[Gimp-developer] Re: [Gimp-user] 2.4 schedule estimate?
Hi Simon, Selon Simon Roberts [EMAIL PROTECTED]: Having written some software in the past, I know this is an unfair question, but I'm interested in the color management that's slated for 2.4. Can anyone point me at a time (rough waving damp fingers in the air) estimate for when that might be released? gimp-developer is a more appropriate forum for the question (although I hope that everyone there is also here). I haven't been involved in GIMP development for a while now, so I can't really tell. Alternatively and also, anyone using the 2.3 version (described as unstable) which is available? Is it reasonable to try to use that? I would say yes. It's not feature frozen, but it's already quite usable. Does the color management part of it work? And is the CM part likely to remain constant, or is it subject to total rewrite, rendering my learning and calibrations/profiling wasted? It's pretty much finished, as far as I know. Cheers, Dave. -- Dave Neary Lyon, France ___ Gimp-developer mailing list Gimp-developer@lists.XCF.Berkeley.EDU https://lists.XCF.Berkeley.EDU/mailman/listinfo/gimp-developer
Re: [Gimp-developer] Re: [Gimp-user] 2.4 schedule estimate?
On Wednesday 18 January 2006 07:44 am, Dave Neary wrote: Hi Simon, Selon Simon Roberts [EMAIL PROTECTED]: Having written some software in the past, I know this is an unfair question, but I'm interested in the color management that's slated for 2.4. Can anyone point me at a time (rough waving damp fingers in the air) estimate for when that might be released? gimp-developer is a more appropriate forum for the question (although I hope that everyone there is also here). I haven't been involved in GIMP development for a while now, so I can't really tell. Alternatively and also, anyone using the 2.3 version (described as unstable) which is available? Is it reasonable to try to use that? I would say yes. It's not feature frozen, but it's already quite usable. Does the color management part of it work? And is the CM part likely to remain constant, or is it subject to total rewrite, rendering my learning and calibrations/profiling wasted? It's pretty much finished, as far as I know. I don't know how far along this is in CVS but the last snap shot I looked at, 2.3.6, still needed significant work to have usable color management facilities. You will definitely notice that there are significant new features and what is there is working. But the following (from memory so may not be complete) still needs to be done as of 2.3.6: 1. There is only limited support for embedded profiles. Only png files worked. So tiff and jpeg users will have limited functionality. 2. Embedding profiles and doing color space conversions is not supported. So you must use an external application for this such as lcms or ImageMagick 3. It is currently coded to use the default RGB profile as the source profile for all images. I don't know if this the case with pgn files since it will - load and save embedded ICC profiles from/to PNG images. But this does allow users of jpeg and tiff images to use what is there. You just have to manually set the correct source profile for your image in the preferences dialog. This is a little confusing if you don't know what is going on but this is apparently for testing during development and this will not be how this will work when this is released. 4. There are no facilities to handle CM during the printing process. I am looking forward to the next snap shot. Hopefully it will have a more complete CM implementation. Cheers, Dave. -- Dave Neary Lyon, France ___ Gimp-developer mailing list Gimp-developer@lists.XCF.Berkeley.EDU https://lists.XCF.Berkeley.EDU/mailman/listinfo/gimp-developer ___ Gimp-developer mailing list Gimp-developer@lists.XCF.Berkeley.EDU https://lists.XCF.Berkeley.EDU/mailman/listinfo/gimp-developer
Re: [Gimp-developer] python plugin
Hi, Emanuele Zattin [EMAIL PROTECTED] writes: my name is Emanuele Zattin and i'm developing a plug-in implementing a texture synthesis algorithm described in a paper (Texture Synthesis by Non-parametric sampling by Efros and Leung). Here attached you can see my actual implementation which works on grayscale images. It is anyway VERY slow, which was anyway predictable judging from the complexity of the algorithm. I was wondering if translating it to C would make it much faster or if there are some more speedup techniques i'm just missing. You could try to speed it up by optimizing your pixel accessors. Perhaps use a RowIterator as is done in the colorxhtml plug-in? Sven ___ Gimp-developer mailing list Gimp-developer@lists.XCF.Berkeley.EDU https://lists.XCF.Berkeley.EDU/mailman/listinfo/gimp-developer
Re: [Gimp-developer] Re: [Gimp-user] 2.4 schedule estimate?
Hi, Hal V. Engel [EMAIL PROTECTED] writes: I don't know how far along this is in CVS but the last snap shot I looked at, 2.3.6, still needed significant work to have usable color management facilities. That is correct, it is far from being finished. 1. There is only limited support for embedded profiles. Only png files worked. So tiff and jpeg users will have limited functionality. IIRC, the support in the TIFF and JPEG plug-ins is as good or bad as in the PNG plug-in. What makes you think that only PNG would work? 2. Embedding profiles and doing color space conversions is not supported. So you must use an external application for this such as lcms or ImageMagick This is definitely on the TODO. 3. It is currently coded to use the default RGB profile as the source profile for all images. I don't know if this the case with pgn files since it will - load and save embedded ICC profiles from/to PNG images. But this does allow users of jpeg and tiff images to use what is there. You just have to manually set the correct source profile for your image in the preferences dialog. This is a little confusing if you don't know what is going on but this is apparently for testing during development and this will not be how this will work when this is released. This will change, the display filter modules cannot yet access any image properties so the embedded color profile is currently being ignored. 4. There are no facilities to handle CM during the printing process. That's left to the print plug-ins but shouldn't be hard to add. I am looking forward to the next snap shot. Hopefully it will have a more complete CM implementation. Noone is actively working on this, so it is unlikely that the next snapshot will improve the situation. I hope to be able to start working on it at some point, but I can't promise anything and I would prefer if I could just help someone else to finish this. Sven ___ Gimp-developer mailing list Gimp-developer@lists.XCF.Berkeley.EDU https://lists.XCF.Berkeley.EDU/mailman/listinfo/gimp-developer
Re: [Gimp-developer] Re: [Gimp-user] 2.4 schedule estimate?
On Wednesday 18 January 2006 11:47 am, Sven Neumann wrote: Hi, Hal V. Engel [EMAIL PROTECTED] writes: I don't know how far along this is in CVS but the last snap shot I looked at, 2.3.6, still needed significant work to have usable color management facilities. That is correct, it is far from being finished. 1. There is only limited support for embedded profiles. Only png files worked. So tiff and jpeg users will have limited functionality. IIRC, the support in the TIFF and JPEG plug-ins is as good or bad as in the PNG plug-in. What makes you think that only PNG would work? Only that the change log only lists PNG as having the ability to load and save embedded profiles. Since the display filters don't use the embedded profiles users have no way to know what it working and what is not with regard to embedded profiles. So I only have the change log to go by which only lists PGN files and not TIFF or JPEG as being able to read and write embedded profiles. 2. Embedding profiles and doing color space conversions is not supported. So you must use an external application for this such as lcms or ImageMagick This is definitely on the TODO. 3. It is currently coded to use the default RGB profile as the source profile for all images. I don't know if this the case with pgn files since it will - load and save embedded ICC profiles from/to PNG images. But this does allow users of jpeg and tiff images to use what is there. You just have to manually set the correct source profile for your image in the preferences dialog. This is a little confusing if you don't know what is going on but this is apparently for testing during development and this will not be how this will work when this is released. This will change, the display filter modules cannot yet access any image properties so the embedded color profile is currently being ignored. The display filter is probably the most difficult part of this to implement and it is close as it only needs to start accessing the embedded profiles. 4. There are no facilities to handle CM during the printing process. That's left to the print plug-ins but shouldn't be hard to add. Even if this didn't make it into the print plug-in if #2 (above) is implemented users will have a usable work around. I am looking forward to the next snap shot. Hopefully it will have a more complete CM implementation. Noone is actively working on this, so it is unlikely that the next snapshot will improve the situation. I hope to be able to start working on it at some point, but I can't promise anything and I would prefer if I could just help someone else to finish this. Sven ___ Gimp-developer mailing list Gimp-developer@lists.XCF.Berkeley.EDU https://lists.XCF.Berkeley.EDU/mailman/listinfo/gimp-developer
Re: [Gimp-developer] Re: [Gimp-user] 2.4 schedule estimate?
Hi, Hal V. Engel wrote: Only that the change log only lists PNG as having the ability to load and save embedded profiles. Since the display filters don't use the embedded profiles users have no way to know what it working and what is not with regard to embedded profiles. So I only have the change log to go by which only lists PGN files and not TIFF or JPEG as being able to read and write embedded profiles. I believe the TIFF (and JPEG?) loaders have been loading and saving embedded profiles as parasites for some time. They've just been ignored in all other regards. The display filter is probably the most difficult part of this to implement and it is close as it only needs to start accessing the embedded profiles. The filter itself is relatively trivial - the existing proof filter needs only minimal changes to do the job. What will be trickier is developing profile- and transform-caching strategies. The display filter is effectively log-jammed, however, until it becomes possible to fetch a reference to the image from within the filter. That's left to the print plug-ins but shouldn't be hard to add. Indeed. Once there's an infrastructure for plugins to query the GIMP for the profiles to be used for a particular task, the rest becomes relatively trivial. All the best, -- Alastair M. Robinson ___ Gimp-developer mailing list Gimp-developer@lists.XCF.Berkeley.EDU https://lists.XCF.Berkeley.EDU/mailman/listinfo/gimp-developer
