Club Amiga de Montreal

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

/ Club Amiga de Montreal - CAM / CAM_CD_1.iso / files / 471.lha / ImageLab_v2.4 / ImageLab.Doc < prev next >

Wrap

Text File | 1991-02-07 | 49.8 KB | 1,570 lines

------------------------------------------------------------------------- ####### # # ##### ##### ####### # ##### ###### # ## ## # # # # # # # # # # # # # # # # # # # # # # # # # # # # ####### # #### ####### # ####### ####### # # # # # # # # # # # # # # # # # # # # # # # # # # ####### # # # # ##### ####### ####### # # ###### ------------------------------------------------------------------------- ImageLab V2.4 Copyright 1991 by Gary Milliorn All Rights Reserved Distribution for non-commercial purposes is ok, as long as the program and accompanying files remain unchanged. ------------------------------------------------------------------------- 1.0 OVERVIEW ImageLab is a program which may be used to explore many common image processing operation, from simple averaging to Fast Fourier Transforms (FFT's). ImageLab makes use of the Amiga's blitter to perform fast operations where possible. ImageLab is copyrighted, but is freely distributable for non-commercial purposes. A shareware fee of $25 or so is suggested if you find it useful, and would help encourage future additions. That, in fact, was the motivation for me to produce V2.4. ImageLab is an evolving program. If you find it interesting, have a few suggestions, bug reports (gads!), etc. you can contact me at: mail: Gary Milliorn 5915 Shanghai Pierce Rd. Austin, Tx 78749-1927 bix: gmill I can't promise everything, but I do endeavor to make it as good as I can. 1.1 GENERAL ImageLab opens a window on the WorkBench screen, which allows access to its menus even when the picture you use is 320 or less. The menu strip is also attached to any picture you load, but it may be difficult to use on small screens, since there are so many menus. There are command keys for most of the important functions, so this may not be such a onerous restriction. The easiest way to use ImageLab in this manner is to position the screen partially down the screen and start functions from the exposed WorkBench window. An alternate method is to position the image screen at the top, move WorkBench to front, and drag it partially down. However, application programs cannot reposition the WorkBench window, so you must do this manually. A this possibility is to keep both screens at the top and flip between them using the AMIGA-N and AMIGA-M key or the space bar. 1.1.1 COMMON FEATURES The following applies throughout ImageLab unless otherwise specified: ImageLab uses the ESCAPE key to terminate an operation in progress. Whenever lengthy operation are being performed, the pointer changes to a small spinning clock. When this is true, you can press ESC to stop whatever it is doing; however, this may leave your picture in a partially altered state. ImageLab uses the RETURN and ENTER keys to simulate the "Ok" gadget which appears in requester windows. Similarly, the ESCAPE key simulates the cancel gadget. Functions can be applied to the entire picture, or to a selected area. To select an area, use the mouse to "draw" a selection rectangle by pressing the left mouse button at one corner and dragging it (with the button still down) to the opposite corner. To cancel an area selection in progress, press the ESCAPE key or the MENU button. Functions which have alternate operations (such as Open, Gamma, et. al.) perform the alternate function when either ALT key is held (naturally). All menu items which call up a requester for additional information have '...' appended to them. 1.2 REQUIREMENTS * Amiga 500..3500 * MC68000 or better. * 512K memory (the more, the better). * AmigaDos V1.2 software or newer. * ARP Library version 39.1 or newer. 1.3 KNOWN LIMITATIONS ImageLab allocates some memory from the C-heap, instead of the system memory pool. This means you will probably need a large stack space to run extensive operations. If you experience problems, try setting the stack size to 25K or more. ImageLab's undo function is not very complete. ImageLab cannot take advantage of a MC68881 or MC68882, alas. 1.4 FUTURE ITEMS * HSI in color palette. * HSI separation and merge. * Programmable convolution kernel sizes. * Multiple image screens. * Better 'undo'. * ??? 1.6 MISCELLANEOUS Copyright 1991 by Gary W. Milliorn All rights reserved. Developed using Manx C 5.0d, 32-bit mode. ----------------------------------------------------------------------------- 2.0 INSTALLATION As of V2.3a, ImageLab requires the ARP library to work properly. By now, most people know about ARP so I'll not cover it's installation. ImageLab only requires that the "arp.library" exist in the LIBS: directory. To install the ImageLab program in your system, you can use the install shell script to perform the default installation procedure. Review the file first, and if it is acceptable to you, just enter "execute install" at any CLI and it will run automatically. Otherwise, review the following paragraphs to manually install whatever portions you would like to have. 2.1 IMAGELAB PROGRAM Except for the ARP library, ImageLab does not require any of the other files supplied with the distribution archive to be present. By default, a directory is created on your SYS: directory, and ImageLab is placed in there. You can use the WorkBench or CLI to move it to a more suitable directory, such as "C:", "SYS:bin:", or wherever you prefer non-C-A programs to reside. 2.2 IMAGELAB FONTS ImageLab comes with a font, "ceres", a bold-faced, proportional, 9 point font that (I think) looks nice when used with menus. Its installation is strictly optional, as ImageLab handles fonts of any style and size, up to a reasonable point (Helvetica/50 is right out). Use your own favorite, if you want. If no particular font is specified when ImageLab is started, it first tries to open this font, then tries Topaz/9, then reverts to the default system font. 2.3 IMAGELAB HELP ImageLab can be provided with on-line help by storing the help file ("ImageLab.help") in the same location where the ImageLab program is stored (generally, "SYS:ImageLab/ImageLab.help"); alternately, it can be stored in the "S:" directory (ImageLab will locate it automatically). Installing the help file is strictly optional. ----------------------------------------------------------------------------- 3.0 STARTING IMAGELAB ImageLab can be run via the CLI or via WorkBench. Operation is basically identical, except that some of the defaults are changed. 3.1 CLI STARTUP When started from the CLI, the ImageLab command line has the following format: ImageLab [-a str][-biv][-c #][-f font] [[dir/]file] the options are: -? : summarize CLI usage. -a str : set IFF `ANNO' chunks to 'str' on write. -b : enable the bell. -c ## : set the clipboard unit to ## (default is 0). -f font : specify font. Fonts are specified as "name/size", as in "Helvetica/9". -i : enable icon generation. -v : verbose. If set, ImageLab prints debugging info. dir/file : load initial image from `file' (in the specified directory). File requester directory defaults to `dir'. Icon creation is disabled when ImageLab is started from the CLI. To enable it, you must specify the "-i" option. 3.2 WORKBENCH STARTUP ImageLab can be started up via the WorkBench. Its icon is a tool, which allows it to receive filenames by clicking on a Project icon such as ImageLab creates, or selecting both to open "alien" files, such as those created by Deluxe Paint, DigiView, etc. The ImageLab icon accepts the tool types listed below. To set them, use the WorkBench "Info" function. ANNO=str : set IFF `ANNO' chunks to 'str' on write. BELL=ON|OFF : if ON, create bells; otherwise, do not. CLIP=## : set the clipboard unit to ## (default is 0). FONT=name : set the ImageLab font to "name" Fonts are specified as "name/size", ICON=ON|OFF : if ON, create icons; otherwise, do not. VERBOSE=ON|OFF : if ON, ImageLab writes debugging information to its WorkBench window. Do NOT use without a "WINDOW..." specification. WINDOW=CON:etc.. : create a message window of the specified size. Mostly just for debug support. Do NOT remove! If you had also selected a PROJECT when ImageLab was started, it tries to open that file as an IFF image file. Only one file may be specified this way, however. Icon creation is enabled when ImageLab is started from the CLI. To disable it, you must specify the "ICON=OFF" option. NOTE: ImageLab *MUST* have a "WINDOW=" tooltype when it runs, even if you don't use the VERBOSE option. If this annoys you, get ConMan and edit it to be "WINDOW=CND:0/130/640/60/ImageLab Console" (a deferred open console) and the window won't open unless any messages are written. ----------------------------------------------------------------------------- 4.0 KEY REFERENCE The following table lists the keyboard keys which can be used in ImageLab: ESC - terminate an operation. Same as 'CANCEL'. RETURN - select an operation. Same as 'OK'. ENTER - same as return. HELP - invoke the help utility. F1-F5 - reserved. F6 - Disable Paint (enables Area Select tool). F7 - Paint: Draw mode. F8 - Paint: Trace (fill-in) mode. F9 - Paint: XOR mode. F10 - Paint: Random mode. SPACE - bring ImageLab Menu window to front. UP - scroll image up 1 line. DOWN - scroll image down 1 line. LEFT - scroll image left 1 line. RIGHT - scroll image right 1 line. SHIFT-UP - scroll image up 10 lines. SHIFT-DOWN - scroll image down 10 lines. SHIFT-LEFT - scroll image left 10 lines. SHIFT-RIGHT - scroll image right 10 lines. ALT-UP - scroll image up one screen. ALT-DOWN - scroll image down one screen. ALT-LEFT - scroll image left one screen. ALT-RIGHT - scroll image right one screen. CTRL-UP - position the screen up 10 lines. CTRL-DOWN - position the screen down 10 lines. AMIGA-0 - Clear Area. AMIGA-1 - Smooth Convolution. AMIGA-2 - Sharpen Convolution. AMIGA-3 - Sobel Transform. AMIGA-4 - Roberts Transform. AMIGA-5 - Binary Transform. AMIGA-6 - Threshold Transform. AMIGA-7 - Flip Horizontal. AMIGA-8 - Flip Vertical. AMIGA-9 - Pseudo-Color Palette. AMIGA-A - Average Area. AMIGA-B - Convert Color to Black & White. AMIGA-C - Copy selected area. AMIGA-D - Display FFT Transform. AMIGA-E - Edit/paint picture. AMIGA-F - FFT Transform. AMIGA-G - Greyscale Palette. AMIGA-H - Histogram Function. AMIGA-I - Inverse FFT Transform. AMIGA-J - Invert Area. AMIGA-K - Close Picture. AMIGA-L - Laplacian Convolution. AMIGA-M - Duplicate Area. AMIGA-N - Noise Generation. AMIGA-O - Open Picture. AMIGA-P - Palette Edit. AMIGA-Q - Quit ImageLab. AMIGA-R - Revert Picture to Old Version. AMIGA-S - Save Picture. AMIGA-T - Distort. AMIGA-U - Interpolated Copy Area. AMIGA-V - Paste Area. AMIGA-W - Double Area. AMIGA-X - Cut Area. AMIGA-Y - Copy Area. AMIGA-Z - Undo last operation. AMIGA-[ - Darken Image. AMIGA-] - Brighten Image. AMIGA-< - Scale Down Image. AMIGA-> - Scale Up Image. AMIGA-- - Negative Palette. AMIGA-= - Histogram Palette Equalization. AMIGA-_ - Vertical-edge Convolution. AMIGA-| - Horizontal-edge Convolution. AMIGA-, - Occupy All of Screen. AMIGA-. - Occupy Half of Screen. AMIGA-/ - Occupy Little of Screen. ----------------------------------------------------------------------------- 5.0 COMMAND REFERENCE The following sections describe the function of each menu item in detail. Each subject is listed in the order it appears in the menus, in left-to-right, top-to-bottom order. The heading for each subject includes the names of the menus on the levels preceeding it, if any. PROJECT ABOUT The "About" menu item provides general information about ImageLab, such as: * Version number * Shareware status. * Where to contact me to register. * Where to send suggestions and bug reports. * How much free memory is available. * Access to the "help" system. Click on the "OK" gadget when you are done, or click on "Help" to get further information. PROJECT INFO The "Info" item opens a window and displays the following information about the picture file currently open: +--------------------------------------+ | Name: "filename". | Name of the Image File. | File Size: 12345 bytes. | Size of the Image File. | Image Size: LLL x WWW x HH. | Total Size of the Image. | Screen Size: LLL x WWW x HH. | Displayed Size of the Image. | Pixel Range: 0..N. | Active pixel range. | Mode: (none). | Screen Modes: EHB, HAM, HIRES, etc. | Region: (0,0)..(8,8) Size: [9 x 9]. | Selected area: location and size. | FFT: 1024 complexes. | Size of FFT retained information. +--------------------------------------+ PROJECT OPEN... The "Open..." item brings up the ARP file requester. Refer to the ARP documentation for further details on the file requester, but it is fairly straightforward. If there is an image opened already, and it has been modified, then a requester will be displayed. Select "OK" to discard your changes, and proceed with loading the new image. Pressing either ALT key when beginning "Open..." will cause any opened image to be discarded without asking first. Select the IFF file you want to use, and select OK to read in the image or CANCEL to stop loading in a new picture. The file may be any standard IFF image as long you have sufficent CHIP memory to load the image. OPTIONS ALT - if held down while making the "Open..." selection, then ImageLab will close any open, edited picture without asking you to verify that you want to discard the changes. All changes will be lost. SHIFT - if held down while making the "Open..." selection, then one bit plane will be stripped from the file while it is read in. This is useful to process Digi-View files (which have an extra bit plane in them) in ImageLab. This has the same effect as the "Restrict Palette" option, but conserves the memory that would otherwise be needed. PROJECT REVERT The "Revert" menu item re-opens the image using the contents of the last file you opened or saved. This menu item is only enabled when any changes have been made to your image, or to its color-map. Revert recognizes the same options as the Open function (see PROJECT OPEN). PROJECT CLOSE "Close" closes the current image and frees all associated memory (including any FFT data). If the image has been modified but not saved, a warning requester will be displayed first. OPTIONS ALT - if held down while making the "Close" selection, then ImageLab will close any open, edited picture without asking you to verify that you want to discard the changes. All changes will be lost. PROJECT SAVE The "Save" item writes to disk your current image file, including any changes that have been made to it. User-settable options allow ImageLab to create icons and/or include ImageLab specific information in the IFF file. See Also: SAVE FFT SAVE TYPE ICONS PROJECT SAVE AS... The "Save As..." item allows you to save the current image under a different name. Selecting this item brings up the ARP file requester, allowing you to alter the current filename or enter a new one. If OK is selected, then the file is saved as described in the "Save" function. PROJECT QUIT The "Quit" menu item closes the image and exits ImageLab. If the picture has been modified but not saved, a requester will be brought up to give you a chance to save it first. OPTIONS ALT - if held down while making the "Close" selection, then ImageLab will close any open, edited picture without asking you to verify that you want to discard the changes. All changes will be lost. EDIT UNDO The "Undo" operation attempts to undo the last operation that was done. Undo works well with color palette changes, but currently cannot undo changes to the picture itself; use the "Revert" command for that. Only one level of "undo" is supported. EDIT CUT The "Cut" operation copies the selected area to the clipboard, then clears the resulting area to the background color (color 0). If either SHIFT key is held, the area is set to the foreground color (the maximum pixel value, such as 15 for a 16 color screen). EDIT COPY The "Copy" operation copies the selected area to the clipboard. EDIT PASTE The "Paste" menu item copies the contents of the clipboard, if there is any and it is an IFF ILBM FORM, into the image. If the clipboard is empty, you will receive an error message. To paste an area on the image, ImageLab brings the screen to the foreground and puts a draggable rectangle (the size of the paste area) on the screen. Position the area where desired, and click the left mouse button to copy from the clipboard to that area. Click the right mouse button or press the ESCAPE key to abandon the attempt to paste the clipboard -- your image will remain unchanged. EDIT CLEAR The "Clear" operation sets the selected area to the background color (color 0). If either ALT key is held, the area is set to the foreground color (the maximum pixel value, such as 15 for a 16 color screen). EDIT UNSELECT The "Unselect" operation clears the select rectangle from the image, causing all ImageLab operations to be performed on the entire image. EDIT SCREEN The "Edit Screen..." function allows you to change the size, depth, or modes of the current image's screen (it does not affect the image, which will be converted to a scrolling bitmap if the screen shrinks). When selected, a requester is brought up, somewhat like: +-------------------------------------+ | Enter the new screen parameters: | | | | Width: [ 640 ] | | | | Height: [ 200 ] | | | | Depth: [ 5 ] | | | | ( ) HAM (/) LACE | | | | ( ) EHB ( ) HIRES | | | | [ Ok ] [ Cancel ] | +-------------------------------------+ To enable a screen mode (HAM, Interlace, Hi-Res, or Extra-Half-Bright), click in the box until it is selected with a check-mark (the current modes are pre-selected). Size parameters can be changed by editing the displayed values. OPTIONS: ALT - if held while "Edit Screen..." is selected, it will use the size of the currently selected area as the default size instead of the screen size. EDIT IMAGE The "Edit Image..." function allows you to change the size, or depth of the current image. When selected, a requester is brought up, somewhat like: +-------------------------------------+ | Enter the new image parameters: | | | | Width: [ 640 ] | | | | Height: [ 200 ] | | | | Depth: [ 5 ] | | | | [ Ok ] [ Cancel ] | +-------------------------------------+ NOTE: FFT data is preserved across this function, although its applicability is highly questionable. OPTIONS: ALT - if held while "Edit Image..." is selected, it will use the size of the currently selected area as the default size instead of the total image size. TRANSFORM DUPLICATE The "Duplicate" item moves the pixels from the currently selected area to a destination of your choosing. When "Duplicate" is selected, a draggable box is positioned on the screen. Click the right mouse button down to begin positioning it, and release the button when you are over the target area. NOTE: the image is copied off-screen first, so overlapping images can be produced. TRANSFORM RESIZE The "Resize" item causes the selected item to be resized, in a variety of ways. When selected, the screen is brought to the front (if needed) and the mouse drags to alter the size of the selected area (the starting location cannot be changed). Select the approximate area, and release the mouse button to begin resizing the area. Note that it is possible to scale up in one dimension and scale down in another at the same time. TRANSFORM RESIZE COPY The "Resize Copy" resizes an area by expanding or shrinking it to fit the target area. When expansion is needed, pixels are supplied by copying the last pixel encountered (the one immediately above or to the left). TRANSFORM RESIZE INTERPOLATE The "Resize Interpolate" resizes an area by expanding or shrinking it to fit the target area. When expansion is needed, pixels are supplied by using bilinear interpolation to create reasonable pixels in the intervening areas. The difference between "Copy" and "Interpolate" arise when the area is expanded: "Copy" just copies the first pixel to fill the expanded area, while "Interpolate" employs bilinear interpolation to fill in the area. Interpolation produces a smoother picture, but is slower to operate. TRANSFORM RESIZE HALVE The "Halve" option allow you to quickly halve the selected area to half its original size. OPTIONS ALT - if held while selecting the "Resize Halve" function, the destination area will be kept square. TRANSFORM RESIZE DOUBLE The "Double" operation allows you to quickly double the size of the selected area. "Double" employs interpolation for best results. OPTIONS ALT - if held while selecting the "Resize Halve" function, the destination area will be kept square. TRANSFORM FLIP HORIZONTAL The "Flip Horizontal" item flips the selected area around an imaginary vertical axis, producing a left-right mirror image. TRANSFORM FLIP VERTICAL The "Flip Vertical" item flips the selected area around an imaginary horizontal axis, producing a top-bottom mirror image. TRANSFORM INVERT The "Invert" item inverts the selected area. This produces a negative- like area for greyscale images. TRANSFORM PAINT The Paint functions are used to change what the mouse button does. Normally, it is used to select areas, but if any of the paint functions are selected, the mouse is used to change pixel values (and corresponding FFT data points, if any). TRANSFORM PAINT PAINT OFF The "Paint Off" function is the default selected item. When checked, the mouse is used to select areas of interest. TRANSFORM PAINT DRAW The "Draw" function is used to set pixels to the last selected color (using the COLOR PALETTE function). FFT data is set to the real value of the color index. TRANSFORM PAINT XOR The "XOR" function is used to toggle the pixels under the mouse. FFT data is edited by complementing the sign of the imaginary component. TRANSFORM PAINT RND The "RND" function is used to replace the pixels under the mouse and FFT real value with random values. TRANSFORM PAINT TRACE The "Trace" option works like "Draw", except that intermediate pixels are connected. This produces smoothed lines; however, it cannot be used to modify FFT data. TRANSFORM FFT TRANSFORM The "Transform" function invokes ImageLab's two-dimensional FFT routine. FFT's convert time domain information (a picture is a valid representation of this) to frequency domain information. Using the default settings, the transform of an image produces a strong DC component at the center (representing the uniform background) with sine circles radiating outwards (representing the edge frequencies of components in the picture). Refer to any good textbook for a better explanation of this. When the FFT has been computed, you can use the "Paint" functio to introduce errors in the transform, and use the "Inverse" function to try to recover your picture. For fun, try blotting out the strong DC term in the center of the picture. ImageLab uses an FFT instead of a DFT for speed; however, FFT's must be a power of two in both dimensions (2, 4, 8, etc.). If the picture or selected area is not a power of two in size, the FFT function is disabled. To select an FFT sized area, hold down the CONTROL key while drag-selecting an area. The Transform function can take some time for FFT's over about 32 by 64 (on a standard MC68000 Amiga). A bar chart tracks the FFT's progress in the ImageLab display window. NOTE: the only functions which can alter the associated FFT data is the "Paint" function. TRANSFORM FFT INVERSE The "Inverse" operation computes the complex conjugate of the FFT data. If the data has not been modified, you can use Display and your original picture will be regenerated. If it has been altered in some way, your source may substantially return, but with a few errors introduced. The "Merge/XOR" function can quickly highlight changed pixels. NOTE: if Translation is enabled (usually the case), your picture will return rotated 180 degrees. You can use the "Flip Horizontal" and "Flip Vertical" functions (both in sequence) to quickly correct this. TRANSFORM FFT DISPLAY The "Display" function causes ImageLab to replace a selected area of the current screen with a two-dimensional representation of the FFT data you have computed (you cannot select this function unless FFT data is present, whether it was just created or loaded from a file). When selected, the screen is popped forward, and you can use the mouse to position a fixed-size rectangle on the screen. Click the right mouse button over the desired area, and the FFT data will be filled into that area. NOTE: The "Display" function is controlled by the "Log Display" option. You do not have to re-compute the FFT if you only change the display modes. TRANSFORM FFT TRANSLATION The "Translation" function is a chcek-selected option that enables the displacement (or offset) of the FFT function to the center of the FFT area. Without this option enabled, the FFT data places the DC component at the lower left corner, with reflections appearing in all four corners, like so. +-------------+ +-------------+ |*- -*| | | |` `| | . | | | | * | | | | -*@*- | | | | * | |` `| | . | |*- -*| | | +-------------+ +-------------+ No Translation With Translation NOTE: if the translation option is changed, you must re-compute the FFT for it to have any effect. TRANSFORM FFT LOG DISPLAY If the "Log Display" option is checked, the "Display" function will perform a logarithmic enhancement to the FFT data before displaying it. The dynamic range of the FFT falls off sharply along the axes from the DC component at zero. This option is automatically checked when you select the "Transform" function, since that is the usual case. It is cleared when "Inverse" is selected, for the same reason. If, for some reason, you don't want this, you can override the settings before you select "Display". Note that the FFT Transform data will be very black, possibly with a singular white DC term at the center, unless you use the log enhancer. TRANSFORM FFT ZERO REAL The "Zero Real" option sets the real part of the complex FFT data to zero. TRANSFORM FFT ZERO IMAGINARY The "Zero Imaginary" option sets the imaginary part of the complex FFT data to zero. TRANSFORM FFT INVERT REAL The "Invert Real" function replaces the real part with its reciprocal. The imaginary part remains unchanged. TRANSFORM FFT PHASE RANDOMIZE The "Phase Randomize" replaces the phase component of the FFT data with a random value (it must convert the data to polar form and back, so this can be slow). TRANSFORM FFT FREE FFT The "Free FFT" function releases all memory associated with an FFT. If you are pressed for space, this function will free up quite a bit, since 16 bytes are required for each pixel in the FFT area. TRANSFORM CONVOLVE Convolution operations perform graphical convolution on the selected area. Currently, all convolutions are 3x3 in size. OPTIONS: ALT - if held when any convolution function is selected, then the custom matrix editor is brought up with the contents of the selected convolution matrix. You can edit this matrix to suit your needs. NOTE: the built-in matrices cannot be changed, they are fixed and will be reset each time you use them. TRANSFORM CONVOLVE SMOOTH The "Smooth" operation uses graphical convolution to smooth the area selected by using the smoothing kernel: 1 1 1 1 1 1 1 1 1 "Smooth" rounds off rough edges of an image at a cost of clarity. TRANSFORM CONVOLVE SHARPEN The "Sharpen" operation uses graphical convolution to accentuate the differences between the center pixel and the surrounding ones. The kernel used is: -1 -1 -1 -1 9 -1 -1 -1 -1 "Sharpen" accentuates subtle changes of an image at a cost of smoothness. TRANSFORM CONVOLVE LAPLACIAN The "Laplacian" operation uses graphical convolution to highlight outline edges. The kernel used is: -1 -1 -1 -1 8 -1 -1 -1 -1 TRANSFORM CONVOLVE LOW FREQUENCY The "Low Frequency" kernel highlights unchanging areas. The kernel used is: -1 -1 -1 -1 1 -1 -1 -1 -1 TRANSFORM CONVOLVE VERTICAL EDGE The "Vertical Edge" convolution operation uses the convolver to highlight vertical edges, but not horizontal ones. The kernel used is: -1 0 1 -1 0 1 -1 0 1 TRANSFORM CONVOLVE HORIZONTAL EDGE The "Horizontal Edge" operation uses graphical convolution to highlight horizontal edges (but not vertical ones). The kernel used is: -1 -1 -1 0 0 0 1 1 1 TRANSFORM CONVOLVE CUSTOM... The "Custom..." option brings up a custom matrix editor. This window displays nine cells in a three by three format. You can select and edit the value in each of the cells. Each cells value will be multiplied by the corresponding pixel value, and the sum of all nine will be stored at the center pixel. TRANSFORM FILTER SOBEL The "Sobel" transform is a highly effective outline tracing function, which examines the gradients present at each pixel location in the picture or selected area. OPTIONS ALT - if held when the "Roberts" function is selected, then the resulting value is scaled down. With the limited resolution of the Amiga display, this is typically never needed. However, sharp edges could result in overflow, so if the result appears corrupted, try again with the ALT key active. TRANSFORM FILTER ROBERTS The "Roberts" transform is a variation of the "Sobel" transform, both of which are effective at outline tracing. The Roberts transform uses the L->R, U->D, and diagonal gradients at each pixel location in the picture or selected area. OPTIONS ALT - if held when the "Roberts" function is selected, then the resulting value is scaled down. With the limited resolution of the Amiga display, this is typically never needed. However, sharp edges could result in overflow, so if the result appears corrupted, try again with the ALT key active. TRANSFORM FILTER BINARY... The "Binary..." function replaces each pixel with white (or the color of the highest palette value, which is typically white) if it is greater than or equal to that limit, or with black (color #0) if it is less than a specified limit. The limit is specified by a sliding bar, which allows you to specify the color index number (0..max). The entire display's color-map updates when you move the slider, but only the selected area will be changed when you select "Ok." The effect of the "Binary" functions is much like orthographic film: the result is pure black and white. Of course, the resolution is cut off sharply. TRANSFORM FILTER THRESHOLD... The "Threshold..." function reduces the range of a picture or selected area by clipping pixels which fall outside a user-defined range. Values outside this range are clipped to the minimum or maximum, as appropriate. The entire display's color-map updates when you move the slider, but only the selected area will be changed when you select "Ok." To select a threshold, a requester is brought up which has two sliders on it: adjust the minimum and maximum values to your desired range. NOTE: the requester will reject attempts to set a minimum value greater than a maximum value. TRANSFORM FILTER PIXELIZE... The "Pixelize..." function replaces each pixel with <n> copies of itself, where <n> is from 2 to 10. This effect "computerizes" images, by craeting a blocky image of low resolution (i.e., IBM compatible). TRANSFORM FILTER MINIMUM The "Minimum" function replaces each pixel with the minimum value found in the surrounding eight pixels. TRANSFORM FILTER MEDIAN The "Median" function replaces each pixel with the median value found in the surrounding eight pixels. Since a sort is required to find a median, this function is slower than the corresponding "Minimum" or "Maximum" functions. TRANSFORM FILTER MAXIMUM The "Maximum" function replaces each pixel with the maximum value found in the surrounding eight pixels. TRANSFORM FILTER DISTORT... The "Distort..." function introduces distortion into the picture or selected area. Distortion is simulated by displacing the pixel in a random direction at random intervals (the distortion probability). A requester allows you to specify the approximate percentage probability that distortion will occur, from 0 (never) to 100 (always). TRANSFORM FILTER NOISE... The "Noise..." function introduces impulse noise into the picture or selected area. Noise is generated by writing the highest index pixel (typically white) at random intervals (controlled by the noise probability). A requester is displayed which allows you to specify the approximate percentage probability that noise will occur, from 0 (never) to 100 (always). OPTIONS ALT - if held while selecting the "Noise..." function, then the impulse noise generated will be color #0 instead of the maximum pixel value (typically, this is black instead of white). TRANSFORM AVERAGE The "Average" functions perform area averaging of the picture or of a selected area. Averaging is performed by using a 3x3 control matrix to select points of interest surrounding each pixel in the selected area. For each non-zero term in the matrix, the corresponding pixel value is added to a sum. Then the center pixel is replaced by the average value. The effect of averaging an area is to reduce noise by blurring the area, in a manner similar to (but faster) the 'Smooth' convolution kernel. The various averaging options select different severities of smoothing. The various submenus select different control matrices. See the individual sections for details on the matrix used. OPTIONS ALT - if held when any of the average functions is selected, then the custom matrix editor is brought up with the contents of the selected averaging matrix. You can edit it to suit your needs. NOTE: the built-in matrices cannot be changed, they are fixed and will be reset each time you use them. CTRL - if held while selecting an averaging option, then the sum of the selected pixels is always divided by 9 (the kernel size), instead of the number of pixels selected by the matrix. This has the effect of squashing the result into a darker image. TRANSFORM AVERAGE HORIZONTAL "Horizontal" averages are based upon the two pixels neighboring to the left and right, using the following 3x3 matrix: 0 0 0 1 1 1 0 0 0 TRANSFORM AVERAGE VERTICAL "Vertical" averages are based upon the two neighbors above and below, using the following 3x3 matrix: 0 1 0 0 1 0 0 1 0 TRANSFORM AVERAGE H&V The "H & V" average combines the previous two methods to base the average on those four pixels, using the following 3x3 matrix: 0 1 0 1 1 1 0 1 0 TRANSFORM AVERAGE DIAGONAL The "Diagonal" average varies the previous one by using the four pixels at each corner, using the following 3x3 matrix: 1 0 1 0 1 0 1 0 1 TRANSFORM AVERAGE ALL The "All" averaging control matrix (the default) averages each pixel based upon all its surrounding pixels (this produces the best average.) It is based upon the following 3x3 matrix: 1 1 1 1 1 1 1 1 1 TRANSFORM AVERAGE CUSTOM... The "Custom..." average option brings up a custom matrix editor. This editor displays nine cells in a three by three format, somewhat like so: +-------------------------+ | Enter Custom Matrix: | | | | [ 0 ][ 0 ][ 0 ] | | [ 0 ][ 0 ][ 0 ] | | [ 0 ][ 0 ][ 0 ] | | | | [ Ok ] [ Cancel ] | +-------------------------+ You can select and edit the value in each of the cells. Enter a non-zero value (typically one) to enable that location (pictorially speaking) to be used in the averaging, or leave it as zero to ignore that location. Enter a value greater than one to emphasize a particular location. TRANSFORM INTENSITY BRIGHTEN The "Intensity Brighten" item is used to brighten the image, by adding one to each pixel value (and limiting the maximum value). TRANSFORM INTENSITY DARKEN The "Intensity Darken" item is used to darken the image, by subtracting one from each pixel value (and limiting the minimum value to zero). TRANSFORM INTENSITY OTHER... The "Intensity Other..." item is used to brighten or darken the image, by adding or subtracting a value from each pixel value (and limiting the minimum and maximum values). A requester is brought up, allowing values from -20..20 to be entered (zero has no effect). TRANSFORM INTENSITY SCALE UP The "Scale Up" function causes the intensity to be scaled (more strongly than the "Intensity" functions) by using a multiplicative function. "Scale Up" scales each pixel value by 2 (overflow values are truncated at the maximum pixel intensity). TRANSFORM INTENSITY SCALE DOWN "Scale Down" halves the intensity of a selected area. TRANSFORM INTENSITY SCALE OTHER... The "Scale Other..." option works like the "Scale Up" and "Scale Down" functions, by allows larger scalars to be entered. A requester is brought up, allowing values from -20..20 to be entered (zero has no effect). Negative values correspond to Scale Down values, while positive values cause scaling up. TRANSFORM INTENSITY ROOT The "Root" option dramatically scales down large intensity range by replacing each pixel by its square root. TRANSFORM MERGE The "Merge" function is used to combine the current picture with another loaded from disk. The submenu selected controls the manner in which the pictures are merged. When any merge option is selected, a file requester is activated, allowing you to specify the second picture. Typically, this should be the original picture, or at least one with the same dimensions and depth of the original source, but this is not required. When "OK" is selected, a new screen containing the selected picture is activated, and then the corresponding pictures are merged. TRANSFORM MERGE OVERLAY The "Merge Overlay" function copies all pixels in the merge picture onto the original one wherever the original one contains a zero pixel. This is somewhat like a genlock overlay. TRANSFORM MERGE ADD The "Add Overlay" function performs an arithmetic merge of the two pictures, limiting the pixels to the maximum allowed value if overflow occurs. TRANSFORM MERGE SUBTRACT The "Subtract Overlay" function performs an arithmetic merge of the two pictures, subtracting the merge picture from the source picture. Pixels are limited to the minimum allowed value if underflow occurs. TRANSFORM MERGE AND The "AND" function uses the Amiga's blitter to perform a logical AND (source & merge -> source). TRANSFORM MERGE OR The "OR" function uses the Amiga's blitter to perform a logical OR (source | merge -> source). TRANSFORM MERGE XOR The "XOR" function uses the Amiga's blitter to perform a logical XOR (source ^ merge -> source). TRANSFORM MERGE OTHER... The "Other..." function allows custom blitter operations; to use, enter the blitter minterm as a decimal number (or in hex if you preceed it with '$'). Refer to the Amiga RKM's for details on blitter minterms. PICTURE HISTOGRAM DISPLAY The "Histogram Display" function counts the number of each pixel values in the picture or selected area, and displays the totals. The results are displayed as a graph in the main (WorkBench) window. Histograms can be produced of HAM pictures, but the information is scaled down from 4096 pixels to 256 pixels when the X axis is plotted. PICTURE HISTOGRAM EQUALIZE The "Equalize" function uses the calculated histogram information to adjust the color palette values to obtain a more uniform color distribution. PICTURE HISTOGRAM BAR GRAPH The "Bar Graph" option, if checked, causes histograms plots to be generated using bar graphs. If not selected, line graphs are produced. PICTURE COLORS PALETTE The "Palette..." option brings up a color palette on the screen. You can select a particular color from the block on the palette window, and alter its red, green, and blue components using the three sliders labelled R, G, and B. If you select "OK", the changes are added to your picture. If you select "Cancel", the old palette is restored. PICTURE COLORS GREYSCALE The "Greyscale" option sets the picture's color map to a linearly increasing range. This option automatically adjusts to different picture depths, but since the Amiga's video DAC's are only 4 bits wide, only 16-color pictures can make proper use of greyscale images. Pictures with more than 16 colors will have repeated grey colors, while those with less than 16 colors will have skipped values. PICTURE COLORS NEGATIVE The "Negative" option reverses the order of the color map. If the picture currently has a greyscale color map, the picture will appear as a negative; otherwise, it is just an interesting effect. Negative values can sometimes make it easier to pick out features and details of the picture. Using this function on color pictures is... also interesting -- you will generally want to convert it to black and white first. SEE ALSO: PICTURE COLORS COLOR TO B&W PICTURE GREYSCALE PICTURE COLORS SATURATE The "Saturate" function converts the palette to values such that all colors are saturated (i.e, if a color is mostly red, it becomes pure red). OPTIONS ALT - if held while selecting this function, then the intensity level is also intensified; this forces each pixel to a basic primary color (red, green, blue, yellow, magenta, cyan, white or black). PICTURE COLORS PSEUDO-COLOR The "PsuedoColor" option converts the palette to values such that dark colors are tinted blue (i.e., "cold") and bright colors are tinted red (i.e., "hot"). Average colors are given a green hue. The effect is much like a thermograph, and tends to accentuate the difference between pixels. OPTIONS ALT - if held when the "Pseudo-Color" function is selected, then the color map is replaced with completely random values. This is mostly for fun. PICTURE COLORS COLOR TO B&W The "Color to B&W" option converts a color picture to a black and white one (a grey scale, actually). This function is the primary tool used to allow ImageLab to process color IFF files (it is also known as the Un-Ted Turner function). "Color to B&W" uses the standard EIA weighting levels to obtains an evenly distributed grey scale. Note: if your picture has more than 16 colors, the greyscale will cause the picture to appear grainier -- this is a limitation of the Amiga DAC architecture. PICTURE COLORS LINEARIZE The "Linearize" option converts the palette to a monotonically increasing palette. ImageLab requires such a palette for its image processing function to perform properly. For example, if your four color palette is: 000 555 fff aaa the "Linearize" function will re-order it as: 000 555 aaa fff Note that this function requires rewriting all pixels in the picture in order to avoid changing its appearance. This function will be required for some DigiView pictures which do not use a linearly increasing palette. PICTURE COLORS DELETE COLORS... The "Delete Colors..." function beings up a requester which has a button for every possible color, numbered from 0..n. To delete a represented color, click on an unselected button and it will be highlightes. When "Ok" is selected, every selected color will be deleted from the picture. The remaining colors are then packed further down into the palette. "Delete Colors..." allows you to remap the color palette, possibly reducing the number of colors (and therefore, graphics bit planes) required. See "Edit Screen" and "Edit Image" for related functions. NOTE: since every pixel must be changed for this operation, it can be lengthy. PICTURE COLORS MAP COLORS... The "Map Colors..." function operates similarly to "Delete Colors", except that only two colors can be selected. When "Ok" is selected, those two colors are exchanged in both the color map and the picture. See: DELETE COLORS... PICTURE GAMMA CORRECTION The "Gamma Correction" applies a series of corrective factors to the greyscale color map, based upon empirical values for the Amiga monitor. These factors vary quite a bit; to alter them, hold down the ALT key when selecting "Gamma Correction" and a requester will allow you to enter other values for gamma, c, and I0. The default values are: gamma = 0.5 c = 1.75 I0 = 1.00 PICTURE HAM CONVERSION The "HAM Conversion" option is enabled only when the loaded image is a HAM picture. Selecting this function converts the picture or selected area to a 4-plane greyscaled image. "HAM Conversion" correctly processes selected area that do not begin in the left-most column by scanning all pixels to the left of the selection area and tracking the resulting RGB values. PICTURE HALFTONE... The "Halftone..." function allows the conversion of the current screen to a halftoned image. A requester allows the selection of a two-cell or three-cell halftone (note that this implies that the resulting picture will be two or three times as large as the selected area). The original screen is replaced with a 1 bit-plane deep screen containing the half-toned image (in black and white). OTHER POSITION SCREEN TO FRONT Moves the image screen to the front. OTHER POSITION SCREEN TO BACK Moves the image screen to the back. OTHER POSITION OCCUPY ALL Positions the image screen to occupy all of the display. The menu key for this function is LEFT-AMIGA- ','. OTHER POSITION OCCUPY MIDDLE Positions the image screen to occupy half of the display. The menu key for this function is LEFT-AMIGA- '.'. OTHER POSITION OCCUPY LITTLE Positions the image screen to occupy a small part of the display. The menu key for this function is LEFT-AMIGA- '.'. OTHER AUTO FRONT The "Auto Front" option, if checked, automtically flips the image screen to the front and back after each ImageLab function is executed. This should make it easier to access the menus without having to use the screen position functions or Left-Amiga-M/N each time. BUGS: it causes the screen to flip for trivial functions also, which is very annoying. OTHER SAVE TYPE COMPRESS PICTURE The "Compress Picture" item specifies how pictures will be saved. If the item is not checked, the file will be saved using no compression. If checked, the standard byte-run compression will be used. The state of this menu item is set each time an image is loaded, and reflects the method of compression used in the file. This option is examined when images are written out, so you can use this to change the compression method, if needed. OTHER SAVE TYPE SAVE FFT DATA The "Save FFT Data" option allows ImageLab to write out any FFT data with the image file. The data is saved in a FORM ILBM file with the usual image data, and the FFT data stored in a "CFFT" chuck. By default, this is disabled, since "CFFT" is a custom, non- registered IFF ID. This *should* be transparent to other picture display programs, but, alas, that is not always true. OTHER RESTRICT PALETTE The "Restrict Palette" function is primarily used to deal with image files which have a palette larger than is necessary to represent the picture (i.e., the picture has 16 colors, but the palette has 32 locations reserved). This occurs often with some DigiView pictures, which creates an extra bit-plane for internal control functions. If this option is checked, then the palette is halved (regardless of the total number). The "Restrict Palette" option is reset (unchecked) whenever a new picture is loaded (or "Revert"ed), or when the screen parameters are edited. See also: EDIT PARAMETERS OPEN REVERT OTHER ICONS The "Icons" item controls whether ImageLab creates an icon whenever a file is saved (via "Save" or "Save As..."). If "Icons" is checked (sic), an icon is generated that resembles the Deluxe Paint (tm) icon, except that the default project is ImageLab. Icon generation defaults to unchecked if ImageLab is run from a CLI, and to checked if it is run from WorkBench. Icon generation can also be initialized by the options: -b (from the CLI) ICON=OFF (from the WorkBench) OTHER BELL The "Bell" item controls whether ImageLab signals an alarm on the completion of any operation. If the menu is checked, then the Amiga's bell function is called. This will simply flash the screen by default, but several public domain functions are available which will intercept this call and substitute an audible bell or a sampled sound. The primary use of the bell function is to alert you when some of the lengthier functions have completed. Some functions, especially the FFT function, may take several minutes to complete. The bell is off by default, since it would tend to be annoying for trivial functions. It can be enabled by selecting the menu item until the checkmark is present. The bell can be enabled by default by using the options: -b (from the CLI) BELL=ON (from the WorkBench)