Kernel Unit Length in Lighting Effects
Hello
Working on filters for one moment now I was asking myself if there are any progresses (or projects) about implementing Kernel Unit Length in Lighting Effects.
Following this Niko Kiirala tutorial http://fedev.blogspot.com/2008/07/filter-effects-lighting-effects.html "Kernel Unit Length is not yet used in Inkscape. It is used to make the result resolution-independent. Now that it isn't used, with high zooms you can see odd plateaus forming."
If that's not so important in daily web graphics, from an illustrator point of view that's a problem not only for zooming but also for scaling up some effects or using wide blurs as basic composites for lighting.
Thanks for any info and thanks for so much fun with filter effects !
ivan
Mon, 9 Mar 2009 22:14:46 +0000 (GMT) Ivan Louette <ivan_louette@...48...> kirjoitti:
Working on filters for one moment now I was asking myself if there are any progresses (or projects) about implementing Kernel Unit Length in Lighting Effects.
Following this Niko Kiirala tutorial http://fedev.blogspot.com/2008/07/filter-effects-lighting-effects.html "Kernel Unit Length is not yet used in Inkscape. It is used to make the result resolution-independent. Now that it isn't used, with high zooms you can see odd plateaus forming."
If that's not so important in daily web graphics, from an illustrator point of view that's a problem not only for zooming but also for scaling up some effects or using wide blurs as basic composites for lighting.
Well, I can tell I've got no specific plans to implementing that. Yes, it's part of Filter Effects implementation that's still unimplemented and there are some other parts too...
Besides, that's not quite the full story about those plateaus. The other part is: there are only 256 height levels available and with large resolutions we'll inevitably run out of those. Kernel Unit Length just allows for lowering the resolution - well, kind-of.
Thanks a lot for your reply !
If I understand well the 256 height levels available cause the same kind of problem encountered with 3D displacement maps and which was solved by the use of 16 and 32 bits grey levels maps.
Are those 256 height levels maps a limitation of SVG ?
ivan
________________________________ De : Niko Kiirala <niko@...1267...> À : inkscape-devel@lists.sourceforge.net Envoyé le : Mardi, 10 Mars 2009, 12h39mn 37s Objet : Re: [Inkscape-devel] Kernel Unit Length in Lighting Effects
Mon, 9 Mar 2009 22:14:46 +0000 (GMT) Ivan Louette <ivan_louette@...48...> kirjoitti:
Working on filters for one moment now I was asking myself if there are any progresses (or projects) about implementing Kernel Unit Length in Lighting Effects.
Following this Niko Kiirala tutorial http://fedev.blogspot.com/2008/07/filter-effects-lighting-effects.html "Kernel Unit Length is not yet used in Inkscape. It is used to make the result resolution-independent. Now that it isn't used, with high zooms you can see odd plateaus forming."
If that's not so important in daily web graphics, from an illustrator point of view that's a problem not only for zooming but also for scaling up some effects or using wide blurs as basic composites for lighting.
Well, I can tell I've got no specific plans to implementing that. Yes, it's part of Filter Effects implementation that's still unimplemented and there are some other parts too...
Besides, that's not quite the full story about those plateaus. The other part is: there are only 256 height levels available and with large resolutions we'll inevitably run out of those. Kernel Unit Length just allows for lowering the resolution - well, kind-of.
Tue, 10 Mar 2009 12:37:01 +0000 (GMT) Ivan Louette <ivan_louette@...48...> kirjoitti:
If I understand well the 256 height levels available cause the same kind of problem encountered with 3D displacement maps and which was solved by the use of 16 and 32 bits grey levels maps.
Are those 256 height levels maps a limitation of SVG ?
I believe what you're describing is exactly the same problem. Input image for lighting effects is essentially a height map, which in turn is a special case of 3D displacement map.
256 levels is not a limitation of SVG itself, but a limitation of Inkscape rendering engine. Using more than 8 bits per colour component would help with this problem, but likely would require quite a bit of work.
Thanks again !
Then I suppose that some banding problems in blur and gradients are of a similar origin relative to the rendering engine and they will be solved together in the future.
ivan
________________________________ De : Niko Kiirala <niko@...1267...> À : inkscape-devel@lists.sourceforge.net Envoyé le : Mardi, 10 Mars 2009, 14h02mn 54s Objet : Re: [Inkscape-devel] Re : Kernel Unit Length in Lighting Effects
Tue, 10 Mar 2009 12:37:01 +0000 (GMT) Ivan Louette <ivan_louette@...48...> kirjoitti:
If I understand well the 256 height levels available cause the same kind of problem encountered with 3D displacement maps and which was solved by the use of 16 and 32 bits grey levels maps.
Are those 256 height levels maps a limitation of SVG ?
I believe what you're describing is exactly the same problem. Input image for lighting effects is essentially a height map, which in turn is a special case of 3D displacement map.
256 levels is not a limitation of SVG itself, but a limitation of Inkscape rendering engine. Using more than 8 bits per colour component would help with this problem, but likely would require quite a bit of work.
Ivan Louette wrote:
Thanks again !
Then I suppose that some banding problems in blur and gradients are of a similar origin relative to the rendering engine and they will be solved together in the future.
You may want to make sure a feature request is filed under https://blueprints.launchpad.net/inkscape
In any case, as Niko mentioned, it will probably require quite some work, so you probably shouldn't hold your breath. But if you have any specific problems, be sure to mention them, it can also be due to "sloppy" coding (for example, some years back Inkscape used to round more often than necessary in some places).
Thanks !
I would so much being able to do the work myself or at least to contribute !!! But at the moment my technical skills don't overpass what I am doing in SVG (just around filters).
I will see tomorrow or later to fill a feature request. I didn't see anything there about this question concerning the rendering engine.
ivan
________________________________ De : Jasper van de Gronde <th.v.d.gronde@...528...> À : Ivan Louette <ivan_louette@...48...> Cc : Niko Kiirala <niko@...1267...>; inkscape-devel@lists.sourceforge.net Envoyé le : Mercredi, 11 Mars 2009, 14h25mn 53s Objet : Re: [Inkscape-devel] Re : Re : Kernel Unit Length in Lighting Effects
Ivan Louette wrote:
Thanks again !
Then I suppose that some banding problems in blur and gradients are of a similar origin relative to the rendering engine and they will be solved together in the future.
You may want to make sure a feature request is filed under https://blueprints.launchpad.net/inkscape
In any case, as Niko mentioned, it will probably require quite some work, so you probably shouldn't hold your breath. But if you have any specific problems, be sure to mention them, it can also be due to "sloppy" coding (for example, some years back Inkscape used to round more often than necessary in some places).
On Tue, Mar 10, 2009 at 2:39 PM, Niko Kiirala wrote:
Besides, that's not quite the full story about those plateaus. The other part is: there are only 256 height levels available and with large resolutions we'll inevitably run out of those.
Is this why we see "topographic map" effect?
Alexandre
participants (4)
-
Alexandre Prokoudine -
Ivan Louette -
Jasper van de Gronde -
Niko Kiirala