path: root/DevIL/TODO

DevIL TODO:

-----------

Denton:

Bugs:
- Conversion to IL_ALPHA seems to be crashing (WindowsTest.cpp).
- Conversion to IL_ALPHA can be flipped?


By next release:

- NVTT.dll as a delay-loaded DLL.
- Update docs with next DXT code, new mipmap/cubemap access, new SaveL code.
- Update DevIL website with file formats (psd-saving, etc.).
- Add Nebula Device to projects page.
- Add VTF Edit to projects page.
- Add http://vrml.cip.ica.uni-stuttgart.de/dune/install.html to projects page.
- Verify all formats are listed in il_io.c Doxygen documentation.
- http://www.fileformat.info/format/drhalo/sample/index.htm - Check these.
- Update Portable Any Map name in manual and code. http://www.bbhscanners.com/support/infinity_WF/faqs/faq_001.html
- Update http://en.wikipedia.org/wiki/Tagged_Image_File_Format name as above.
- Same with MNG
- Update Targa extensions in manual.
- Update license page on the website.


Easier:
- Find out what is different in the VTF headers that are 64 bytes.
- Make DXT1 decompression function that does not do alpha.
- Implement ilIsValid calls for ROT.
- Make sure ilFixImage is called everytime something is loaded.
- Add BLP loading of alpha sizes other than 0 and 1 for BLP2 RAW.
- Add BLP loading of mipmaps for all other types (jpeg will be the hardest).
- Find out why BLP2 files do not follow DXTC specs when width or height < 4.
  It looks like they are not wasting space with the small mipmaps.
  An example is a 2x8 image in DXT5.  Their mipmap length claims to be 16, instead of 32.
- Recompile libtiff with extra options (but not jpeg...).
- Clean up ilSetAlpha.
- DxtcReadColor and DxtcReadColors need to expand the colours from 565 to 888 properly.
- Add faces to WindowsTest.
- Add reading other files in directory to WindowsTest, so you can easily view all files in a dir.
- Finish BLP loading (indexed mipmaps, jpeg data).
- dcraw support
- http://en.wikipedia.org/wiki/2xSaI
- Clean up WDP code.
- Get WDP code to do conversions within the WDP library (check to see if type/format set by user).
- Get WDP code to properly do float, short, etc.
- Make DecompressDXT1 and DecompressDXT1a functions separate (.blp files could use this).
- Keep TPL DXT data.
- DecompressDXT1 (and more) do not set an error if returning IL_FALSE (should be an internal error, I think).
- Add ilIsValidFits* functions to il_io.c.
- Add ilIsValidWdp* functions.
- iGetString may not work properly with Unicode strings.
- Look at adding ilIsValid* for mng, iff, wbmp, wal and pxr.
- Added ilSetError error codes to il_files.c.
- Get rid of check for precache.
- If IL_NO_JP2 is defined, lines 306-307 shouldn't just cause the entire thing to fail.  We should just skip that entry.
- Change ilConvertBuffer to be able to convert paletted images as well.
- Fix MSVC8 solutions for text files.
- ilSavePnmF needs to set FName if not called by ilSavePnm.
- Make TPL_CMP case in iLoadTplInternal code more modular (also in il_dds.c).
- Get rid of iluSwapColours call in ilutConvertToSDLSurface.
- Use GL/glext.h.
- Update Eclipse projects.
- Add Code::Blocks projects.
- Check to make sure all VTF formats are supported.
- Look at DDSTextureLoader.cpp for new DDS formats.
- http://www.leunen.com/cbuilder/rotbmp.html
- 3d mipmaps and better 2d: http://www.opengl.org/discussion_boards/ubbthreads.php?ubb=showflat&Number=236590&fpart=2
  http://zach.in.tu-clausthal.de/teaching/cg2_07/literatur/frame_buffer_objects.html
  http://objectmix.com/graphics/745877-glgeneratemipmapext.html
- Make libsquish and NVTT code compress volume textures (do each slice separately).
- Check .dcx code - does it even load multiple images?
- Make ilBlit either error with color indexed images or work with them.
- iluScale: Make ilTexImage or whatever actually use Temp->Data instead of copying.
- Get rid of globals in il_dds.c and ilu_scale2d.c.
- Make ilApplyPal also work for file streams and lumps.
- Make a version of ilApplyPal where a palette buffer and palette type are specified.
- Renormalize FITS and DICOM files (make it an option?).
- Add scaling functions to IL (and wrappers in ILU).
- Fix DXT2 and DXT4 loading.
- Add _ilLoadExt/_ilSaveExt Unicode in il_states.c.
- XCF (Gimp) support


More involved:
- ILinfo struct still has number of images...
- Use loaded mipmaps in OpenGL and DirectX.
- JPEG and PNG in .bmp files?
- It looks like ilSetRead/ilSetWrite don't actually do anything.
- Run Valgrind on library - all functions.
- Make iSaveJp2Internal and iSaveExrInternal work with palettes (make ilConvertBuffer work on palettes as well).
- Make .ico/.icns uses Mipmaps field instead for smaller icons (can icns have multiple images?).
- Work on http://en.wikipedia.org/wiki/Apple_Icon_Image page.
- http://mirrored.xentax.com/gfe/index.html (lots of game graphics formats)
- Strife textures
- Starcraft GRP textures - http://web.archive.org/web/20041011165218/http://user.cs.tu-berlin.de/~mickyk/cv.html
- Commenting out lines 729-731 in pngread.c of libpng (1.2.9?) will allow reading of certain malformed Vista PNG-compressed .ico files with extra data at the end of the stream.  An example can be found in test/in and was created using Microangelo.
- Add .ico/.icns saving.
- Readd CMAKE support
- Better static lib support in MSVC++ 8/9
- Median cut quantization
- Lib EXIF support
- APNG support
- Koala support
- Flashpix (.fpx) support
- Multipage PNM files
- Fortran support
- D language support - There exists a program to convert C headers to D.
- Better DirectX 10 support
- Open URL in WindowsTest not working
- Should probably check for errors in iGetWdpHead and etc...
- ilGetAlpha returns malloc'ed memory that cannot be freed by the caller.
- Support for DX10 .dds files - one new format is found in HDRRendering sample in nVidia DX10 SDK.
- More checks for headers of unknown file types
- Set CompSize in ReadData of il_dds.c for DDS_LINEARSIZE flag?
- Look at resize filters in VTFEdit.
- Get rid of globals in il_dds.c.
- TIFF with JPEG encoding
- Bicubic scaling
- Freebasic support
- Multiple images in FITs files (through extensions).
- Reorder loading/saving in il_io.c
- Get il_vtf.c to do less memory allocation.  Just use the precache function.
- In il_gif.c, make sure subimages with offsets will not lead to an overflow.
- il_bmp.c / il_dds.c / il_dds-save.c may not be safe for Big Endian for 16-bit data.
- Change how mipmaps are stored (do not use Next pointer in ILimage).
- Check ASCII PPM code for the ability to read data that wraps around to multiple lines.
- Check return on all ilFixImage calls.
- Printing not working from Windows test.
- Can ilConvertImage deal correctly with signed image formats?
- Add function to create cubemaps.
- Request compressed texture data in ilutGLSetTex2D?
- Upload cubemaps directly to OpenGL/D3D.
- Change some calls (such as in il_tiff.c) to ilGetString to non-Unicode.
- Check saving of Luminance (8 bpp/1 Bpp) BMP
- Check compatibility with all the files in the BMP Suite.
- IL_COLOUR_INDEX:  must be implemented in iluReplaceColour(ILubyte Red, ILubyte Green, ILubyte Blue, ILfloat Tolerance);
- iluEqualize must be implemented for various Bpc
- Update the state stack in IL.  Right now, it only saves part of the states when you push/pop.
- Redo Compress in il_dds-save.c.
- Check 16-bit images being loaded have their lower bits properly set.  An example is the il_dds.c entry on 2009-01-15.
- Fix image loading problems listed on http://www.seriema.net/thumbview/index.php?page=compatibility.
- Look at http://en.wikipedia.org/wiki/Comparison_of_graphics_file_formats
- PNG and JPG compression in .bmp files?
- Scan line table for RLE .tga files.
- Allow more formats for saving .tga (does not do IL_ALPHA for instance) - also in iTargaSize.
  Make sure iConvertImage is called.  Right now, it would not be.
- XBM support.
- Check XPM support for different types: http://en.wikipedia.org/wiki/XPM_(image_format)
- Find test images for all .tpl formats.
- Create new palette types for luminance, luminance-alpha and alpha (.tpl has luminance-alpha) (also change ilutConvertToSDLSurface and maybe others).
- iGetActiveNum needs to be rewritten somehow to allow for mipmaps under a cubemap in an animation chain (extreme example).
- TPL saving
- Test Windows Mobile code.
- OpenMP support: http://stackoverflow.com/questions/326487/multithreaded-image-processing-in-c
  http://www.ddj.com/architect/184405586
- Shear rotation: http://www.codeproject.com/KB/graphics/rotatebyshear.aspx
- Icon Library (.icl) support.
- Test all DDS types.
- New (DX10) DDS types
- Gamecube BMD and BTI support (http://www.amnoid.de/gc/)
- Conversion from half to int is producing artifacts.
- Support for mp3 frames with Unicode strings


Lots of work:
- Linux Unicode support is partially broken.
- Write own .exr routines.
- Native WDP support
- Enable descriptive error strings in IL.
- Librsvg
- Give estimates for buffer sizes for ilSaveL.
- MNG saving
- Metadata
- Thread safe version
- Add flag for determining whether DDS data is outdated.
- TIFF saving to file streams and lumps
- Use fseek/ftell 64-bit?
- Check for any external library pointers that are assumed to be 32-bit.
- Possibly make palettes always 768 (or 1024) bytes so that it is not possible to access outside by bad data?
- Add padding options, so that each line could be DWORD-padded, for instance.
- Add dithering.
- Work on making DevIL more stable.  There's probably a few places where I don't check the return value of ilNewImage, for instance.
- Add support for callbacks so that users can create progress meters for loading and saving in their applications.  We could extend this idea further and allow cancellation of loading/saving images.
- global use of restricted pointers where available (for now only il_bmp.{c,h} uses it)
- iluScaleColours is BUGGED, doesn't handle the image with the correct type! more Bpc (except luminance and paletted one)
- Add a real Layer facility.
- IL_LUMINANCE_ALPHA support must be tested in ilApplyPal();
- ilApplyPal should work on any type of image, not only IL_BYTE or IL_UNSIGNED_BYTE
- Look at adding FreeType support to ILU.
- Work on the signed/unsigned conversions a bit more (IL_BYTE versus IL_UNSIGNED_BYTE, etc.).  A good link for defining signed formats is http://www.sgi.com/software/opengl/advanced98/notes/node31.html .
- Add TWAIN support.
- Add clipboard and printing support in Linux.
- Finish support for regions in ILU.
- Add support for regions in IL.  This could be used to save only a specific portion of an image.
- Add support for user-defined filters (possibly of a user-defined size).
- Add more filters.  Some sites to look at are http://www.jasonwaltman.com/thesis/introduction.html http://www.dai.ed.ac.uk/CVonline/transf.htm http://www.fortunecity.com/lavendar/kane/39/ffpg.htm and http://www.opengl.org/developers/documentation/Version1.2/1.2specs/convolution_border_modes.txt .  The last link describes ways to handle borders for the convolution filters.  It might be worthwhile to allow these kinds of options.
- Possibly add support for other colour spaces besides RGB (CMYK, YUV, etc.)
- Implement a scripting "language" with LUA for ILU.  This way, people can generate scripts to run a certain set of filters on all images.
- Look at other rescaling algorithms:
http://java.sun.com/products/java-media/jai/forDevelopers/jai1_0guide/Geom-image-manip.doc.html
http://java.sun.com/products/java-media/jai/forDevelopers/jai1_0guide/Geom-image-manip.doc.html#55431
http://www.npac.syr.edu/projects/nasa/MILOJE/final/node36.html
http://www.ultranet.com/~aramini/design.html
- ilKeyColor ilAddKey...
- Find a way to make the filters (especially the convolution filters) work in three dimensions.
- Make better versions of the DevIL ports of the NeHe tutorials.  Several of the tutorials don't call ilInit, for instance.
- Change parameters of iread to use ILsizei.
- Make sure all bugs in bug tracker are squashed.


Questions:
- Should IL_CONV_PAL convert to more than RGB?
- Possibly readd VC6 projects?
- Add BeOS support back in?
- Create Dev-C++/MingW/Cygwin .a files for releases?
- Does ilIsValidTga generate an unncessary IL_INVALID_EXTENSION error?
- Why is IL_HALF even used in ilut_directx9.c?
- Fix problems loading .dds files on Big Endian machines (still a problem or not?)
- Should we be calculating DXT endpoints before converting data to 565 format?
- Use Extreme DXT compression code?
- Support half data internally?
- Will ilDetermineType fail if file reading functions are overriden?
- Is ilSaveCHeader fine with Unicode?
- Should ilFlipSurfaceDxtcData and more work on mipmaps/cubemaps as well?
- Should ilImageToDxtcData work on cubemaps?
- Should we seek back to the initial position on filestreams?
- Should we return the number of bytes read in loading lumps?
- Should we be renormalizing float and double data in FITS files?


From old TODO:
For 1.7.0
- fixed OpenGL Loading (breaks old code which inverted the images)
- iluReplaceColour should have a color to describe which color to set.
- ilBindImage must return a result. It may fail!
- ilInit() call iSetImage0 and ilBindTemp. This shouldn't happen and image names must start from 0
- Rename iGetIntegervImage to ilGetImageIntegerv and make it public

Matej:
- Make a test suite (IL ok, now ILU tests)
- Make DevIL parts that require external libraries modular
- Make the library include files correct (fix the config.h problem)
- Look into TIFF loading (support for images with greater bpp should be possible)
- Fix examples