Zootaxa Digital Imaging Guide

1) Introduction.
Zootaxa strives to publish quality manuscripts in a timely manner. The quality assessment comprises both the scientific content of the manuscript and the quality of the illustrations. Illustrations deemed insufficient by any of the editors will be returned to the authors until publication quality material is presented. It is not the duty of the editors to prepare or mend digital files. The editors only reject substandard work. Evidence of inferior illustrations published elsewhere shall not be used to argue for acceptability in Zootaxa. Peruse recent issues to judge minimal quality standards for Zootaxa.

Here a somewhat lengthy guide to digital imaging specifically designed for Zootaxa is provided. The file preparation guide is given specifically for Photoshop, most likely the most widely used imaging application. Other applications may be used as long as a publication quality file is achieved. Zootaxa neither requires the usage of nor endorses any particular product.

QUICK SUMMARY

SUGGESTED FURTHER READINGS.

SCANNING, DIGITAL SOURCE FILES

2) Source files.
2A) Scanning of transparent and reflective artwork.
Transparent material includes slides and negatives from 35 mm to 8x10 inches, whereas any kind of print photograph or drawing is considered reflective. Scanning any of these materials requires assessment of the following parameters: color space, resolution, file type.

COLOR SPACE. Black and white material should be scanned in greyscale mode for pictorial (half-tone) images, and as bitmap (black and white only) for line art. Color material is best scanned in RGB (Red Green Blue) mode [Note for the technically inclined: Adobe 98 RGB, Prophoto are good choices]. RGB is preferred over CMYK (Cyan Magenta Yellow Black) as it has a larger color gamut (more colors), and is the color space used by all monitors. Additionally, jpegs created from CMYK files are unreadable in many programs. Zootaxa accepts both RGB as well as CMYK files for color images. PLEASE send B&W images as B&W files (No RGB/CMYK/index color SEM images, unless colorized).

RESOLUTION. For Zootaxa, the required resolution for half-tone images is 300 dpi (dots per inch), for bitmap files it is 600 dpi. The maximum size of illustrations in Zootaxa is 17 x 25 cm = 6.693 x 9.843 inches. Therefore, the maximum image size in pixels at 300 dpi 2008 x 2953 pixels, and at 600 dpi = 4016 x 5906 pixels. The scan resolution is taken for the size of the original, not the target size. Example: A 35 mm slide is 36 mm wide. For a full width picture in Zootaxa, the slide has to be scanned at (17 cm/3.6 cm) x 300 dpi = 4.72 x 300 dpi = 1417 dpi. As most images will be cropped, scan the image at up to twice the minimally required resolution, for a slide, for instance, use 2000 dpi. Most scanning software allows any resolution to be used when scanning. Some older software may only have a number of steps, in the latter case, always choose at least the next higher setting then what is required.

FILE TYPE. Save all half-tone files as .tif files without LZW compression. Save bitmap files as bitmap. DO NOT SAVE ORIGINAL SCANS AS JPEGS, GIFS, OR ANY OTHER COMPRESSED FILE FORMAT.

[Note for the technically inclined: compressed file formats such as jpeg reduce the amount of image information, i.e., delete some of the data. This deletion is carried out every time an image is saved in jpeg mode, hence the image gets progressively degraded. This degradation is visible, particularly in print, when color or B&W stripe patterns appear, instead of continuous tonal gradients.]

SCAN ADJUSTMENTS. Most scan software allows some pre-scan adjustments. These can often be accessed through a tools palette; otherwise consult the documentation for your scanner/software (example below is for the Nikon SuperCoolscan 4000). The histogram will show the spread of greyscale values from 0 = black to 256 = white [this range is for an 8 bit scanner. There are some scanners that will allow more values, up to 16 bit. Use 16 scanning and files if you will need to make major adjustments to images, or if you need fo bring out detail in the deep shadows or brightest highlights.]. Usually, at the bottom of the histogram there are three triangles: black, grey, white.

Place the black/white triangles at the end points of the histogram. The grey triangle will shift proportionally. If the histogram does not show a nice bell curve, adjust the position of the grey triangle towards the modal class of the histogram. Often, the grey point is best set in the mid-slope adjacent to the modal class of the histogram. There are, however, no set rules: see what is pleasing to the eye and shows all the features you want to illustrate.

In some cases (for instance a dark object photographed on white paper), there will be a range of light values showing all the detail of the paper structure. As the paper structure bears no germane information, all the whitish values can be made all-white. Slide the white triangle to the left, until highlight information in the dark specimen is lost. Then move the white triangle back a few points (5–10). This adjustment will make the structure of the darker object more prominent, i.e., help to emphasize the features you want to show. Similar adjustment can be carried out for unimportant dark areas.

Raw scan preview without any adjustments: notice lack of contrast and the grey paper background.	Lightness adjusted so that white point and blackpoint are at ends of histogram. Notice that the paper is still grey
White point further moved to left, so that all paper tones are all-white.	Slight adjustment of tonality by moving greypoint to left (1.23). Final image is much more punchy than original figure.

Caprellid isopod on hydrozoan. Note that the dark background does not contain any germane information: it can be made all-black. A dark area was selected with the black eyedropper where arrow is on image. This will be the all-black reference point.

Right: Same image, after dark/black point was clicked: Notice that the RGB curves are steeper, and that the caprellid stands out much more from the black background.

Occasionally, detail in the shadows or highlights is insufficiently shown. Some scan software allows adjustment of the curve. Curve adjustments are mostly trial and error, so try adjustments until the contrast is as desired. You can always cancel. An example of curve adjustments in Photoshop is shown below.

SUMMARY: Do adjust the greyscale (the triangles). If necessary do some curve adjustment, though additional fine-tuning can be carried out in Photoshop. The better the original scan, the better the final product.

COLOR ADJUSTMENTS. This is the most difficult part. If you are somewhat experienced, you can do initial adjustments in the scan sofware. Often you can adjust the curves for each of the colors/alpha channels: red, green, blue.

Alternatively, some scan software allows designation of a white reference point and a black reference point with eye-dropper icons. When hunting for the lightest and darkest points of your image, watch the brightness values in the info area for R, G, and B. Once the highest values have been obtained, click the eye-dropper for the white point. Do the corresponding actions for the black point. This procedure works fairly well.

Raw preview of histological section, notice overall yellow-pink coloration of image.	White eyedropper was selected and clicked in the brightest area, above and to the right of the eye marked by the arrow. Notice that the color cast has disappeared, but the image looks overexposed. The curves plot shows that blue and green have now a steeper curve than red.
Mid-grey point is moved to the right (0.56) giving more contrast in the image. This is a very good starting point for scanning and further refinement in Photoshop (cleaning, levels).	Same image with ROC applied. Notice odd result. ROC does not work well with many scientific images.

USE OF ROC, ICE3.
Restoration of color (ROC) works fairly well with old, faded images. Use it for salvage operations. It may give odd results for some scientific photos (such as, histological images, see above). Use the color adjustment with black and white eye-dropper described above instead. ICE3/ICE4, the automatic dust and scratch removal algorithm, works, but removes too much critical detail. DO NOT USE ICE3/ICE4. Instead, use a Photoshop filter with better adjustment possibilities (filter: noise: dust and scratches): see below.

2B. Digital original files.
FILETYPE. If possible, save digital captures as RAW file (or proprietary equivalents such as Nikon NEF, or Canon CR2), the second best option is uncompressed .tif files, least preferred is highest quality jpeg. If the files were not captured as .tifs, convert the files to umcompressed .tifs before altering the files in any way.

RESOLUTION. Save all files at the highest available resolution. Often this is still too low for the required resolution for Zootaxa. Upsampling of images is best carried out with GenuineFractals; resolution can also be increased in Photoshop (under the menu options Image: Image size), though results are somewhat lower in quality. Expect visible image problems when the file is upsampled more than four times in linear dimension from the original size (first generation SEM digital captures).

FILE PREPARATION AND MANIPULATION FOR HALF TONE IMAGES

3. Cleaning images.
All scanned hard-copy images must be cleaned from dust, scratches, and any other irregularities. Digital captures may also need to be cleaned, as dust can settle on the light sensor, or anywhere on the optical pathway. Artifacts often seen in Automontage-type images also need to be mended. Before scanning, blow off the dust with compressed air or a blow brush. This will save a lot of time in post-scan cleaning. IMPORTANT. The scientific integrity of the image MUST be maintained. Any file manipulation that may be interpreted as altering scientific content must be explicitly indicated in the manuscript (such as the use of single color channel in B&W image to highlight certain structures).

To remove small dust, use the Photoshop filter under the menu options noise: dust and scratches. A good balance between removing dust while preserving image detail is found with the following settings: diameter = 2, threshold =12–14. Photoshop gives a preview of the result. If too much/too little is removed, first adjust the threshold, while keeping the diameter small.

Next, view the original scan in Photoshop at 100%. Any visible irregularity has to be corrected. Good results are obtained with the healing brush (the band-aid icon in Photoshop 6 through CS6) and the cloning tool (the rubber stamp in Photoshop). Suitable brush size is 9–13 pixels in diameter with feathering set to 25%. Check the Photoshop manual on how to adjust brush sizes; there are major differences between earlier versions (<=5), and more recent releases (>=6).

4) Adjust overall contrast/tonality.
You MUST use the full range of contrast. Overall contrast is best adjusted using Photoshop-Levels (command-key L: Top left image below). The histogram will show the spread of greyscale values from 0 = black to 256 = white. At the bottom of the histogram there are three triangles: black, grey, white.

• The black triangle sets the black point: anything to the left of the black point will be shown as pure black. The darkest information-bearing point should be no more than 6 points away from the black point.
• The white triangle shows the white point: anything to the right will be shown as pure white. The lightest information-bearing point should be no more than 6 points away from the white point.
• The grey triangle identifies the 50% grey point.

Place the black/white triangles at the end points of the histogram (top right image below). The grey triangle will shift proportionally. If the histogram does not show a nice bell curve, adjust the position of the grey triangle towards the modal class of the histogram. Often, the grey point is best set in the mid-slope adjacent to the modal class of the histogram (bottom left image below: note grey point is at 1.21). There are, however, no set rules: see what is pleasing to the eye and shows all the features you want to illustrate. Click ok. If you recheck Levels (bottom right image below), the histogram will be somewhat jagged. Don't worry, it will smooth when downsampling image towards the end.

In some cases (for instance, a dark object photographed on white paper), there will be a range of light values showing all the detail of the paper structure. As the paper structure bears no germane information, all the whitish values can be made all-white. Slide the white triangle to the left, until highlight information in the dark specimen is lost. Then move the white triangle back a few points (5–10). This adjustment will make the structure of the darker object more prominent, i.e. help to emphasize the features you want to show. Similar adjustment can be carried out for unimportant dark areas. For examples, see above under scanning adjustments.

Sometimes, adjustments using Levels will not do. In extremely contrasty pictures, neither the white nor the black areas show needed detail (image below left). Use Photoshop Curves (command-key M). Place an adjustment point in the dark area and move the curve up, i.e., make the bottom left portion of the curve steeper: notice the improved detail in the dark, though the white areas are washed out even more. Place a second adjustment point in the upper right hand portion of the curve and pull it down. Now the top right portion of the curve plunges down steeply. The overall shape of the curve is now sinusoid with the middle being flatter (image below right). Experimentation will indicate the best result. After applying curves, recheck Levels.

5) Adjust overall color.
This is the most difficult part. With some care, reasonable results can be obtained, and those are generally sufficient for Zootaxa. If your work critically depends on highly accurate color representation, it may be easiest to seek assistance from graphics professionals. Zootaxa is not responsible for making color adjustments to your files.

[Note for the technically inclined. For optimum color adjustment a properly calibrated monitor and a thorough understanding of .icc-profiles and color spaces is necessary. Calibration of a monitor should be carried out with an external calibration spider such as Monaco Optix or ColorMunki. As supplemental reading Fraser et al. (2005: Real World Color Management, Second Edition. Peachpit Press, Berkeley, 582 pp.) and Rodney (2005: Color Management for Photographers. Focal Press/Elsevier, Amsterdam. 461 pp.) will prove invaluable. Taking advantage of LAB color-space in many cases will produce superior results compared to work in RGB. E.g., increases in color saturation in a and b channels do not affect lightness/brightness of the image areas; un-sharp masking in l-channel will only accentuate brightness differences, but will not introduce color patterns. See Margulis (2006. Photoshop LAB Color. Peachpit Press, Berkeley. 366 pp.) for details.]

There are many ways of adjusting color in Photoshop. Possibly the simplest way is using Curves (command-key M). Curves allows to designate a white reference point and a black reference point with eye-dropper icons. When hunting for the lightest and darkest points of your image, watch the brightness values in the info area for R, G, and B (Palette: Info). Once the highest values have been obtained, click the eye-dropper for the white point. Do the corresponding actions for the black point. This procedure works fairly well. For examples, see above under scanning.

If this should not work, try to adjust the curves of the individual alpha channels. Above the curves graph is a pull-down menu; initially "RGB" is selected. If your image is too blue, choose "B" for the blue alpha channel and pull the endpoint of blue down; Photoshop will give a preview of the altered image. If then the image is too green, choose "G" from the pull-down menue and lower the green curve. This procedure will require trial and error: go back and forth between the channels until the best possible result has been obtained.
Better results are obtained when lowering the excess color, rather then increasing the missing color. The latter has the tendency to blow out highlights. In general, do not try to change the shape of the curve, leave it as a straight line (keep the variables at a minimum).

After the color cast has been corrected, re-check levels.

Curves is superior to Variations as well as Color Correction, because Curves affects the image over the entire range of brightness values, whereas in Variation/Color Correction, the adjustments have to be done for highlights, midtones, and shadows independently. If you are comfortable using one way of adjusting color and get good results, there is no need to change your ways.

Usually, color saturation has to be increased. Often, a preliminary adjustment is carried out in the scanner software. In Photoshop, use Hue, Saturation (command-key U), and slide the saturation ruler to the right until a pleasing image appears. [Alternatively, in LAB color space, increase the slope of the a and b curves, while keeping the curve run through 0,0.]

6) Local adjustment of contrast/tonality and color.
There may be some areas that are overly bright, but global changes cannot adjust the problem (such as a, hot spot on a SEM image: below). Select the desired area with an appropriate tool, usually the magic wand. The range of tones selected by the magic wand can be adjusted in Windows: Options. Select the magic wand, then check the value set in options. Good results are obtained with values between 10 and 40 (top image below: magic wand selected, tolerance 40). Add additional areas by clicking the magic wand again while holding down the shift key.

IMPORTANT: before adjusting the contrast/tonality or color of the selected area, it must be feathered, which makes the margins of the selected area softer. If this is not done, unacceptable tonal breaks will become apparent. Under select, choose feather. Feathering can be made at values from 0.5 – >100: this determines the linear extent over which the softening of the effect will occur. 0.5 = a very sharp change, 50 = a very gradual change (top image below: feather radius 5 selected; notice smoother margins of selection in bottom image). The best setting is a matter of experimentation and experience, and depends on the particular image.

With the feathered selection, make the desired adjustments: levels, curves, saturation, ... (bottom image below).

In Photoshop CS1-6, one can also use Image-Adjustments-Shadow/Highlight .... In the dialog box, click on "Show more options". In the Shadows and the Highlights box, move the Amount and Tonal width sliders all to the left. Adjust Radius to 10–30 pixels as usually good starting values. In Highlights, move the amount slider to 70–100%, then start sliding the Tonal width triangle until the highlights are no longer blown out. Adjust Amount and Radius to check for improved results. Do the same for Shadows.

7) Cut outs.
Often multiple specimens/views are assembled on one plate, and the specimens can be cut out from the background. When cutting out images, you have to do the cutting in any but the bottom layer. [If you erase unwanted parts in the bottom-most layer, the selected background color is filled in, rather than transparency.] You may have to duplicate the image layer of the original file (Layers: duplicate). Cutting out can be accomplished in a variety of ways, including magnetic lasso, magic wand, or eraser. With all selections, before removing background, feather the selection, usually fairly narrowly with tollerance = 0.5. For the eraser, feather the eraser brush at least somewhat. Larger brushes need less feathering than smaller ones. Adjust spacing of the brushes to 5% (default is 25% and is too wide resulting in irregular edges).

8) Assembling plates.
Multiple images can be combined to a plate. Before combining elements into a plate, make most of the adjustments in the separate files: cleaning, levels - curves, color correction. Usually, the individual elements will be somewhat scaled in the composite plate. This scaling with improve the tonality of each element = smooth the levels histogram.

Copy - Paste or drag elements into the target file; the elements will automatically be placed into new layers. If you need to re-size elements, select all (command-key A), choose Image-Transform-Scale, and hold down the shift key while dragging one of the corners until the desired size has been obtained. The shift key action will force the image to be modified proportionally, thus avoids distortion of the image.

Use separate layers for scale bars, separation bars, and lettering. For complex plates, you can group layers into directories/folders, e.g., folder with images, folder with part letters, folders with arrows, etc.

Layered files can be saved in Photoshop format (.psd) or .tif-files. These are lossless formats, hence are suitable for further image manipulation. Keep a copy of the layered file as the master file at very least until the manuscript has been published.

9) Remove excess borders.
Crop image so that any unwanted borders are no longer visible. There is no need to keep a white border around the image, but a black border should be maintained if desired. Crop with the cropping tool, or use Image: Canvas size.

FILE PREPARATION AND MANIPULATION FOR LINE ART

10) Scanned images.
For basic scan parameters see above, section 2. The source image must be of good quality: lines solid black, not smudged; precise stippling; even and deliberate cross hatching if used. Images should be drawn significantly larger than they will appear in print, 1.5 – 2 times larger usually is sufficient. Quality pens should be used.

Images need to be cleaned as for half-tone images, though dust and scratches filter cannot be applied to bitmap images. Levels and curves do not apply. Composite figures can be produced as for half-tone images, and labeling is added as needed.

11) Vector graphic files.
Line art generated in vector graphics programs (Illustrator, Powerpoint, etc.) must be submitted as .tif or .bmp files. When you generate line art in vector graphics programs, you do not need to worry about resolution, as the geometric shapes are stored as mathematical expressions, not as pixel images. HOWEVER, once you transform them into pixelated images, then resolution of the pixelated images must be considered. The pathways to transform a vector graphics file into a pixelated file vary significantly from program to program: some will allow saving the vector graphics file as a .tif or .bmp file, with others, the file is saved-as an encapsulated postscript (EPS) file that is imported into Photoshop and saved appropriately. At some point(s) of the transformation series, there will be options that will affect resolution. As there are too many different variations, no further details are provided here.

If a line-art tif file is too low in resolution, DO NOT UPSAMPLE. Start over from the original vector graphics file, and re-export at higher resolution.

SUBMISSION FILES

12) Pre-submission check of final files.
Duplicate the master files into a separate folder (final files). Adjust the file size using Image: Image size, with resampling option box checked. Image width should be 17 cm with 300 dpi resolution. The resampling should REDUCE the image size, not increase it. If an increase should be necessary, you have scanned your image at too low a resolution, or not up-sampled your digital capture sufficiently before further file preparation. Go back to step 2 and start over.

Recheck Levels and sharpening for each image layer. This should not need any adjustments. Note that the histograms are smooth now, whereas before they may have been jagged. IMPORTANT. Adjust levels/curves BEFORE reducing file size, otherwise jagged levels will become apparent.

Layered files should be flattened (Layers: Flatten image) and saved as TIF files. To make electronic transmission easier, files MAY be transmitted as LZW-compressed TIF files, or as highest quality jpegs. IMPORTANT. Although jpeg file format is permissible to ease electronic transmission, DO NOT MAKE ANY IMAGE ADJUSTMENTS IN JPEGS. The only permissible jpeg-function is "save as jpeg: maximum quality".

13) "Sharpening" - Unsharp Mask.
Most images will require some sharpening. Shapening increases the contrast between adjacent areas of strong differences in brightness, hence emphasizes borders. DO NOT USE "SHARPEN" FILTER'; USE "UNSHARP MASK". Unsharp mask allows more flexibility in the amount of sharpening. It has three parameters: diameter, threshold, amount:

Unsharp mask can be carried out before or after downsampling to 17 cm @ 300 dpi. If done before, the levels histogram will be very smooth in the final image, though the image becomes slightly softer after downsampling. If done afterwards, the levels histogram becomes slightly more irregular, but the desired sharpness level is retained. [For color images, unsharp-masking in the l-channel of LAB color space often produces better results with no color artifacts.]

NikSharpener is a good utility/Photoshop-plug-in for output sharpening.

For details on digital sharpening, see Fraser, B. & Schewe, J. 2010. Image Sharpening with Adobe Photoshop, Camera Raw, and Lightroom. Second Edition. Peachpit Press, Berkeley. 345 pp.

14) Submission files.
Duplicate all your final image files, and place them into a "submission" directory (third copy of images). Open the images, and reduce image size (Image: Image size) to 72 dpi at the original size (maximum width = 17 cm), with the resampling check box checked. Save images as good medium quality jpegs (quality 6 of 10). PLEASE save jpegs in B&W/RGB mode; although your system may be able to handle CMYK jpegs, many cannot do so, hence doing this may increase the review period of your manuscript.

PRINT IMAGE FILES

15) You will receive further instructions once your ms is close to being accepted for publication.

GLOSSARY