═══ 1. Bitmap File dialog ═══ This dialog allows you to load a bitmap from a file, or save a bitmap to a file. Filenames and directories may be selected from the listboxes via the keyboard or the mouse, just like in any other File dialog. You can handle bitmap files of a variety of different formats. For example, OS/2 Bitmaps, CompuServe GIF files, ZSoft PCX files and many others. To see an extensive list of the level of support for the various bitmap formats, and the format specific options available in each, see the Bitmap file format support section of this help. You can select which files are listed in the files listbox using the File type combo box. You can pick, , and various s. The program will look at the file extension and determine the format of the file and use the relevant bitmap file handling code. The Optional bitmap format specific options field allows you to specify a string which is used to give hints to the relevant bitmap file handling code. For example, you can tell the GIF file reader to ignore corrupt data and return how far it got by giving the "errok" option. ═══ 2. Bitmap file format support ═══ This program uses the Generalised Bitmap Module. This peice of code can read and write a variety of bitmap formats. The application program is unaware of the file format of any bitmaps, and is just presented with the bitmap data in a single standard way. Some bitmap formats allow additional options to be supplied and these options are listed highlighted like this. This section fully defines the level of support :- Formats supported are : OS/2 / Windows Bitmap CompuServe Graphics Interchange Format ZSoft PC Paintbrush Image Format Microsoft/Aldus Tagged Image File Format Truevision Targa/Vista Amiga IFF / ILBM Interleaved bitmap format YUV12C M-Motion Frame Buffer Portable Greyscale-map Portable Pixel-map IBM KIPS IBM Image Access eXecutive X Windows bitmap RiscOS Sprite IBM Printer Page Segment Others The Bitmap File Dialog can handle bitmaps in any of the formats documented above. ═══ 2.1. OS/2 / Windows bitmap ═══ The following file extensions .BMP .VGA .BGA .RLE .DIB .RL4 .RL8 are recognised as OS/2 1.1, 1.2, 2.0 or Windows 3.0 bitmaps. Input The bitmap support will read any of the above if uncompressed or run-length compressed. CCITT G3 MH Fax compressed files, or RLE24 compressed files can't be read. index=N Can be used if the file is an OS/2 bitmap-array file to specify which bitmap in the array to read (default N=0 - the first in the array). inv Causes the palette to get inverted after reading. Provided to allow for inconsistencies between Windows and OS/2 programs in how the palette of 1bpp files should be interpreted. invb Causes the bitmap bits to get inverted after reading. Output The this module can write both OS/2 1.1 and OS/2 2.0 / Windows 3.0 style bitmaps. By default OS/2 2.0 or Windows 3.0 bitmaps are written. The OS/2 2.0 / Windows 3.0 style of bitmaps written are written with a header size (cbFix) of 40, which is compatible with both Windows 3.0 and OS/2 2.0. 1.1 Causes the bitmap to be written OS/2 1.1 style. 2.0 or win Causes the bitmap to be written OS/2 2.0 / Windows 3.0 style. inv Causes the palette to get inverted before written. Provided to allow for inconsistencies between Windows and OS/2 programs in how the palette of 1bpp files should be interpreted. invb Causes the bitmap bits to get inverted before written. The newer MultiMedia Windows bitmap format that allows 16 and 32 bpp support is not currently supported. The Bitmap file format support help panel lists other supported file formats. ═══ 2.2. CompuServe Graphics Interchange Format ═══ The .GIF file extension is recognised as a GIF file. Input 1 bpp GIF files are read as 1 bpp data. 2, 3 and 4 bit GIF files are read and returned as 4 bpp data. 5, 6, 7 and 8 bit GIF files are read and returned as 8 bpp data. GIF files provide no support for 24 bpp images. Both GIF87a and GIF89a level files may be read. The code will correctly handle deferred clear-codes in the LZW data stream. Files with these often break earlier programs that claim to load GIF files. index=N Can be used to specify which image in the GIF file to read (default N=0). errok If a corrupted or truncated GIF file is encountered, normally an error is returned. With this option, as much data as was decoded before the error is returned, and no error occurs. Output The GIF87a style of GIF files are written, and since the GIF standard is backwardly compatible, all programs that claim to load GIF files should handle the output generated by this program. This program does not generate GIF files with deferred-clear-codes, as these have been shown to break many older GIF readers. xscreen=N,yscreen=N Specifies what values to put in these fields of the GIF file header. The default is the image size. background=N Specifies what background colour index to write in the GIF file header. The default is 0. xpos=N,ypos=N Specifies what values to put in these fields in the GIF file header. The defaults are 0 and 0. This code correctly handles 1bpp GIF files. Some programs make incorrect assumptions about the initial LZW code table size in the 1bpp case, and so fall over. UBU is guilty of this. This code has been tested on a wide selection of files from a wide variety of sources. It has been tested against the output of public domain PPMTOGIF.EXE, and tested for compatibility with UNIX X Windows xloadimage and xv. The Bitmap file format support help panel lists other supported file formats. GIF and 'Graphics Interchange Format' are trademarks (tm) of Compuserve, Incorporated, an H&R Block Company. ═══ 2.3. ZSoft PC Paintbrush Image Format ═══ The .PCX file extension is recognised as a Paintbrush file. Only a subset of the possible planes/bits-per-plane combinations are supported. Input This program reads 1,4,8 and 24 bpp files. The run-length encoding scheme is supported. Compressed runs of pixels are not supposed to straddle scan lines. Some other applications violate this rule. By default, if this code finds that a run of pixels is longer than the scan line, it will use the leftover pixels on the next scan line. Some images will not load properly, as they require the loading program to discard those pixels in the run that extend beyond the end of the scan line. This behaviour can be enabled using the trunc option. Output This program writes 1,4,8 and 24 bpp files, using run-length compression. It will always ensure run-lengths of pixels do not straddle scan lines. This code has been tested on some PC Paintbrush originated .PCX files. The Bitmap file format support help panel lists other supported file formats. ═══ 2.4. Microsoft/Aldus Tagged Image File Format ═══ The .TIF file extension is recognised as a TIFF file. There is support to handle 1,4,8 and 24 bpp files. Input This program reads uncompressed, Packbits and LZW compressed files. This code will read all Baseline TIFF 6.0 files, except CCITT Modified Huffman encoded bi-level files. index=N Can be used if the file contains more than one image to specify which image to read (default N=0 - the first). Output This program usually writes uncompressed data. Data are written in a single large strip. Files written by this code are Baseline TIFF 6.0 compliant. artist=TEXT Causes an Artist tag to be added with text TEXT. lzw Asks GBM to use LZW compression. software=TEXT Causes the Software tag to be added with text TEXT. The default Software tag to be added if none is specified is Generalised Bitmap Module. make=TEXT Causes the Make tag to be added with text TEXT. model=TEXT Causes the Model tag to be added with text TEXT. host=TEXT Causes the Host tag to be added with text TEXT. documentname=TEXT Causes the DocumentName tag to be added with text TEXT. pagename=TEXT Causes the PageName tag to be added with text TEXT. imagedescription=TEXT Causes the ImageDescription tag to be added with text TEXT. pal1bpp Causes 1 bpp data to be written as a palettised file, with 2 palette entries. Technically this is not TIFF 6.0 Baseline compliant. The default is to write out a bi-level file with the brightest colour as white. This is Baseline compliant. The TEXT should not include spaces. There are 2 well known corruptions of the TIFF format. Some TIFF writers write the ColorMap tags length to be 2^N, whereas it should be 3*2^N. This code will detect this tag having a bad length, and correct for it transparently. UBU is/was guilty of this. Some TIFF readers/writers believe that ColorMap entrys are bytes. In fact they are words. When reading TIFF files, if this code finds that all the palette entrys are in the range 0-255, as opposed to 0-65535, then it multiplys them all by 256, thus transparently overcoming the problem. When writing TIFF files, this code writes the 8 bit palette entry in both the low and high bytes of the tag. Thus programs that look at the high byte work fine. Programs that use the full word work fine (the colours get 0.5% brighter). Programs that incorrectly look at the bottom byte work fine too. At least one version of Image Support is/was guilty of this. JPEG compressed data, and data in the Y-Cb-Cr colour space is not supported. CCITT G3 Modified Huffman / Facsimile compressed data is not supported. The Bitmap file format support help panel lists other supported file formats. ═══ 2.5. Truevision Targa/Vista bitmap ═══ The following file extensions .TGA .VST .AFI are recognised as Targa/Vista files. This handles the uncompressed Targa/Vista formats. This code only supports 8 bpp and 24 bpp images. Input This code will read files with both top-left-origin and bottom-left-origin. Not all Targa readers correctly handle bottom-left-origin files. When reading 16 bpp files (which are 5 red + 5 green + 5 blue + 1 alpha), they are loaded and presented as 24 bpp bitmaps with only the top 5 bits of each of the 8 bit bytes valid. The bottom 3 bits are zero. When reading 32 bpp files (which are 8 red + 8 green + 8 blue + 8 alpha), the alpha channel is ignored. Output yup and ydown Causes the data to be written in bottom-left-origin or top-left-origin form. The default is bottom-left-origin. 16 If the data is 24 bpp, then the 16 flag causes only the most significant 5 bits of the red, green and blue to be written in a 16 bit pixel. 24 Has no effect, as this is the default. 32 If the data is 24 bpp, then the 32 flag causes the data to be padded with 8 zero bits of alpha channel to make a 32 bit pixel. The Bitmap file format support help panel lists other supported file formats. ═══ 2.6. Amiga IFF / ILBM Interleaved bitmap format ═══ The following file extensions .IFF .LBM are recognised as interleaved bitmap files. Normally ILBM files are 1,2,3,4 or 5 bpp, because the early Amigas supported upto 32 colours at once. 6,7 and 8 bpp files are also found nowadays too. Extra-HalfBrite, or EHB is a mode where N colours have palette entries, but the file may contain 2*N colours. The second set of N colours are half the brightness of the first half. In this way, for example, 64 colours may be displayed on an early Amiga. Hold-And-Modify, or HAM is a type of screen format where each pixel is either one of a small number of colours in a palette, or is held the same as the pixel to its left except that either the red, green or blue is modified. In this way a much larger range of colours can be displayed. HAM modes sometimes have fringes down the vertical edges of shapes in the picture, as sometimes it is necessary to modify the lefthand pixel several times to get to the desired colour, if a suitable entry is not in the palette. HAM6 is a HAM mode where there are 16 base palette entries and a possible maximum of 4096 colours. This is found on early Amigas. HAM8 is a HAM mode where there are 64 base palette entries and a full 24 bit range of colours. Also because of the larger palette, fringing is less than in HAM6. This mode is present on later Amigas. Sliced-Hold-And-Modify, or SHAM is a mode where the palette is changed on every (or every other) scan line. This also helps reduce fringing compared to normal HAM6. SHAM is actually sliced HAM6, and might really be called SHAM6, and SHAM8 is not known to exist. Input This code can read files that contain uncompressed or RLE compressed data. 1 bpp files can be read ok. 2,3 and 4 bpp files are read and returned as 4 bpp data. 5,6,7 and 8 bpp files are read and returned as 8 bpp data. EHB images can be read and are returned as 8 bpp data. HAM6, HAM8 and SHAM files can be read and are returned as 24 bpp data. Only the BMHD, CMAP, CAMG and BODY chunks are used when reading the file. Output This code can write 1, 4 and 8 bpp files. 24 bpp files can be written also, this code performs a mapping to HAM6 before writing to the file. This loses quality, but, is still better than a normal 8 bpp file. This code writes RLE compressed data, as this usually results in much smaller files. Only the BMHD, CMAP and BODY chunks are written to the file. When writing HAM data, a CAMG chunk is written also. You can optionally override parts of the BMHD header. xpos=N,ypos=N Overrides the origin information in the file. Defaults are 0,0. transcol=N Overrides the transparent colour field. Default is 0. xaspect=N,yaspect=N Overrides the aspect ratio information. Defaults are 1,1. This header information is not always honored by ILBM readers. xscreen=N,yscreen=N Overrides the screen size information. This is not often used by ILBM readers. The default is the image size. Note that when some early programs try to load an >5 bpp ILBM file, they can switch to a 32 colour (5 bpp mode) and only read the 5 least significant planes of the file. This generally looks bad. Some GBM created IFF files have been found to be loadable by AMOS programs although not by Deluxe Paint. Apparently Deluxe Paint will crash if the image size does not match one of the Amigas graphics modes, such as 320x256. This support has been tested using a large range of sample files. But, some sample files supplied to the author have yet to be loaded. In particular some (24bpp?) DCTV originated files are not yet understood. The Bitmap file format support help panel lists other supported file formats. ═══ 2.7. YUV12C M-Motion Frame Buffer ═══ The .VID file extension is recognised as a YUV12C file. Although YUV12C files contain 12 bit YUV encoded colour image data, this code pretends they contain 24 bit RGB encoded colour data. Input The files are converted from YUV to RGB data before the data is presented to the program. Output The RGB data is converted to YUV when it is written to disk. YUV12C files must be a multiple of 4 pixels wide. When trying to write a bitmap that is not a multiple of 4 pixels wide, this code truncates the last 1, 2 or 3 pixels left over. xpos=N,ypos=N Overrides the origin information in the file. Defaults are 0,0. This support has been tested using the M-Control Program 2.0, Browser sample application. The Bitmap file format support help panel lists other supported file formats. ═══ 2.8. Portable Greyscale-map ═══ The .PGM file extension is recognised as a Greymap file. Credit for file format to Jef Poskanzer. Only the Binary P5 form of this format is supported. These files hold no palette information. Input The files are returned as an 8 bit image usually with a greyscale palette. r,g,b or k Forces the image to be read as shades of red, green, blue or grey (the default). Output Normally 8 bpp image data as the greyscale equivelent of the bitmap data. r,g,b or k Forces the red,green or blue component of the image to be written, or alternatively to write the greyscale equivelent (the default). The Bitmap file format support help panel lists other supported file formats. ═══ 2.9. Portable Pixel-map ═══ The .PPM file extension is recognised as a Pixel-map file. Credit for file format to Jef Poskanzer. Only the Binary P6 form of this format is supported. The Portable Pixel-map format is only capable of holding 24 bpp RGB data, and this code fully supports this. The Bitmap file format support help panel lists other supported file formats. ═══ 2.10. IBM KIPS ═══ The .KPS file extension is recognised as a Pixel-map file. The .KPS file is not enough to define a KIPS image. The palette is stored in either an associated .PAL or .KPL file. Input pal or kpl Tells this code to look for a .PAL or a .KPL file to find the palette (default is .PAL) Output pal or kpl Tells this code to write a .PAL or a .KPL file containing the palette. (default is .PAL) The Bitmap file format support help panel lists other supported file formats. ═══ 2.11. IBM Image Access eXecutive ═══ The .IAX file extension is recognised as a IAX file. This image format does not include a palette or a file header. In particular, the image file size cannot be determined from the file! This code assumes an image width of 512 pixels, and determines the image height by dividing the file size by the width. The data is presented as a greyscale. Input r,g,b or k Forces the image to be read as shades of red, green, blue or grey (the default). width=N Overrides the default assumption that the image width is 512 pixels. Output Normally 8 bpp image data as the greyscale equivelent of the bitmap data. r,g,b or k Forces the red,green or blue component of the image to be written, or alternatively to write the greyscale equivelent (the default). The Bitmap file format support help panel lists other supported file formats. ═══ 2.12. X Windows Bitmap ═══ The .XBM file extension is recognised as a X Bitmap file. X Bitmap files are ASCII text files, that look to be directly source include-able in C programs. This program is very generous in its parsing of the files. This program supports X10 and X11 1bpp bitmaps. Some .XBM files with text strings inside have been encountered. These looked to be sprites or icons and are not supported. When reading X Bitmaps, 1's are returned as black, 0's as white. When writing X bitmaps, the darkest colour is written as 1's, the lightest as 0's. This code has been tested on some .XBM files. The Bitmap file format support help panel lists other supported file formats. ═══ 2.13. RiscOS Sprite ═══ The .SPR and .SPRITE file extensions are recognised as a RiscOS sprite file. Technically, due to the way the RiscOS works, files don't have extensions. But for convenience, I have arbitrarily said .SPR, to allow sprites to fit in with all the other image formats. Input This code will read 1,2,4 and 8 bpp files. 2bpp files will be expanded to 4bpp on reading. Sprite files can contain an image, or an image and a mask plane. This code will ignore the mask plane. If no palette is present in the file, GBM will assume a default palette similar to that used by the RiscOS Wimp. There is a limitation in that the first-bit field of the sprite header must be a multiple of 8 (highly likely). index=N Can be used if the file contains more than one sprite to specify which sprite in the file to read (default N=0 - the first in the file). Output This code will write 1,4 and 8bpp files. A palette will be written into the file. Only the image plane is written, ie: no mask plane. When outputting 8bpp bitmaps, the usual wierd Archimedes palette is written and the bitmap data is mapped to this palette. This is because the Archimedes cannot support arbitrary 256 colour palettes. The VIDC chip only has 16 palette registers, with 16 fixed overrides on (or shades of) each. GBM writes 16 palette entries to the file. Sprite files can be found with 64 palette entries (with 4 overrides on each). However, it is seen that these 64 palette entries are actually just 16 entries with the 4 overrides on each. Hence this is equivelent to 16 entries with 16 overrides, except some of the overrides are explicitly stored in the file. Tested reading 4bpp sprites with no palette. Tested reading 8bpp sprites with no palette and with 64 entry palette. Sprites generated by GBM have been loaded on RiscOS. Remember: Use SetType to change type of file from PC-file to Sprite, by SetType option of filer menu or "*SetType Sprite". The Bitmap file format support help panel lists other supported file formats. ═══ 2.14. IBM Printer Page Segment ═══ The following file extensions .PSE .PSEG .PSEG38PP .PSEG3820 are recognised as PSEG files. PSEG files are used as a way to include image data in SCRIPT documents. PSEG files only contain 1bpp data, which is always ink on paper, ie: black on white. Input PSEG files should be downloaded from the host in binary mode. No EBCDIC/ASCII conversion should be used in the download process. The palette is always returned as black 1's on white 0's, because PSEG files do not include any palette. fixrec Some PSEG files have been observed to have shorter records than specified in the length field (2nd and 3rd bytes) of a record. Although this doesn't matter when on VM, it can confuse non-VM readers, such as GBM. GBM can be told to attempt to compensate for this corrupt record structure using the fixrec option. Output When writing PSEG files, any colour palette is lost. The resulting file is a binary file and should be uploaded to a VM machine as such. No EBCDIC/ASCII conversion should be used. However, on a VM host, the record structure of a PSEG is necessary for it to be correctly interpreted by the VM tools. On non VM systems, files do not have a record structure. Therefore after uploading, the PSEG files must be 'fixed'. You need a tool, like FIXL3820 EXEC to do this. Basically, each PSEG files record starts with a 0x5a character, and this can be used to delineate records. PSEG files are a multiple of 8 pixels wide, so this code writes a file with the width the next multiple of 8 pixels wide. The right edge is padded with white pixels. double There is a scale factor in the PSEG header. By specifying this option, the scale factor is written as double the usual value. The Bitmap file format support help panel lists other supported file formats. ═══ 2.15. Others ═══ Support for all the bitmap file formats may be added at any time, without application writers being aware of it. Any new formats will be listed in the dialog filetype combobox, although the help may not (yet) have an entry for them. The Bitmap file format support help panel lists other supported file formats. ═══ 3. Bitmap File format Credits ═══ This code is the work of :- {{{ Andy Key Internet: ak@vnet.ibm.com IBM-VNET: AKEY AT HVTVM5 Electronic addresses valid at least until September 1994. I would like to thank all the numerous people who have helped by providing sample bitmap files and file specifications.