Am 28.09.07, 20:31 +0200 schrieb Cyrille Berger: > > > I don't know about ICC enought to comment. But HTML do not allow > > > extension on the spec, it's just that some main stream web browsers have > > > extension to the spec, meaning some websites use them, meaning other web > > > browsers have to cope finding information for the extension, while 99% of > > > what is in the extension could have been done with a full implementation > > > of the spec. TIFF and OpenEXR, I can't count people asking us to support > > > one of this undocumented extension. > > > > In text formats, if no other meachanism is provided, often new things > > start to appear in comments ;-) > yeah :) or tags are added anyway. But I am offering a mechanism, comes here > to > ask for the change !
You mean to include in the spec? > > For instance the ICC profile in Tiff (v6.0 1992) is defined in the ICC > > spec (1995?) not vice versa. Tiff was easy enough extensible to do such > > things, Exif, geo referencing ... Ok, Adobe registers new TIFF tags on > > request. But this makes no new standard. > Sure and they don't require documentation, so half of those tags are only > used > by a few applications. This is always the case. But better having a common base than people switching to something completely different. For ICC it is documented and well supported. > > Blender with its layer and compositing stack, exhibited in the OpenEXR > > images is nowhere described in the OpenEXR spec. Even though it is > > possible with this format. > Yes and now people are asking for Krita to support multilayer OpenEXR and I > have a hard time finding information. While with a central place with all the You can step throu CinePaint's OpenEXR plug-in since 0.22-x. Basically one have to parse channel names and point '.' separators and form the layers. I putted everything in a layer and omitted the inbetween structure. It is just saved later again. It is not a 1:1 mapping. What problems do you see with that? > information it's much easier ;) And from what I have heard, the multi layer > things is supposed to move into the spec. Interessting > I could also add as an example, Exif, which includes a field where > manufacturer are allowed to insert whatever data they want, and some of them > are inserting metadata in that field which could be described in the spec. That behaviour is a big problem general with camera workflows. The software makes often a difference. So companies feel a need to protect their buissness knowledge. As soon as a good common place software makes such differences equal, the secrets become nonsense. The same with many other workflows. > I do think the create group is a reasonnable group so that if someone wants > something, I think it will be easily added. This would lead to a basic and a extended section in the specification. A basic reader has to support just sRGB and names. A extended reader/writer has all possibilities and should write only understood things inside the colour descriptions. kind regards Kai-Uwe Behrmann -- developing for colour management www.behrmann.name + www.oyranos.org + www.cinepaint.org _______________________________________________ CREATE mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/create
