|
In Batch Mode, the Processing and Downscale options are on the Batch Settings tab. In Watch Mode, these options belong to the watch folder properties. Double click on a folder at the Watch Selection List to see the options.
The Process group contains the image processing options. Currently you can only process grayscale and color images. You can request any of the following filters:
| • | Smooth image: Make the images smoother. You specify the amount from 1 to 255. The reason you want to make images smoother is the output file size. Every sharp dot or edge in the image increases the output file size. It is not necessarily true that smoothing improves the image quality, but it could. Generally it makes the quality worse by making the image blurry, so be careful with this option. A slight smooth actually can improve the quality of grainy images with significantly reducing the file size. Clear images can not be improved, this will only make them blurry. Smoothing always reduces the amount of details in the image. |
| • | Adjust auto levels: This is an automatic histogram-based level adjusting filter, which usually works well for documents where the background should be white and the text should be black. If your scanned images have a gray background or gray text, then this is exactly what this filter was created for. It automatically adjusts the black-white balance, making the image more natural. It also works for many photos, but not all of them. For example, if you photograph a black train, your image will be dark, but that is normal, you want it to be like that. Similarly, if you take a photo of a snowy landscape, you want it to be very white. In both cases, this filter will make the image gray, and that is not what you want. Remember, this filter is automatic in terms of doing mathematical calculation and averaging colors, not in terms of being intelligent. |
| • | Convert to bitonal: This is a filter that converts grayscale or color image to black & white. The black & white output will use two colors only, 100% black and 100% white, no shades of gray. The algorithm automatically calculates a value of threshold and considers every pixel darker than the threshold black, and every pixel lighter than the threshold white. Even if your image has a gray background and dark gray text, the output image will have a perfect black text in white background, unless the input image has almost no contrast and it is impossible to separate the text from the background. For the best quality, do not mix Adjust auto levels and Convert to bitonal, because the combination will probably be worse than what you expect. Adjust auto levels is an automatic brightness correction, which Convert to bitonal is an automatic contrast correction, and often the two things do not mix well. Convert to bitonal was engineered to work well even if the background is dark gray and the text is black, or if the background is light gray and the text is dark gray, and it never works well with photos or halftones, no matter what you do. Convert to bitonal drastically reduces the output file size. |
The Downscale Options group contains the downscale-related settings. Check the Downscale images box if you want to perform a downscale, uncheck otherwise. Downscale can significantly reduce the output file size by reducing the resolution of the image, thus reducing the amount of details. This feature is especially good for screen reading, such as CD-ROM and Web publications, when the images will not go to a professional printing shop. If you turn on downscaling, you have the following options:
| • | Downscale images: Choose To resolution or By percentage. |
| o | To resolution downscales every image to a specified resolution, unless the image already has a lower resolution than that. Choose this if you have a specific resolution requirement in your mind. For example, if you want to downscale 600, 400 and 300 DPI images to 200 DPI alike. |
| o | By percentage, in contrast, always downscales by a given factor. Choose this option if you want to resize all images by a certain factor to save output file size. For example, you can downscale every image to half size.
Note that downscale never changes the physical (inch or millimeter size) of the image, it only reduces the pixel size, thus the number of pixels per inch (DPI value, resolution). |
| • | B&W images: Specify the downscale resolution or factor for B&W images. You have a choice to downscale B&W images less aggressively as grayscale and color images. In B&W images, you only have black and white pixels, do no color shades. In grayscale images, you have at least 256 different shades that carry a lot of information alone. So downscaling a grayscale image will not be as clearly visible as downscaling a B&W image. A B&W image becomes unreadable below 200 DPI and unprintable (in professional quality) below 300 DPI. On the other hand, grayscale images are readable even at 72 DPI on the screen, and look extremely good at 200 DPI. Color images are even better, because the color code carries a lot of information as well, so a digital photo can be significantly downscaled for screen reading. |
| • | Gray/color: For the above mentioned reasons you might want to use a higher downscale factor or lower resolution for grayscale and color images than for B&W images. |
| • | Anti-alias B&W images: Anti aliasing is an interpolation that makes B&W images lower resolution grayscale. Anti aliasing only works for B&W images, you can not use them for grayscale. Note that anti aliasing increases the output file size drastically, and therefore it is not recommended if you thrive for small output files. |
| • | Interpolate gray/color images: Interpolation is currently only available for grayscale and color images. Interpolation makes the image quality much better, without increasing the output file size, but with the price of a very computing intensive process that slows down the conversion significantly. If there is no interpolation, downscale simply throws away pixels. If you downscale to the half, every other pixel will be tossed away. If you had a dot there, you will lose it. In general, you lose a tremendous amount of detail during a non-interpolated resize. On the other hand, if you downscale with interpolation, nothing will be thrown away, pixels will just get smaller or lighter, but they will not disappear, unless they were extremely small and light on the input image. With downscale you always lose details, no question about that, but without interpolation you lose a great deal of detail, you can even lose the periods at the end of the sentences; also, the smooth edges and lines in your input images will become extremely jagged and pixel-like. With an interpolated downscale, you lose a minimal amount of detail only, not a single sharp dot will be lost, and smooth edges and lines in your input images will remain smooth, actually even smoother than before. It also means that the image will be darker and more blurry after a resize, but that is not noticeable unless the downscale goes to the extreme, something like 10 times or more. |
Notes: Downscaling and hidden text work perfectly together. Moreover, we take extra caution to do the OCR before the downscale, so the OCR be performed on the high resolution source image, even if the output is a very low resolution image.
Note that every processing and downscale makes the quality of very good quality input images worse. Some image processing can improve bad quality input images significantly. All image processing and downscale alters the image and its content in a way that might or might not be acceptable for you.
|
ELAN GMK Twitter feed
