Home
Image Tools Group

 

 

Part 2.

Other parts:
Nowadays there are two modifications of the GIF-files: GIF87a (GIF87) and GIF89a, developed 1987 and 1989 respectively. The ultimate is the GIF87a enhancement.
By 1987 the following features of the GIF87a modification were defined: The GIF89a modification increased the opportunities of screening control, by addition of the following parameters: Basically the GIF89a differs from its precursor GIF87a by the ability to relay the set of the graphic images on the run. The official documentation on GIF89a is dated on July, 31, 1990. Every GIF89a standard deviation is considered illegal. Just because of this a viewer sometimes can not recognize the file of the seeming to be widely known GIF-format.
The GIF89a format, which for the first time was supported only by Netscape 2.0 browser, made it available to use animation in Web. You can create a sequence of images in GIF89a that are also used as frames of the animation clip. The image animation is based on the one-by-one image screening. This format also allows setting the screening time for each frame and the number of iterations. Nowadays the GIF-animation is one of the most popular ways to create a simple Web-pages animation, because all widely known browsers support this format. If the browser cannot recognize this format, then it just screens the first frame of the image.
Let's look through the main GIF-files' characteristics:

Color
The color of each image point is defined as an index in color table (palette) stored separately from the graphic data. Better to say that the image data item for a certain point contains not a color but address for the color table to define color.


This image consists of 5 pixels with the following indexes: 3, 1, 1, 0 and 2, that point to the color table defining the RGB-value for these indexes.
Depending on the image coloration the amount of the colors in the color palette may vary from 2 to 256. The color indexes in a shorter color palette will take less volume respectively - 1 bit for 2 colors, 5 bits for 32, 8 - for 256. The original colors, for some reason missing in the color palette, may be substituted with the closest available color, or be emulated with the closest colors bitmap diffusion. The size of the color palette and the number of its colors are choice of a user - the appearance and neatness of the image is his own case.
The color palette depth has a great influence on the destination file size. The reduction of the color depth of a GIF-file leads to the reduction of its size and the loading speed increasing.

Diffusion
Diffusion allows creating a different colors blend or a missing color mapping by mixing different color pixels. Diffusion together with the depth of the color palette influences the GIF-file size.
The picture shows how to imitate purple, using just red and blue.


The pixel mixing takes place only where it is necessary, thus the needed effect is provided.
There may appear spot and stripes, making the image sharpless and artefactial when blending: the Web-designer should carefully use "safe" color palette containing 216 colors that is used both in Macintosh and Windows systems. If the GIF-image, being screened in the 8-bit system, contains a color missing in available color palette, the blending is going to take place.
The diffusion increases the file size, but very often it is the only way to more or less adequate original color palette screening after the reduction. In other words the diffusion allows to a greater extent to cut the depth of the GIF-file color palette, thus providing its "lightening".

Compression
The GIF-format uses the same compression algorithms as ordinary archive programs (ZIP, for example), that's why no data loss happens during writing and reading GIF-images. But unlike archivers, GIF-files are packed and unpacked automatically. GIF-format uses LZW-compression (after Lempel-Ziv-Walsh) that allows to effectively compress files, that contain a good amount of homogeneous fill (logos, inscriptions, schemes).
The GIF-files compression algorithm calculates the number of the equally colored points following sequentially in horizontal, and stores the reference for the number and color of those points, but not the each point reference. Nevertheless the compression algorithm for the pattern is not going to be effective, because every next point is different.
When creating an image one should take into consideration the following features of the LZW-compression. The graphic data compression degree in GIF-format depends not just the data's repeatability and predictability (the monochrome image comes with less size, than confusedly "noised"), but also on the direction of the image, because the image scanning goes line-by-line. The picture shows the number of bit in GIF-files of the 100x100-pixel image in comparison.

Interlacement
There are two different types of data storage inside the GIF-file. One of them is to store the image lines one after another from the first to the last one (a noninterlaced storage). The other is to store the image lines in a certain order (the interlaced storage). For the second type the storage order is fixed. The lines are stored in 4 passes:

The type of an image's storage is a file parameter and is set when creating or editing the file. Depending on the image storage type the screening comes different - the image is either rolled out, or it may become shown step-by-step and is enhanced from pass to pass. The file visuazlization in process of image receiving is typical for the WWW-browsers. This feature specifies an obvious connection between the storage scheme and the screening process.
Thus a 1/8th of the image is screened in the first pass. Every next step increases the amount of the lines filled, thus giving an original appearance to the image in four passes. It should be expected that after the first pass there's going to be seen only every 8th line, while others will be covered by the background image. Nevertheless, the browsers when screening an image provide interpolation by pixels with a certain value on each step. This approach allows imaging the image's outline fast enough. When getting the upcoming data, the separate line's pixels change their color, thus giving an effect of slow image showing.
This feature is widely used in Internet. First you can see a picture with a coarse resolution, but when getting new data the quality of an image gets better. The opportunity of the interlaced step-by-step screening is very popular among Web-designers in Web, where the loading speed is often a sensitive point.

Transparency
One of the basic advantages of GIF89a format in comparison with other graphic formats is that it supports transparency. One or more colors may be set transparent in GIF-format, thus staying hidden in Web-browser and some other applications. This GIF-format feature is widely used in Internet, especially when there's a necessity to make the Web-page's background be seen through some parts of the image. The transparency is provided by the additional Alpha-channel, that is stored together with the file.
Nevertheless the transparency in GIF-file is far from perfect. In certain cases there may appear a ghost all over the image. For example, smoothing tool is used for creating the blends. In smoothed images the blend comes really smooth, because the boundaries between the image and the background are smoothed, and the smoothed colors make a ghost or noise over the image as a result.

The delay period
The delay period is a period of time between the previous frame screening completion and the next frame output start. Before the next frame output start, the recovery of the screen area, covered by the image is carried out, in accordance with the mode settings.

The user's entry waiting mode
The user's entry waiting mode enables to interactively manipulate the frame changing while playback. The delay period and the entry-waiting mode may be set for a certain frame. At the same time the frame changing is executed when either of the following occurs - the end of the delay period or the user's entry. Browsers never support this mode.

Commentaries
The commentary addition to the GIF-file is used for the text interpretation of the image and it's not the part of the image. The size of the commentary is not restricted, but the too large number of commentaries leads to the file size increasing.

The text lines addition to a file
The text lines addition to a file enables the text visualization in a simple form of a graphic image fragment. For that the symbol mesh of cells, which is defined by the parameters in this block of a file, is used. The text data is submitted as fixed-pitch symbols - one symbol per cell, in the most appropriate font and size. The data is submitted unless the end of data is reached or the mesh of cells is filled. Browsers never support this mode.

The screen area recovery mode
The screen area recovery mode sets the way of handling the graphic fragment after its screening. The following modes are defined:

The inside GIF-files entries, specific for a certain application
The inside GIF-files entries, specific for a certain application allows to almost infinitely extend the number of the opportunities upon working with this format. One GIF-file may contain several different blocks for different applications. It is possible to create a block, ignored by all applications except one it was intended for. For example the Netscape browser uses such a block for the image changing loop.

Image Tools Group 2004.

Home / Previous part / Next part / Back to top