, build up the colorslowly by using an opacity of50 percent or less.

Step 4Individual regions can also becolorized by choosing BalanceColors, under the Image menu andAdjust submenu, and sliding theadjustments to the left (for moresaturation) or right (for less) untilthe desired color is achieved.

!

his highly detailed lithograph by Charles Dana Gibson can bereproduced in a variety of ways, depending on how much detail can

be sacrificed for a smaller file size.

A lot depends on output. Printing it on a 300-dpi laser printer requires aless detailed image than output to film or an imagesetter. When the imagegoes to press, printing it on an uncoated paper requires less definitionthan printing it on a coated paper. A smaller image requires less defini-tion than one that runs full page.

Scanning Fine Line Art


TTo hold the fine lines of this etching, one option is toscan the line art as a grayscale image. In fact, anyonewith a low-resolution scanner almost always gets abetter scan with grayscale than with line art mode.Why? Because resolution is affected by bit depth.The greater the bit depth, the greater the apparentresolution. This means that grayscale im-ages with a depth of 8-bits, alwaysappear to have higher resolution thana 1-bit line art image. A 300-spigrayscale scan of an image shouldshow as much detail as a 600-spiline art scan of the same image.Although the grayscale image hasa lower resolution than the line artimage, it has a larger file size than theline art image.

Grayscale images have other advantages overimages scanned as line art. They can be manipulatedand rotated more easily, and are not bound by sizingconsiderations like a bitmapped line art image is.Bitmapped scans can only be scaled in a page layoutprogram by a mathematical proportion of the outputfrequency. Typical output resolutionsof imagesetters are 1,200, 2,400 and3,600-dpi. Thus, a 120-spi, 1-bit imageimported into a page layout programshould be scaled at 50 percent,100 percent, 200 percent,500 percent or 1,000 percent foroptimal results when printed on a2,400-dpi imagesetter.

The downside to using grayscaleimages is that they need to behalftoned at print time at typical printresolutions of 150 to 200-lpi. As a result, sometypes of line art images, particularly those with pre-cise detail, may look slightly fuzzy. Line art withregular patterns may also create moirés whenhalftoning.

Scanning an image in line art mode may be themost appropriate choice if you’re pressedfor time, simply because it’s usuallyfaster to scan at this setting. Mostscanners return a line art image inabout half the time required fora grayscale image. This can bean important consideration ifyou’re scanning many imagesat one setting.

If your scanner supports high resolu-tion scans, scanning at the resolution ofyour imagesetter’s output (1,200 or even2,400-dpi) creates the sharpest looking output you canpossibly get 2,400-spi scans are serious stuff, but boydo they ever look good! This type of scan is used byprofessional publishers for high-quality art books.Remember, you can achieve an effective 2,400-spiwith a 600-spi scanner simply by reducing yourscan to 25 percent of its original size.

To summarize: Use grayscale for low-resolution scans of line art. Forhigher resolutions, 1-bit mode isbest. Consult the table below:

Scanner Imagesetter Suggested SuggestedMax SPI DPI Scan Mode Line Screen

300 300 1-bit600 300 not recommended because of oversampling

300 600 1-bit600 600 1-bit

300 1,200 grayscale 133600 1,200 1-bit1,200 1,200 1-bit

300 2,400 grayscale 150600 2,400 grayscale 1501,200 2,400 1-bit

Recommended Settings for Line Art Scanning

300-spi, Grayscale at 624k

600-spi, Bitmap at 324k

300-spi, Bitmap at 96k

600-spi, Grayscale at 424k

! Scanning Borders and Frames


Scanning solid black artwork to be used as a border or frame requireslittle other than proper alignment so that corners, vertical lines and hori-zontal lines are all square.

However, scanning borders that have been previously reproduced re-quires clean-up and conversion to suitable electronic artwork. This can bedone easily with an image-editing program such as Photoshop. The finalborder art can then be made more workable by converting it to line artwith a program such as Adobe Streamline, which converts bitmappedartwork into editable line art.

Conversion Using Photoshop

OriginalThis original print came from aturn-of-the-century specifier cata-log of compositors’ borders anddecorations. Notice in this initialscan how the paper has yellowedand the colors are muted. If yoursoftware supports it, you can elimi-nate the yellowed backgroundwhen scanning by mapping thebackground color to pure whitewith Set White Point.

Step 1Bring the scan into an image-editing program. In this case, Iused Photoshop. By selecting thebackground with the Magic Wandtool, you can isolate it from theborder and create a mask.

Step 2Create a pure white background byselecting the mask and filling itwith pure white. Save this as anRGB file. For many purposes, thisscan would do just fine as is, butbecause I wanted a “spot” colorversion for use in a design I wascreating in PageMaker, I took thisversion of the border throughadditional steps as depicted.

I created an electronic mechanical or spot color version of the border byusing the channel calculation commands under the Image menu. In thiscase I (a) duplicated the red channel, (b) calculated the difference be-tween the red and green channels and inverted the results to obtain thered berries, and (c) duplicated the green channel and moved it into theblue channel.

Because I wanted a mechanical comprised of solid art, I used the Levelscommand to eliminate any “grayness” in the final border.

Calculate/Duplicate

Calculate/Differenceand Invert

Calculate/Duplicate

Red Channel

Green Channel

Blue Channel

Cyan Plate

Magenta Plate

Yellow Plate

Photoshop

Streamline

Step 3

Step 4

Conversion Using Adobe StreamlineThe Streamline version shown here was made fromthe final electronic mechanical shown in Step 4.

The electronic mechanical created with Photoshopworks, but it’s an incredibly large file, weighing in at4.2MB — and that’s just half of a complete border!

Adobe Streamline converts color images into line art,resulting in a smaller file size, which in turn meansfaster printing and more flexibility when incorporat-ing your scanned border into other art.

Using Streamline couldn’t be easier. Import theimage into the program, set a few controls to tellStreamline what kind of conversion you want (outlineor inline, color or black and white, tight or loosetolerance, etc.). In a short time you have Adobe Illus-trator compatible line art. Save the file to disk andyou’re done.

!

Tip: I suggest you avoid usingpreviously printed artwork as anoriginal. You can pick up debrisand texture from the paper whenyou scan, necessitating post-scanclean-up. The printing process alsomay have resulted in smeared inkor distortions such as shearing theangle of the bar code or FIM.

You’re also taking a chance of notmeeting postal regulations whenyou copy a non-USPS original —possibly an illegitimate one. It’sjust not worth taking a chance —get a USPS-approved original.

Scanning Postal Codes and Signatures


n electronic file of postal codes can be a real time-saver fordesktop publishers wanting to imprint their envelopes and post-cards with this information. The U.S. Postal Service makes it

relatively easy for you to acquire the art you need as your originalfor scanning.

First, apply for a business reply mail (BRM) or business reply card(BRC) permit from the USPS. A few business days later, you will receivea printout or hard copy original with the Facing Identification Mark(FIM) and the POSTNET bar code for your particular ZIP code as wellas the postage indicia or permit imprint that goes where the stamp isnormally affixed.

Before the days of desktop publishing, this printout was pasted down as atraditional mechanical that was then taken to a printer for printing. How-ever, you can use your scanner to incorporate this artwork into as manyelectronic mechanicals as you wish. The trick is to be sure your scan isperfectly square and that there is no variation in the width of the bars.

You’ll want to scan these items as a single piece of artwork, just as theUSPS furnishes them, with the alignment guides intact. Although thisresults in a larger file, it aids greatly in the placement of your scan in apage layout program.

Bar CodesFor artwork such as this, where accuracy and crisp-ness are of prime importance, high-resolution 1-bitscanning in line art mode works best. Grayscale scanstend to be too fuzzy. Scan at a resolution of 600-spi orhigher to capture the artwork. You may have to rescanand adjust the original several times to be sure theartwork is perfectly square. (Ofoto users have anadvantage because their software has a feature thatautomatically straightens images.)

Whether it’s a postcard or an envelope, I suggest you take a laser print ofthe final artwork to your local U.S. Post Office and have them OK thelayout. They should be able to test the bar codes and FIM on their equip-ment to make sure they can be accurately read.

SignaturesIf you send a lot of faxes, having your signature on hand to place intoyour documents can be a real time-saver. Scan your signature at an spithat is a multiple of your output device. For instance, if your faxmachine outputs images at 200-dpi, you want to scan your signature at200-spi. Save it as a TIFF file, and it’s ready to import into a page layoutor word processing program.

Scan these regions from the Post Officeart in their entirety (as shown) to assistwith accurate placement in yourmechanicals.

A

! Using a Polaroid Camera With a Scanner


hen in a pinch for a quick image, Polaroid offers an instantsolution. Take a picture of your subject with a Polaroid camera,

scan it into your computer and voila — you’ve captured your image!Polaroid’s new professional Polacolor Pro 100 instant print film is ex-tremely accurate and provides highly detailed results. However, be care-ful when using Polaroid B&W instant film — it requires a special liquidcoating to prevent the image from deteriorating. You’ll want this coatingto have plenty of time to dry before putting the image down on the glassplaten — it’s very sticky and hard to remove!

OriginalTake your picture, rememberingto keep the exposure even, thecamera steady and subject(s) infocus. Compose the subject matterto completely fill the image areaof the film to minimize the needfor cropping when you scan. Thisparticular print was made usingPolacolor type 59 instant printfilm, a standard for “everyday” usewith a 4" x 5" camera. After makingsure your print is completely dry,place it on the glass platen of yourscanner.

Scan at 300-spi at 100% — about the maximum resolution limit ofPolacolor instant print film. (It should be noted that Polacolor Pro 100can handle up to 500-spi since it is much sharper than “ordinary”Polacolor film). This 200% enlargement of my original 1:1 scan showsthat the detail available in a Polacolor Type 59 print is quite good. I haveeven used Polacolor prints for full-bleed 8.5" x 11" backgrounds! Ifnecessary, bring the completed scan into a photoediting program and tonecorrect as you would for any other scanned image.

Polaroid even has a new “all-in-one” technology called Print to Press,which includes a calibrated mini-desktop scanner for their 4" x 5"Polacolor 100 Pro film. Shoot your image, scan it, and place it into yourDTP program — with calibrated results! Pretty neat stuff.

Related topics: Polaroid homepage

W

http://www.polaroid.com

! Creating Custom Computer Screens


Everyone likes to customize their computer. Especially if it doesn’t costanything! Nothing could be easier. Follow these instructions to create acustom screen that will appear every time you start up your Macintosh.A bonus is that this method doesn’t use up any precious system memorysince a startup screen is flushed from memory once the system boots.

Nearly any kind of original will do since the computer’s screen resolutionis significantly smaller than what is typically used for print. For mystartup screen I used a 4" x 6" snapshot of my daughter with her cousin.I scanned it in at 300-spi and scaled it to cover an area of 1,200 x 1,800pixels — big enough to fill even the largest computer screen. Once you’vescanned your image, clean up and color correct in an image-editingprogram, if necessary.

Step 1Scan your original to exact screen size or larger than intended screen size(see table at left). Crop and resize in Photoshop to achieve the correctpixel dimensions. Use these measurements for standard display monitorswith 8-bit color.

Step 2Convert to Index/Color Mode/8-Bit/System Palette with dither. Thenchoose the Save As command from the File menu.

Step 3Name the file “StartupScreen” and choose PICT Resource as the FileFormat. Save the file StartupScreen in your System folder, and that’s it!The next time you boot your machine, you will see the startup screenimage displayed.

Creating a Macintosh Startup Screen

Creating a Windows 3.1 Screen BackgroundThe same kind of custom background screen can easily be achieved on aPC. You need to scan the original to exact size or larger than the intendedscreen size and color correct as necessary in an image-editing program asdescribed in the Macintosh version of creating a startup screen. Thismethod does take up some system memory, so take heed.

Step 1Size your image carefully! If your screen background image is too large

to fit on screen or if it doesn’t fit in memory, Windowswon’t load it. Check this chart showing the amount ofsystem memory required to handle an 8-bit colorimage to be sure your PC can handle the image youwant to load on your screen.

Step 2Crop and resize in Photoshop to achieve correct pixel dimensions foryour particular screen. Convert to Index Color Mode/8-Bit System Pal-ette with dither. Choose the Save As command from the File menu andname this file Myfile.BMP or any other legal DOS name. Choose FileFormat: BMP and save in the C:\Windows subdirectory.

Step 3Open the Control Panel in the main program group and choose the Desk-top icon. Select your BMP file from the Wallpaper File list box andcheck Center. Reboot Windows to see your screen background.

Macintosh 12" Display 196k

Macintosh 13" & 14" Display 307k



Creating a Permanent MacintoshComputer Screen BackgroundIt’s easy to create a custom background screen for your Macintosh. Butbefore you begin, you’ll need to purchase a control panel/INIT such asDeskPicture from Now Utilities to load the background onto your com-puter. Once that’s taken care of the rest is simple.

I used the same photograph of my daughter and her cousin to create acustom background screen for my Macintosh in the following demo.

Step 1Before you begin, take note of the amount of memory a back groundscreen will take up in your system folder. Use the table at right to deter-mine the amount of memory required for the desired screen size.

Step 2Follow Steps 1 and 2 for Creating a Macintosh Startup Screen, exceptwhen sizing for your screen, subtract 20 pixels from the height of theimage to allow room for the Macintosh menu bar and save the image as aPICT file instead of PICT Resource. Name this file anything you want,but be sure to check your application to see if special names are required.

Step 3Install and launch the control panel/lNIT to load the picture into memory.Reboot your Mac to see your new background.

12" Display 512 x 384 pixels

13" & 14" Display 640 x 480 pixels



Most people are impressed when they see a picture of your loved oneslighting up the screen at startup. Needless to say, it can remind youwhat’s really important in life.

13" VGA Display 640 x 480 pixels 307k

16" Super VGA Display 800 x 600 pixels 480k

19" Super VGA Display 1024 x 768 pixels 786k

see also:

Creating a Macintosh Background

Creating a Windows Background

! Reducing Scratches & Cracks

cratches and cracks in your photos are a real pain to clean up.Flatbed scanners make the situation worse as they only light an

image from one side, causing shadows and highlight artifacts! The trickis to scan the image twice — once upright and once again upside down.Bring both into Photoshop, rotate the upside down scan 180° (to make itright side up) and merge the two images into one image by pasting oneimage on top of the other using the Layers palette as shown below. Makethe transparency of the top image 50% and adjust the placement of thetop image until both are exactly aligned. Zoom in if necessary. Oncealigned, select “Merge Layers” and voilá — instant cleanup.

S

Scanned right-side up Scanned upside down

♥If done correctly, the two images will tend to cancel the highlight/shadowartifacts introduced by the scanner — and retouching is now made mucheasier! The combination image of my wife Liz shown above is entirelyunretouched except for the use of this technique.

© 1996 Michael J. Sullivan

!

W

B&W Scans from Color

© 1996 Michael J. Sullivan

hen would you ever scan in color an image intended for B&Wreproduction? Well, when you want to obtain better results than simply

scanning in grayscale, that’s when. Here’s how:

First, scan the image as best as it can be in color. You may want to save itfor later use. For this particular image of yours truly I wanted to enhancethe contrast in the shirt and darken the sky. A brief check of the individualRGB channels revealed that the blue channel wasn’t adding anything tothe image. In fact, other than the jacket, the data contained in the Bluechannel was dull as can be. Most of the data I wanted was contained inthe Red and Green channels. (Note that your images will have uniqueproperties of their own.)

I started first by copying the Red channel to a new,empty image that had the same dimensions as thecolor scan. I placed the Red data on the backgroundlayer. Next I copied the Green channel from the colorscan and placed it on a new layer above the Red. Aftertrying various settings, I set the transparency to 60%giving slightly more emphasis to the data in the Greenlayer. Lastly I merged the two layers into one (usingthe Merge command). The result is shown below.

Scanned as Color

New Image Converted from Color

Note how much better the the shirt looks — it now has the contrastneeded to print well. Also note that the sky is darker too. In all, theimage is vastly improved over the straight grayscale scan. The computeris the best “darkroom” I’ve ever owned!

Scanned as Grayscale

This traditional grayscale scan from a colorimage resulted in a lack of contrast in the shirtand delivered dull gray sky — we can do better!

! Choosing Optimal Color Bit-depth for GIF Files

Scanning for the world-wide-web is no different than scanning forprint — except that one has to consider reducing the file size to facili-tate downloading the image — online, speed is everything!

The graphics below demonstrate that the most important considerationis the number of colors and the amount of noise in the image — heightand width are secondary!

Greyscale (8-bit) = 2,433 bytes

17 colors = 1,660 bytes(68% size of greyscale)

6 colors = 1,246 bytes(51% size of greyscale)

(75% size of 17 color)

8 bits = 2,656 bytes

3 bits = 1,494 bytes(56% size of above)






tips & techniques - haywood & sullivan

Documents