Merciful 1

home *** CD-ROM | disk | FTP | other *** search

/ Merciful 1 / Merciful - Disc 1.iso / software / l / lsd_sprint_docs_disks / lsdsprintdoc'sdisk07.dms / in.adf / ImageLink.doc.pp / ImageLink.doc

Wrap

Text File | 1990-09-07 | 58.9 KB | 1,463 lines

ImageLink Contents Acknowledgements..................................................... i Contents............................................................ ii About This Manual.................................................... 1 Getting Started...................................................... 3 Overview..................................................... 3 The Main Window.............................................. 3 Selecting Modules............................................ 4 Standard Operation................................................... 7 Linking Modules.............................................. 7 The Conversion Process....................................... 7 The File Requester........................................... 8 Starting A Conversion....................................... 10 Color....................................................... 10 Dithering................................................... 11 Compression................................................. 12 Scaling..................................................... 12 Image Geometry.............................................. 13 Image Bounds................................................ 14 Aspect Ratios............................................... 14 Scaling Methods............................................. 14 Module Status............................................... 16 Aborting A Conversion....................................... 16 Advanced Techniques................................................. 19 The ARexx Interface......................................... 19 Quiet Mode.................................................. 19 Command Summary............................................. 20 Sample ARexx Script......................................... 22 ii Single Frame Recording...................................... 24 MicroIllusions Transport Controller......................... 25 Other Uses For ARexx........................................ 26 ARexx Error Codes........................................... 26 Tips Tricks and Technical Info...................................... 27 Pixels, BitMaps and ColorMaps............................... 27 BitMap Organization......................................... 29 Color Reduction............................................. 31 Color Quantization.......................................... 31 Dithering................................................... 32 Scaling and Aspect Ratios................................... 32 ARexx Scaling Syntax........................................ 33 Working With Large Images................................... 34 Storage Issues.............................................. 34 Other Operating System Formats.............................. 35 Appendices Module Detail and Discussion................................ A-1 Installation................................................ B-1 Error Messages.............................................. C-1 References.................................................. D-1 License and Warranty Registration........................... W-1 iii About This Manual This manual was composed entirely using Amiga personal computers. The cover art was rendered using Sculpt-Animate 4D from Byte by Byte, converted to IFF using ImageLink and color separated using Professional ScanLab from ASDG. The over and entire manual were typeset using Professional Page from Gold Disk. Each page has a border of 2 inches along the outer edge. This space contains section names for quick identification of material of interest, as well as icons and notes to emphasize and highlight the material in certain sections. The manual is divided into the following sections: 1. About This Manual 2. Getting Started 3. Standard Operation 4. Advanced Techniques 5. Tips, Tricks and Technical Info A. Module Detail and Discussion B. Installation C. Error Messages D. References W. License and Warranty Registration The "Getting Started" section provides an overview of the ImageLink user interface. Novice users should read this section first. The section "Standard Operation" describes the con- 1 version process from start to finish. "Advanced Techniques" describes how to write batch scripts in ARexx. The section "Tips, Tricks and Technical Info" provides further insight on optimizing the conversion process as well as what is really going on behind the scenes in ImageLink. This includes some technical information on the organization of various BitMap formats. Detailed descriptions of various modules are included as appendices. The ImageLink package comes with documentation for a base set of modules. Additional modules will include documentation for insertion in this section of the manual. Throughout this manual, the term "button" is used as a synonym for Intuition gadgets. Any text to be typed in by the user, including all script commands, is represented in a standard typewriter font, e.g.: [Type This]. The term "ghosted" refers to the state where a button cannot be selected and is obscured by a pattern of dots. Please take the time to read the license and warranty section in the back of the manual, and send in your warranty registration. 2 Getting Started Overview ImageLink is a powerful system for the conversion of multiple format bit-mapped images. To convert images an input and output format are selected. ImageLink links its special modules together to perform the necessary conversions. During the conversion, the number of colors in the image can be reduced or the image can be scaled to a new size. ImageLink provides the user with a number of flexible options for operation. Each image format has a corresponding input or output module within the ImageLink system. The ImageLink interface is designed to be both easy to use and expandable at any time through the addition of new modules. The Main Screen After ImageLink is run from either the WorkBench or the CLI, the main ImageLink screen appears (see Figure 1). The screen is divided vertically into two sections. These show the input and output modules available for selection (alphabetically sorted) and which are currently selected. ImageLink can 3 be configured from external ARexx scripts to start with default modules of your preference. This procedure is outlined later in the chapter on Advanced Techniques. If ImageLink is to be installed on a hard disk refer to Appendix B. Selecting Modules Once the ImageLink main screen is opened, modules for the input and output formats may be selected with the mouse. As new modules are selected, their names will appear in the boxes below the lists of available modules. The arrows and scroll bars can be used to scroll through the list of all available modules if there are more modules available than will fit within the display area. Once both an input and output module are selected, the button at the bottom left of the display, labeled "LINK", will change from its previous "ghosted" state. This indicates that two valid modules have been selected and the conversion process may begin. The program can be terminated by selecting the "EXIT" button. The basic ImageLink package contains modules for several conversions. More detailed descrip- 4 tions can be found at the end of this manual: Module Depth Width Height Type ------ ----- ----- ------ ---- Caligari Broadcast File 32² Unl Unl Input DigiView RGB File 21¹ Unl Unl Input Compuserve GIF File 8 Unl Unl Both IFF ILBM File 24 Unl Unl Both Macintosh PICT File 32² Unl Unl Both PC PaintBrush PCX File 8 Unl Unl Both Sculpt Direct 24 Unl Unl Input Sculpt Raw RGB File 24 Unl Unl Both Truevision TARGA File 32² Unl Unl Both Impulse Turbo Silver File 24 Unl Unl Both Note that "Width" is the maximum width image a module will either read in (input) or write out (output). For width and height, "Unl" means "Unlimited" (in theory - the practical image size limit in ImageLink is 32768 X 32768) and "Depth" reflects the maximum number of colors (as bits per pixel) a module will handle. "Type" indicates whether a module is provided for input, output or both input and output. Refer to the section on "Tips, Tricks and Technical Info" for more insight into these parameters. ¹DigiView RGB is promoted to 24-Bit format internally by ImageLink. ²The Caligari, PICT and Targa support 32-Bits of data, but only 24-Bits are significant for the display. 5 This page intentionally left blank. 6 Standard Operation Linking Modules When the input and output modules are selected, the conversion may be started. Selecting the button labeled "LINK" at the bottom left of the screen will change the display to the Module Action/Status Screen which is below in Figure 5: The display is divided into two sections describing the current status of the Input and Output modules. The word "Initialize" will appear in each section beside the label "Status:", indicating that the modules are being loaded by ImageLink. After a brief pause, the word "Idle" will appear. This indicates that the modules have been correctly loaded and are awaiting the command to call them into action. The selection of modules may now be changed by selecting the "UNLINK" button and repeating the selection process. The Conversion Process In the general case, converting images requires that the "START" button be selected. Certain modules described in later sections may actually initiate the conversion process through some other action (e.g. the Sculpt Direct Module) or 7 the process may be initiated via a script as described in the section on Advanced Techniques. The File Requester Certain modules will require a file name before proceeding with a conversion. The general convention is that modules which are "File" modules will require file names, and modules which are "Direct" modules will not. If a module requires a file name, a requester will appear filling most of the ImageLink screen, prompting for the selection of an appropriate file as shown below in figure 6³. A brief line of text will appear at the top of the requester indicating which module is making the request. For example, the requester in Figure 7 was generated by an Input module. If a module needs to read in or write out a file in the course of the conversion process, it will request a file name in this manner after it has been "Linked". The input module always makes it's request before the output module. ³The file requester used in ImageLink was written by Bruce Dawson and Colin Fox. 8 The box on the right contains a list of all available devices and volumns on the system. Selecting any one of these will cause the file requester to begin examining the files or directories contained on the selected volume or device. The file names will appear in the large box on the left. Files appear in the list as they are found on the disk. Selecting the scroll bar or arrows will sort the list alphabetically. Files will have a number to the right representing the size in bytes of the file. Directories will have the word "(dir)" to the right. Selecting a File Name File names may be selected with the mouse as soon as they appear. The selected name will appear in the box labeled "File". Selecting the button labeled "OK!" will confirm the selection. Double clicking on the file name will select and confirm the choice with one action. File names may also be entered by hand after selecting the "File" box and directly typing in the name of a file. Selecting the "OK!" button or simply pressing 'return' will confirm the selection. If the entire file name list is built, it may be viewed using the scroll bar and arrows on the right side of the list of file names. Selecting the "Parent" 9 button will go back to the parent of the directory currently being examined. Selecting another directory name will build a list of files and directories contained in that directory. Changing File Names After the file names have been selected, they may be changed by selecting the module file selection/status box which is labeled "Source:" for the input module and "Destination:" for the output module. Starting a Conversion Selecting the "START" button will begin the conversion. The "START" button will become "ghosted" and the "UNLINK" button will change to an "ABORT" button which can be used to interrupt the conversion. There may be a brief pause while certain input modules perform special housekeeping operations. Refer to the section "Module Detail and Discussion" for information on the behavior of particular modules. Conversion Color The Conversion screen will appear as shown below in figure 9. The left half of the display describes the process which will be used to convert the image and the right half describes the amount of color information available from, and to be used by the input and output modules. If an output module does not support a particular conversion method, the corresponding button under the heading "Conversion:" will be ghost- 10 ed. Possible conversion methods include Color Map, Color Value, and Grey Scale. Refer to the section on "Tips, Tricks and Technical Info" for more information regarding conversion methods. As different conversion methods are selected, the box on the right hand side will reflect a list of all possible color choices for the output module under that format. Selecting a color choice from this list will change the output colors appropriately. The input module is always a fixed color value based on the image type being converted. Selecting the "REVERT" button will reset all values to their initial defaults. Selecting "ABORT" will return ImageLink back to the Module Action/Status screen without performing the conversion. Dithering Dithering may be turned off by selecting the button labeled "None". Other dithering methods may be selected, but are not appropriate when converting from a lower to a higher number of colors or when converting between images of the same number of colors. Refer to the section on "Tips, Tricks and Technical Info" for more in- 11 sight into dithering and dithering methods. Compression If an output module supports selectable compression methods, they may be enabled using the buttons labeled "Method 1" and "Method 2". The default compression type is "None". If a module doesn't allow selectable compression, but creates compressed output by default, the button labeled "Method 1" will be highlighted. Refer to the section "Module Detail and Discussion" for information regarding compression methods supported by specific output modules. Scaling When all conversion methods have been chosen, selecting the "OK" button will confirm the choices and bring up the scaling screen as shown below in Figure 10. The left half of the display describes the image geometry and selected bounds of the input and output data. The right half describes the method used for scaling the image data as it is converted. The values reflect what default settings the input and output modules have chosen based on the information available to them (e.g. 12 the dimensions of the input data and the output format/device). Certain modules (most notably certain input modules such as the Sculpt RGB File input module) may not know the exact dimensions of the data they will be dealing with. In this case, they will either present values of zero for the width and height or automatically take on the default dimensions of the output. The user must assure that these values are valid before proceeding with the conversion. Selecting the "REVERT" button will reset all values to their initial defaults. Selecting "ABORT" will return ImageLink back to the Action/Status screen without performing the conversion. Image Geometry The fields on the left describe the actual ("Original") width and height of the input or output image, the selected bounds of the input and output image (also called the "device size"), and the pixel aspect ratio of the input and output image data. In certain cases, these fields may be changed by selecting them and typing in new values. If a module does not support a fixed size, or if a module's actual data size is undetermined (as it is with the Sculpt Raw RGB File), then the "Original" width and height may be changed by selecting the fields and entering new values. 13 Image Bounds The "Bounds" fields describe offsets into the input or output image or device. If a module does not support offsets, these fields will not be modifiable. If the output module supports offsets, the button on the right labeled "Center Output Image" will be enabled. Selecting this button will automatically calculate the required offsets for the output image to be "centered". Aspect Ratios Aspect ratios for both input and output may also be modified. Modules will place values here based on the best information available to them, but sometimes image files will "lie" about their proper aspect ratios. Therefore, this mechanism is provided to correct for improper aspect information. These aspect values are used for calculations used with the various scaling methods ImageLink supports. Refer to the section "Module Detail and Discussion" for more insight into correct aspect ratios for specific formats. Refer to the section "Tips, Tricks and Technical Info" for more insight into the meaning of pixel aspect ratios. Scaling Methods The different scaling methods supported by ImageLink include: Copy Input To Output: The default method (as supported in ImageLink Release 1.0). This says "Copy the input to the output, clipping the input 14 data to the bounds of the output, not accounting for any aspect ratio differences." Fill Output From Input: "Copy the input to the output, clipping to the bounds of the output, padding the input data to fill the output as necessary, not accounting for any aspect ratio differences." Resize Straight: "Rescale the input data to exactly match the bounds of the output, not accounting for any aspect ratio differences." Resize To Match Width: "Rescale the input data so that the height exactly matches the output, clipping or padding the height as necessary, not accounting for any aspect ratio differences." Resize To Match Height: "Rescale the input data so that the height exactly matches the output, clipping or padding the width as necessary, not accounting for any aspect ratio differences." Copy Width Aspect Adjust: "Copy the input to the output adjusting for any differing aspect ratios, clipping or padding the data as necessary." Resize With Aspect To Width: "Rescale the input data so that the width exactly matches the output, clipping or padding the height as neces- 15 sary, accounting for any aspect ratio differences." Resize With Aspect To Height: "Rescale the input date so that the height exactly matches the output, clipping or padding the width as necessary, accounting for any aspect ratio differences." Module Status After the scaling screen is confirmed by selecting "OK", the conversion process will begin. As the data is converted each module will display its status in the module action/status screen. "Waiting" indicates that the modules are active and waiting for data to be read or written. "Reading" indicates the input module is reading a buffer of data for the conversion. "Resizing" indicates that the image data is being scaled to a new size. "Histograms" indicates that color histograms are bing built for a color reduction and "Cutting" indicate that a color reduction is bing performed. The section "Tips, Tricks and Technical Info" provides more insight into color reduction. Aborting a Conversion "Converting" indicates that the data is being converted to the output module format. "Writing" indicates that the output module is writing the data out in converted format. Reading and writing (and possibly scaling) continue in paral- 16 lel until the entire image is converted. Any errors will halt the conversion and produce and error requester describing the error. The "ABORT" button will abort the conversion after the output module has completed the current write operation. Refer to Appendix C for a discussion of error conditions and messages. When the modules have completed their operation, their status will change back to "Idle". The process may be started again using the same file names by selecting the "START" button, or new file names may be entered by selecting either file name box. The "UNLINK" button terminates and unloads the modules. When this happens, the action/status screen disappears and is replaced by the ImageLink main screen. The modules selection process can now be repeated. Selecting the "EXIT" button will exit the program completely. 17 This page intentionally left blank. 18 Advanced Techniques The ARexx Interface Licenses of Bill Hawe's ARexx package have even more flexible control over ImageLink. Using ARexx, ImageLink may be controlled via external programs or externally generated scripts. Repetitive action may be avoided by using ARexx to set up initial defaults, completely drive the conversion process via scripts or from other applications. The script language is "case insensitive." All letters will be converted to lowe case before ImageLink performs any processing. Quiet Mode The script language can be used for "Batch Mode" processing. This allows the conversion of large amounts of data via unattended operation. To run completely unattended, without using any Intuition screen (and valuable CHIP memory), ImageLink can be run in a special "Batch" mode. CLI users should run ImageLink with the argument "QUIET", e.g.: 1> ImageLink QUIET The word "QUIET" can be entered in upper or lower case. When run in this fashion, ImageLink will run in the background without opening any screens. The only way to control or exit ImageLink at this point is via ARexx. To direct ARexx commands to ImageLink, use 19 the ARexx command sequence: options results; address ;ILINK' lock; if (rc ~=0) then do; say "Could not open ImageLink"; exit; end; address value result; This sets up a private communication channel between the ARexx program and ImageLink. No other script or application may control ImageLink while an ARexx script is running. The channel must be freed by issuing the "UNLOCK" command at the end of the script, or by "QUIT"ing ImageLink. Commands may now be sent to ImageLink via ARexx. The syntax for commands is as follows: Command Summary COLOR [CMAP | CVAL | GREY] [NUMBER] - Set number of output colors, reduction methods etc. INIT - Insure the initial state. Any transfer in progress (not initiated by an ARexx script) will be aborted, all modules will be unlinked etc. INPMOD / OUTMOD [NAME] - Select the 20 Input or Output module. INPFIL / OUTFIL [NAME] - Select the Input or Output file as appropriate. ISCALE / OSCALE [XSIZE, YSIZE, XOFFSET, YOFFSET, XLIMIT, YLIMIT, XASPECT, YASPECT] - Set Scaling factors, input/output bitmap sizes, regions, aspect ratios etc. LINK - Line the two selected modules together. LOCK - Attempt to lock the ImageLink communications channel. This must be the first command issued by an ARexx script and is not valid after the ARexx script has been started. SCALEM [COPYIO | FILLOI | RESZS | RESZW | RESZH | COPAS | RSZAW | RSZAH] - Set the scaling method. Refer to "Tips, Tricks and Technical Info" for details on the syntax. START - Start a conversion. All conversions under ARexx are synchronous in nature. They can only be aborted by an error during the process or by selecting the "ABORT" 21 button when the user interface is active. QUIT - Tell ImageLink to shut down. RECORD [number] - Signal the MicroIllusions Transport Controller to record "number" frames. UNLINK - Unlink the currently linked modules. UNLOCK - Release the communications channel for others to use. VERSION - returns the version number of the active ImageLink program in the format "type number" where "type" is the string "Release" and "number" is a string in the form "major#.minor#", e.g. 2.0. Sample ARexx Script A sample ARexx script with explanation in comments (/* */) follows: /* Turn on result code passing */ options results; /* Find and lock ImageLink */ address 'ILINK' lock; if (rc ~= 0) then do; say "Could not lock ImageLink"; exit; 22 end; /* result contains the private channel */ address value result; /* Get the version string */ version; say 'Version is 'result; /* Insure the initial state */ init; /* Select input / output modules */ inpmod "IFF ILBM File"; outmod "Targa Direct"; /* Link the two modules together */ link; /* Loop here for multiple images */ loop: /* The input module needs a file name */ inpfil "DH1:Pictures/Seat.24"; /* Don't dither the image */ dither none; /* User 24 bit-plane output */ color cval 24; 23 /* Do a straight copy */ scalem COPYIO; /* Start the conversion */ start; /* Single frame record here */ /* Loop back for additional files */ /* Unlink the modules when done */ unlink; /* Free the channel */ unlock /* Tell ImageLink to exit */ quit; /* Normal end of ARexx script */ exit; Single Frame Recording One of the great features of ARexx and the Amiga's multitasking environment is the flexibility it provides for controlling multiple processes. Because there are no tight restrictions on this environment, it becomes fairly straightforward to add new features to an existing application which were not previously possible in it's original design. One such feature in ImageLink is the ability to control single frame recorders 24 from within the conversion process. A single frame recorder allows the recording of frames in an animation to video tape one at a time. When dealing with images which are large and complex due to their color depth etc., most machines don't have the bandwidth to play back animations in real time. The ability to single frame record opens up new avenues of quality at the expense of instant feedback. In the previous ARexx example there is one comment field which reads "Single frame record here". The addition of a command which will signal a single frame controller program may be inserted at this point in the script. Since ImageLink has just completed converting an image, perhaps to some high quality frame buffer, this is the correct time to record the frame to tape and then loop to render the next frame to the tape and then loop to render the next frame and record it to tape. This process may continue until all the animation frames desired have been recorded to tape. MicroIllusions Transport Controller One system which ImageLink supports directly for single frame recording is the Transport Controller from MicroIllusions written by Mike Berro. Once the Transport Controller is running with the proper module for the VTR, the ImageLink ARexx command "RECORD [number]" can be used to signal Transport Controller 25 to record a "number" of frames. This will cause the Transport Controller to record the number of frames specified in the "number" argument. Other Users For ARexx Using ARexx, ImageLink may also be configured as an Image Conversion Server on a network of diverse imaging systems. Data can easily be piped through ImageLink to convert images into formats readable by Apollo, HP, IBM, Macintosh, Silicon Graphics, Sun and a variety of other workstations and PC's. There are a number of sample ARexx scripts provided on the ImageLink distribution disk in the "ARexx" directory. ARexx Error Codes ImageLink ARexx scripts can return the following error codes: 80 - LOCK: ImageLink already Locked. 81 - UNLOCK: Nothing to UnLock 82 - LINK: No Input Module Selected 83 - LINK: No Output Module Selected 84 - START: No Input File 85 - START: No Output File 96 - Transfer Already In Progress 97 - No Such Module Exits 98 - No Such File or Can't Create File 99 - RECORD: Problem with Transport Controller 26 Tips, Tricks and Technical Info BitMaps What is a bitmap? A bitmap (commonly referred to as a raster) is a series of locations in a computer's memory which define the bounds (width, height) and makeup (color depth etc.) of a display or image. The computer display hardware fetches information from this area in order to determine how to draw an image for its display. Pixels Every element in the bitmap can be represented by a Euclidean point (x,y). These elements, known as pixels, map directly to unique memory locations which contain the information related to their visual representation. Each pixel exists at x units to the right and y units down from the origin of the bitmap (0,0). Pixels have a third dimension known as their "depth" or color value. Color Depth is what determines the amount of different colors a pixel is capable of representing. The depth of a pixel is usually some power of 2, there 2n yields the total number of color values possible for the pixel (and usually the entire bitmap). So, 2(5) would yield 32 possible color values, 2(8) would yield 256 possible color values etc. 2(24) will yield 16 million possible colors, also known as "True Color". True color represents the maximum number of color values for each component. Red, Green and Blue (RGB) being 2(8) or 256 possible values each. 27 Color Value There are two ways to interpret the depth of a pixel to determine it's color. The first is to take the pixel value and separate it into RGB components to yield an absolute color value which specifies an intensity of Red, Green and Blue. This is easily done with a pixel whose depth represents a total number of bits which is evenly divisible by three (RGB). Some formats which aren't evenly divisible by three will sacrifice the color resolution of the lower frequency color spectrum (blue) and provide more bits of color depth to red and green. Color Map The second way to interpret depth is as an index into a table of color values, called a ColorMap. Each entry in the ColorMap will contain an RGB color value using the maximum color resolution available in the display hardware. This method is useful when the hardware can resolve more bits of color than it can display at any given time. The Amiga has a color resolution of 12-bits; 4-bits each for R, G and B. In normal display modes, at any given time the Amiga can only display 32 colors (because the hardware has only 32 total color registers). Bitmaps are built using pixels which are 5-bits deep (2(5) yields 32 colors) and every bitmap has a ColorMap associated with it. The ColorMap may contain 32 values out of a total range of 4096. Some systems allow the storage of images with 28 either color values or color maps. BitMap Organization Not all bitmaps are created alike. If they were, there wouldn't be a need for ImageLink. The two most common methods for storing the individual elements of a bitmap are known as the Chunky and Planar methods. Both formats have advantages and disadvantages. Chunky Organization In the case of Chunky, the bitmap is organized as consecutive storage locations in memory. Every location contains the complete Red Green and Blue (RGB) value for the pixel it represents. The computer display hardware will scan down this (contiguous) memory space as it draws each pixel consecutively from the top to the bottom of the display. it is very efficient for general purpose hardware to modify bitmaps stored in this fashion. A modification of a single pixel value will typically involve an operation on a single memory unit (byte, word or long). This operation is common to most popular processor families such as the Motorola 640x0 family found in the Amiga. There are cases when using general purpose hardware where operations on large regions of the bitmap which involve subtle modifications may become costly in CPU usage. This situation becomes more severe as the depth increases 29 since the hardware will have to scan down increasingly large regions of memory even if only modifying one component of color (R, G or B). Planar Organization In the case of Planar organization, every pixel in a bitmap is broken down into its component bits. Each component bit is placed in a separate region of memory known as a bit plane. Some machines such as the Amiga have specialized hardware that operate on planar bitmaps. The hardware maintains pointers to these separate locations and fetches the individual components from these memory regions to build a complete pixel. Each bit represents a binary number (0 or 1). The component bits are combined to form a number which represents either a real color value for the pixel or an index into a color map. When the number represents a color value, the bits are usually grouped in R, G, B sorted order. In this case, planar bitmaps lend themselves well to applications which need color components separated. Many printing and image processing applications require or benefit from having separated components. Having the components pre-separated by the nature of the bitmap organization can save significant processing time later. 30 Color Reduction Many systems cannot display true color images. They are often limited by memory or other hardware constraints. The Amiga, for example, is capable of displaying a maximum of 4096 colors using it's special "HAM" (Hold and Modify) technique. The fewer the colors a system can display, the less capable it is of displaying realistic images. When the number of colors in an image is reduced, there are two methods for reducing the color. The first requires little or no intelligence and simply throws away some range of colors in the image based on their physical location in the color palette. If one of those colors was a major component color of the image, the resulting image will be severely distorted from it's original representation. Color Quantization The second method is more intelligent and requires considerably more computation. The colors of the original image are examined and a histogram is constructed which describes the frequency and distribution of colors in the image. A set of colors is chosen for the target space which takes into account the importance of various colors in the original image. Paul Heckberts' Median Cut algorithm is used. This algorithm successively divides the largest color cubes until enough colors have been selected. This yields results which are far superior than simple color popularity alogrithms. 31 Dithering An additional technique known as dithering may be applied to an image when reducing colors to help provide a truer representation of the original. Dithering involves creating a pattern of two or more colors to approximate the appearance of another color which is not present in the palette. ImageLink uses a method known as Floyd-Steinberg dithering (named after it's two creators). This technique sacrifices spatial resolution for color resolution. When dithering is applied to an image, some loss of dot resolution may occur, but the resulting increase in color resolution increases the perceived quality of the image to the human eye. Scaling and Aspect Ratios There are other issues in converting between image formats besides file layout and number of colors. Individual pixels have an additional characteristic known as their aspect ratio, or shape. This shape is different depending upon the hardware being used to display the image. Ultimately, we wold like all pixels to be perfectly square (aspect ratio of 1:1), but this is not so. The Amiga is more complex in this regard than most other hardware platforms, since it's varying resolution modes for an NTSC compatible signal present a variety of possible aspect ratios for images. The Macintosh is a very simple model - all pixels are always 1:1. So an aspect ratio of 20:10 means that the image pixels are 32 twice as wide as they are high. If this is to be converted to an aspect of 1:1, we must either double the output pixels in the vertical dimension, or halve them in the horizontal dimension. ImageLink provides full feedback and control over image size and aspect ration. Some Amiga software has the capability to render image data at arbitrary resolutions and aspect ratios. The information provided by certain ImageLink modules can be used as advisory information for these software packages. For instance, the Macintosh modules will advise that their aspect ratio (for overscan video images) is 1:1. The Sculpt Direct input module can utilize this information to render at the correct aspect ratio for the Macintosh. ARexx Scaling Syntax The ARexx scaling syntax corresponds directly to the scaling methods presented in the Scaling screen: COPYIO Copy Input To Output FILLOI Fill Output From Input RESZS Resize Straight RESZW Resize To Match Width RESZH Resize To Match Height COPAS Copy Width Aspect Adjust RSZAW Resize With Aspect To Width RSZAH Resize With Aspect To Height 33 Working With Large Images Working with large images can be a problem when color reduction is involved since the entire image must be read into memory. One thing that can help matters is to first perform a resizing operation and then do a color reduction on the resized image. This can considerably reduce the demand for memory and achieve the desired results. For instance, if a 24-bit Macintosh image at 1280x1024 is to be converted to a 352x480 HAM image, nearly four megabytes of contiguous memory will be required to perform the color reduction. If the Mac image is first scaled to a 24-bit 352x480 IFF bitmap, only five hundred kilobytes will be required for the color reduction. Storage Issues Compress wherever possible. Consider storing images in compressed form for long term storage even if the ultimate format needs to be something unrelated to that format. The IFF format offers good compression ratios, especially for "computer generated" (e.g. ray-traced) images. Large images mean large files. Transporting these files to outside service bureaus can be difficult. There are many possibilities for overcoming this problem: * Split the original file 34 * Compress and archive the original file using a popular achieving format (e.g., ARC, LHARC, ZIP, ZOO). * Split and compress/archive the file Split/Combine utilities as well as archieving programs are available in the public domain or as shareware. Consult with a local user group or bulletin board for assistance in acquiring these utilities. Other Operating System Formats When working with foreign operating system formats, there are a number of products and utilities to aid in exchanging files: For reading and writing MS-DOS disks: Dos-2-Dos - Central Coast Software CrossDos - Consultron For reading and writing Macintosh disks: Mac-2-Dos - Central Coast Software These products are commercially available. Consult your local dealer for price and availability. Information can also be easily exchanged with other systems using popular communications software and file transfer protocols over modems or direct RS-232 lines. 35 Module Detail and Discussion This section contains detailed discussion on using the various input and output modules that come with ImageLink. As new modules are acquired, their documentation should be inserted here. New modules are installed into the system by running the install script provided on the module distribution diskette. Certain modules may require additional installation procedures which will be documented in the modules accompanying literature. A-1 The Caligari Module The Caligari module supports the 24-bit imaging capabilities of Octree Software's Caligari Broadcast Modeler. The module will read the ".6rn" files produced by the Rendition rendering software from Numerical Design Ltd. There is no restriction on image sizes. To create the proper file from Caligari, enter the "Brender" menu from the scene module. Choose the object attributes and enter the "Script" menu. Create a script and use the "Save" option to generate all the ".6rn" image files in the script directory. You may then use an ImageLink ARexx script to record the resulting frames to videotape or convert them to other formats. It is also possible to use ImageLink to transfer Caligari image data over to the Targa board as Caligari creates each image. Refer to the ARexx directory on the ImageLink distribution disk for a script with instructions for doing this. A-CAL-1 The DigiView Module The DigiView digitizer provides a low cost method for inputting images scanned with a video camera on the Amiga. While a slow process, the digitizer is capable of maintaining a color resolution of 21-Bits per pixel for a total color resolution of 2,097,152 colors. To maintain this color resolution in DigiView, you must use the "Save RGB" option when saving your image from the DigiView software, otherwise you will simply be saving out a standard Amiga IFF ILBM file with at most 4,096 colors. The image may then be brought into the ImageLink conversion system to be converted to other formats or to be previewed on a frame buffer, sent to another imaging device etc. Note that ImageLink promotes DigiView images internally to 24-Bit images for convenience. The information is simply padded out to 24-bits, with the original 21-bits being the only significant information used in the conversion process. Note on DigiView 4.0 DigiView 4.0 will produce 24-Bit IFF files which may be read in using the IFF ILBM File module. ImageLink will strip the proprietary DGVW chuck from the DigiView file. A-DGV-1 The GIF Module The Compuserve GIF format is a popular format for the interchange of images between a variety of PC's. Very large databases of images can be found on various electronic information services. The format is limited to a maximum of 256 colors and the GIF Module will read and write GIF images in 2, 4, 6, 8, 16, 32, 64, 128 or 256 colors. The input module will read images created in row-interleaved as well as non-row-interleaved format. When reading a row-interleaved image, there will be a brief pause when a conversion is started as the whole image must be read into memory. The GIF output module will write all images out in non-row-interleaved format. There are several public domain GIF viewers available for IBM, Macintosh, Atari and Amiga PC's. A-GIF-1 The IFF ILBM Module Overview The IFF ILBM (Interleaved BitMap) modules which are supplied with ImageLink will convert Amiga images for use on other systems or will convert other image formats to IFF for viewing or other use on the Amiga (color separation etc.) As either an input or an output module, the IFF ILBM module requires a file name. Traditionally, Amiga IFF images have been limited to 2, 4, 8, 16, 32, 64 or 4096 colors. The ImageLink module supports ILBM files with a full range from 2 to 16 million colors. The format used is as described in the 1989 Amiga Developers Conference notes as adopted by Commodore. HAM (4,096 Colors) images are promoted to 24-Bit images (16 Million Colors) internally by ImageLink. ImageLink will also read and write 256 color (8-bit) ILBM files for use on IBM machines and 8-bit Amiga frame buffers as well as 18-Bit IFF files created by the Sharp JX-100 color scanner. Amiga IFF supports a variety of aspect ratio formats depending on the Amiga display resolution and the original source of the image data. The following table provides suggested aspect ratio values for standard Amiga formats as well as for images created with particular Amiga packages: A-ILBM-1 Aspect Ratios for Standard Amiga Formats 320x200 10:11 320x400 20:11 640x200 5:11 640x400 10:11 Overscan values are the same (e.g. 704x480 is 10:11). Sharp Scanners create images with an aspect of 10:11. A-ILBM-2 The PC-PaintBrush (PCX) Module PC-Paintbrush is a popular 2D paint package from ZSoft Corporation for the IBM-PC. ImageLink supports both reading and writing of files stored in the ZSoft PCX format. The initial release of the PC-PaintBrush modules support reading and writing CGA, EGA, and Hercules (Monochrome) images. For color, this means the modules can read and write 2, 4, 8 and 16 color images. Future versions of this module will allow reading and writing VGA 256 color images. A-PCX-1 The PICT Module Overview The Macintosh uses an internal imaging model known as QuickDraw. QuickDraw produces image file in a format known as PICT. The format known as PICT2 introduced support for color and the latest QuickDraw (32-Bit QuickDraw) supports up to 16 million colors (true color). Caveats QuickDraw supports the imaging of structured graphics as well as bitmaps. These structured graphics are represented in PICT files as opcodes (instructions) for drawing lines, curves, arcs as well as text. The Amiga does not support a standard imaging model for structured graphics, nor does ImageLink. Any information in a PICT file which describes either text or other structured graphics is ignored by ImageLink, and only bitmap data is extracted for the conversion process. Conversion Issues There are several methods available for reading and writing Macintosh files on the Amiga. Mac-2-DOS from Central Coast Software allows a Macintosh floppy disk drive to be attached to the Amiga. The more traditional method of hooking the machines up via serial port or modem may also be used. When sending from the Macintosh, be sure to use standard XModem/Kermit etc. Binary transfer. In either case, you will have to be certain that the correct Re- A-PICT-1 source type is created on the Macintosh for the image file. A Resource of type "PICT" may be specified directly in Mac-2-Dos as the file is being converted, or you may change the type using the dialog box from "Get Info" on the Macintosh itself. PICT2 images are typically 2, 8, 256, 32768 or 16 million colors. The ImageLink PICT module will correctly handle 32-Bit QuickDraw images, which are images of either 32768 or 16 million colors. For 32-bit QuickDraw, the reader will read in images created in 32768 colors ("16-bit direct entry") or 16-million colors (either 24-bit or 32-bit). The writer will write out in only the 24-bit format. Both modules will handle 2 or 256 color images. All Macintosh images have an aspect ratio of 1:1 (square pixels). A-PICT-2 The Sculpt Direct Module Overview The Sculpt Direct module provides an interface directly to Byte-by-Byte's Sculpt-Animate series of 3D modeling and rendering software packages. This provides a near seamless integration of ImageLink to Sculpt-Animate, allowing Sculpt to take advantage of all the image conversion capabilities of ImageLink, including immediate support for any frame buffer (such as provided by the Targa Direct module) or other device for which there exists an ImageLink module. Installation The module is provided in the default ImageLink configuration, but may require some installation for a hard disk or customized floppy system. In addition to the module, a device driver is provided to interface with Sculpt-Animate, using the Byte-by-Byte developed Frame Buffer interface. The basic ImageLink installation script will attempt to copy the device driver to the currently assigned DEVS: directory. If you wish to copy it somewhere else, you may find it on your ImageLink disk as devs/ilink-fb.device. Rendering ImageLink must be running with the Sculpt Direct input module and any output module linked before starting Sculpt-Animate. Rendering a frame or series of frames must be done from the "EDIT MODIFY TAKE" menu as A-SCD-1 described in the Sculpt-Animate manual. From the global menu, turn off "RAM Animation", turn off "Save Images", and select the "Frame Buffer" button. Sculpt will bring up a requester asking you to choose the frame buffer driver. Select the "ilink" driver. If the "ilink" driver is not present, you must copy the devs drawer from the ImageLink installation disk to the Devs: directory on your system. When a frame begins rendering, ImageLink will behave as though the START button was selected. Sculpt will not actually begin rendering the image until the Conversion and Scale requesters in ImageLink have been satisfied. The Sculpt screen must be pushed to the back to reveal the ImageLink screen when the Sculpt "Busy Gears" appear. Once the Scale requester is satisfied, the Sculpt screen will automatically pop back to the front of the display. Note that since ImageLink buffers it's data for a conversion, if images are being rendered. This behavior will continue for each buffered group of scanlines until the image is complete. If ImageLink is being run in quiet mode, the settings which have been selected via ARexx A-SCD-2 will be used for the conversion of each frame. No requesters will interrupt the conversion process, making this mode the preferred mode for a single frame recording to tape. Sample scripts are provided on the ImageLink disk in the ARexx drawer for use with the Sculpt Take mode. A-SCD-3 The Sculpt RGB File Modules Creating Images Sculpt-Animate 4D allows images to be rendered to disk in 24-Bits of color. Selecting "BitPlanes" from the "Mode" menu in Sculpt will bring up a series of requesters allowing you to define the dimensions and depth of the image. A depth of 24 must be selected for later use with ImageLink. Sculpt saves out three separate files each for red green and blue. You must tell Sculpt to use the following file name suffixes for the input module to be able to read the image: Red File: .red Green File: .grn Blue File: .blu Only one of the individual .red, .grn or .blu files need be selected as the input file when reading a Sculpt image using the Sculpt RGB File input module. When writing a Sculpt RGB file, ImageLink will append the .grn, .red and .blu suffixes to the filename selected for output. Selecting aspect ratios of 1 for both x and y is also recommended. Caveats Unfortunately, the raw RGB data which Sculpt writes out does not provide for storing other information about the bitmap (size, depth etc.). The exact dimensions of the image rendered must be recorded for later conversion with ImageLink. ImageLink requires that the exact di- A-SCR-1 mensions of the image be entered into the scale requester. When the scale requester appears, the dimensions of the input will be initially set to all zeros. The correct values for the input image must be entered here for the conversion to work properly. It is recommended that values which correspond closely, if not exactly, to the desired output device/format be used for rendering the original image. This will yield the best possible results. Storage and Compression Issued Note that since the RGB data is uncompressed, reading these files progresses more rapidly than other compressed formats (such as IFF). The tradeoff is that the files take up much more disk space than a compressed format would. Allowances for extra disk space should be made when converting from other formats to this format. Effective compression can be achieved by converting these images to 24-Bit IFF ILBM format using ImageLink and the IFF ILBM Module. Improving AntiAliasing in Sculpt Special tip for Sculpt users: To improve the anti-aliasing of Sculpt images in version 2.09c or earlier, render at twice the size of your target device and use ImageLink to scale the data down. A-SCR-2 The Targa Modules The Targa File modules are provided to enable the reading and writing of images in Truevision Targa (TGA) format. These images may support either the Targa 16, 24 or 32 boards from Truevision which provide 16¹, 24 or 32² bits of color respectively. The files created by ImageLink are suitable for export to any system capable of reading images in Targa format, such as IBM based Targa systems, Slide and Film recorders and production houses. If you would like the capability to use a standard Targa board in an Amiga environment, please contact Active Circuits, Inc. about the Targa Direct Module. Conversion Issues To get Targa files into and out of your system without using an IMB Bridgeboard, we recommend using DOS-2-DOS from Central Coast Software which will format, write and read IBM 3.5", 720K format disks. You may also use a direct or modem connection via the Amiga serial port to upload and download to/from an IBM system. Be sure to use a binary transfer protocol when sending the images in this fashion. If you are using a Bridgeboard, you must copy the files across using the "/b" option, e.g.: A>AREAD AMIGAFile MSDOSFile /b A>AWRITE MSDOSFile AMIGAFile /b ¹15 bits define the color (32,768 possible) and one bit defines the live video overlay plane. ²24 bits define the color (16 million possible) and 8 bitplanes define the live video overlays etc. A-TGA-1 The Release 2.0 Targa output module creates uncompressed "Type 2" or compressed "Type 10" Targa files. When a conversion is started using the Targa input module, there will be an appreciable delay before the color/conversion requester appears. This is because Targa files are stored in reverse scanline order from most common bitmap formats and ImageLink must pre-scan the entire file before starting. Type 10 files are written in normal scanline order. The Targa input module will read both Type 2 and Type 10 (compressed) files in either normal or reversed scanline order. The input module will pause before continuing on to the conversion screen if the Targa file is stored in compressed reverse scanline order. This is because the input has to scan the file to build an index for the compressed lines in reverse order. A-TGA-2 The Turbo Silver Modules Overview The Turbo Silver module as provided in the base ImageLink package, allows the reading of images created by Impulse's Turbo Silver rendering package or by the impulse paint package known as Diamond. These packages create images in one of three file formats: RGB4, RGBN and RGB8. The ImageLink module will read images created in either RGBN or RGB8 format. The RGB8 format is used to store images in full 24-Bit color and should be the preferred method for generating images with the Turbo Silver product. The module automatically recognizes which format (RGBN or RGB8) the image is, no further user intervention is required. Creating 24-Bit Files in Turbo 24-Bit images may be created in Turbo by selecting "24 bit RGB" from the "Modes" menu before loading or creating a CEL for editing or rendering. Mimetics Frame Buffer ImageLink Release 2.0 also provides a writer for the Turbo Silver RGB8 format. Mimetics frame buffer owners can take advantage of this feature along with ImageLink's rescaling capabilities to view images from a variety of formats on their own boards. A-TSV-1 Installation To install ImageLink on a hard disk or other Amiga disk volume, double-click on the "Install" icon (always work from a backup copy of your original ImageLink disk). A window will appear, prompting for the name of the destination drive or volume. Type in the name of the disk volume or full path where ImageLink is to be placed. For example, if you wish to install ImageLink as DH0:ImageLink, enter the following: DH0: To install ImageLink as DH0:Bin/ImageLink, enter: DH0:Bin A new directory called ImageLink will be created, and ImageLink and it's modules will be placed there. Any required shared libraries will be copied to the LIBS: directory. if the library already exists, ImageLink will prompt for permission to overwrite the existing copy with its own. The script will then prompt for special ImageLink drivers to be placed in the DEVS: directory (e.g. the Sculpt-Animate driver). Answer "Y" to have the drivers installed and "N" is they are not to be B-1 installed. Note that they may be installed later by copying the files from the devs drawer on the ImageLink installation disk to your DEVS drawer. B-2 Error Messages ImageLink makes every attempt to recover gracefully from error conditions. In the event of an error, a system requester, such as shown below in figure 16 will appear. The requester will describe whether the error was encountered in the input or output module, and the type of error that occurred. Selecting the "OK" button will terminate the error requester and reset ImageLink. There are seven types of errors that may occur. A description and suggested course of action for each error type follow: Data Format - The input data was nt in the correct format. Check that the file was created in the desired format and that the correct modules have been selected. File or Device Error - An error occurred while reading or writing the file. This was likely caused by an AmigaDOS or media error. Check the integrity of the disk volumes being used. Module Init - An error occurred while ImageLink was launching the modules. ImageLink must be started either from the Work- C-1 Bench or from the CLI while CD'ed to the main ImageLink directory. Verify that the installation was performed properly. No Memory - ImageLink could not allocate memory for it's internal buffers. If other applications are running, attempt to free their memory or shut them down. Color reduction requires that the entire original image be read into memory at one time. If there is not enough memory, consider reducing the size of the original image first. Protocol Error - A serious internal ImageLink error occurred. Attempt to re-create the situation reliably and contact Active Circuits, Inc. to report the problem. Terminate - An error occurred while ImageLink was unlinking the modules. Please contact Active Circuits, Inc. to report the problem. Start Illegal - This input module cannot be started by the user, but must be started by some external process or program. Consult the documentation on the input module for the correct procedure. C-2 References The following references may be useful in understanding bitmaps in computer graphics: Fundamentals of Interactive Computer Graphics J.D. Foley, A. Van Dam Addison Wesley Bitmap Graphics SIGGRAPH '84 Course Notes R Pike, L Guibas, D Ingalls ACM SIGGRAPH 1984 COURSE NOTES Graphics in Overlapping Bitmap Layers Rob Pike ACM Transactions on Graphics 1983 Procedural Elements for Computer Graphics David F. Rodgers Springer Verlag Illumination and Color in Computer Generated Imagery Roy Hall Springer Verlag Color Quantization For Frame Buffer Display Paul Heckbert ACM SIGGRAPH Proceedings 1982 Amiga ROM Kernal Manual Addison Wesley D-1 The following references may be helpful in learning how to use and write ARexx scripts: ARexx User's Reference Manual William S. Hawes A Practical Approach To Programming Rexx M. F. Cowlishaw Prentice Hall Modern Programming Using Rexx O'Hara, Gomber Prentice Hall D-2 ============================================================================ DOCS PROVIDED BY -+*+-THE SOUTHERN STAR-+*+- for M.A.A.D. ============================================================================