Image capture problems
Moire (more-ay) patterns are undesired interference artifacts produced by various digital imaging techniques. The effect is caused by the interference of two or more regular structures with different spatial frequencies most often grids or parallel lines. In scanning moire is created through the relationship between the frequency of the dots used to make up the halftone printing process and that of the pixels that make up the sensor. Many scanners come with descreen options that reduce moire when scanning and there are also several manual processes in Photoshop such as scanning oversize, applying a median noise filter, reducing the image then suing Unsharp mask and occasionally gassian blur.
Moire can also occur in digital photography if fine detail on the subject interferes with grid of pixels on the sensor. This is however much more unlikely as most modern cameras feature built in anti aliasing filters that digitally blur the recieved image to match the resolution of the sensor.
Pixelation is another problem when dealing with digital capture. It results when a bitmapped image is enlarged leading to the induvidual pixels that make up the image becoming visable. Compared to print the resolution of a computer monitor is rather low and this leads to comparitely large pixels, this can cause pixelation to be visible along diagonal lines or curves and is particularly noticable when using fonts. Pixelation of this kind can be disguised using several techniques. Anti-aliasing employs partial opacity to emulate smooth curves causing the pixels along a diagonal border to be filled with a color that is in between the color of the letters and the color of the background, tricking the eye into percieving a smooth curve.
Other techniques that have more specific applications are hinting, used to embed interpolation data in certain fonts and sub pixel rendering that is used in colour LCD displays.
Reolution in digital capture relates to the number of pixels that make up an image. Pixel dimentions of an image directly relect on the files size and the maximum size the image can be displayed with exibiting problems with pixelation. For example if we have an image where the dimention along one axis is 3000 pixels, these pixels can be output at 10 inches at 300 ppi (pixels per inch) or at 12 inches at 250 ppi. There is a reciprocal relationship between pixel size, the physical dimentions and resolution that can be expressed as follows:
Number of pixels = phsical dimention x ppi
As the maximum ppi for images on the web is 72 ppi the obvious problem is that images captured at high ppi will appear extremely large when displayed on a monitor. They will also have an extremely large file size and will therefore take an age to load. Photoshop has several interpolation options that can be used when resizing, each approximate values for new pixels generated in a different way.
Nearest neigbour resampling is extremely fast however the results can often be of a low quality. The algorithm looks at a pixel and copies its value. It is often used for preserving the readability of screenshots.
Bilinear sets the value of each pixel by looking at the neighbouring horizontal and vertical pixels. It provides a better result than nearest neighbour but has mostly been replaced by bicubic options.
Bicubic interpolation reads the values of neighbouring pixels horizontaly, vertically and diagonally then calculates a weighted approximation for the new pixel value. Photoshop has two type of bicubic interpolation smoother and sharper, as a general rule smoother is better for upscaling while sharper reduces images more acurately.
Colour casts are (sometimes) unwanted tints caused by capturing images in lighting conditions of different colour temperatures. Daylight at mid day (5,000K) can produce a bluish cast while an incandecent bulb (2,700K – 3,300K approx) can produce an orange tint. Sometimes these casts can contribute to the atmosphere of an image but on other occasions they will need to be corrected. A major advantage of shooting in RAW is that white balance corrections can be made easily before an image is opened for editing in photoshop. When dealing with other file formats colour casts can be removed in photoshop by the use of either photo filter adjustment, curves, levels and auto colour.
When saving work there are several formats available some are more suited for web production and some are not supported by web application at all. Thery provide varying methods of compression and result in different file sizes relative to the uncompressed original. The available colour depth and inclusion of alpha channel information can also be affected by the choice of file type.
The JPEG (Joint Photographic Experts Group) file format provides the greatest compression of the standards disscussed here. At its highest setting with barely noticable image degredation an 18MB, 10 x 8 file at 300 dpi can be reduced to around 1 MB. JPEG achieves this remarkable compression by using “lossy” compression, breaking a continous tone down into 8 x 8 pixel mosiac structures. The heavier the compression more of the image becomes irrevocably degraded.
A comparison of JPEG qualities with high on the right, low on the left
That said even on medium settings JPEG images can be more than adequate for web application and result in images that can be loaded extremely quickly. JPEG compression is better suited to images with soft tonal transition, highly detailed areas sometimes do not compress so efficiently and this can result in visable JPEG “artifacts”. JPEG creates 24 bit RGB images meaning that each value of red, green and blue is represented by a byte or 8 bits of data. Alpha channel information is not stored in JPEG files.
GIF (Graphic interchange format) files use indexed colour. This is an 8 bit colour display mode where specific colours are indexed to one of 256 numeric values. As such it normally used for more graphic images such as logos or diagrams with sharp lines but a limited colour palette. Within its limited colour range however GIF uses lossless compression that reproduces colours without degrading image quality. GIF also supports animation, animated gifs have been used for some time to provide logos or avatars on the web and have recently been applied to art photography in the form of “Cinemagraphs”.
An animated "cinemagraph" courtesy of iwdrm.com
TIFF (Tagged Image File Format) files are characterised by the inclusion of header tags that can specify the geometry of the image. These can include information on size definition, copywrite information or pretty much anything through the use of custom or “private” tags. The main advantage of the format is its flexibility TIFF images can support bilevel (black and white), gray scale, 24 bit RGB true colour or 32 bit CMYK images amongst others. They can also be compressed or uncompressed using either lossy formats such as JPEG or lossless like the Lempel Ziv Welch (LZW) format or a mixture of the two. It can also hold both bitmap and vector information. The downside to all this flexibility is a large filesize (even with JPEG compression TIFF files are significantly larger than straight JPEG files) and that with all the possible variations of TIFF file very few web browsers can display them without plug ins.
The PICT file format was developed by Apple in 1984 and are primarily used for exchanging graphics between various Mac applications. PICT files are encoded in QuickDraw and contain all the commands needed to create the image (Font, Size, Style, Type,Lines, Circles, Bitmaps). The format can be used for vector or bitmap images and can support line art, greyscale or RGB data. This RGB data can 4 bit to 24 bit colour with 32 bit images featuring an alpha channel. With Mac OSX Apple replaced PICT with PDF and as of 2009 Adobe dropped support for the format in photoshop diminishing its used image production on the web.
The BMP file format is a standard file format for use in the Windows OS. The format was developed by Microsoft to allow windows to display the finle onany kind of display device. The file always cantains RGB data but this can be anything from 1 (monochrome) bit to 24 bit RGB with 32 bits including an alpha channel. Like TIFF files BMP files have a header that contains information about size, type and layout but unlike TIFF files the type of information contained is fixed. This makes the structure of a BMP more straight forward however a BMP file cannot represent CMYK data or make use of such a versatile range of compression formats and this has resulted in it losing out to TIFF in pre press applications. BMP’s can be compressed using run length encoded formats but this type of compression does not work well with complex images and continuous tones. BMP files are therefore less suitable for complex images on the web with its emphasis on small file sizes and quick loading.
PNG (portable network graphic) is a file format originally designed to replace the GIF format. PNG’s support palate based images in both 24 and 32 bit colour, greyscale and RGB images. PNG’s can also store transparency data, an induvidual pixel can have its value set to transparent or alpha channel data can be recorded. PNG’s are structured as having a header which identifies the format and determines some of its characteristics and also both critical and ancillary “Chunks”. The header is actually contained in one of these critical chunks, others contain the pallet, the image itself (sometimes spread across multiple chunks) and markers designating the end of the file. Ancillary chunks can contain information on background colour, transparency, pixel dimension, the last time the file was edited etc. Compression is achieved in a two stage system that first filters each line of the image using a chosen algorithm then passing the result to a DEFLATE lossless compression system that eliminates reproduced strings and shortens commonly used symbols. All in all PNG is a smaller, faster loading more versatile format than GIF the only drawback being that it doesn’t support animation without the use of unofficial extensions.
EPS (Encapsulated postscript) files are able to contain a wide variety of bitmap vector and layout information. This information is encoded using the PostScript page description language. EPS files are often used to transfer information regarding entire page layouts across operating systems. They only describe the layout of one page and the information is structured in a way that allows an importing application to discover the size and shape of the page described in the file. An importing application needs to know the size and shape of the page in order to position the page properly and in order to render the page at the proper size. Many Adobe products can save layout information as EPS with all image context data intact and this format is often used when transferring material to print manufacturers who may not use the same Adobe software. It must be noted that EPS files don’t support transparent constructs so any image saved as an EPS will be flattened.