I posted to the list before about this, but I guess I wasn't specific enough. However has anyone else but me noticed that Export to Bitmap doesn't really work? It seems to particularly break when you use gradients with additional interpolation points at arbitrary offsets. I find this particularly amazing since obviously the program is able to rasterize to screen, however it seems incapable of redirecting this same output to a memory buffer than PNGing it. How is Inkscape going to print my drawing correctly if it can't even render it to a bitmap? I only ask this because I am making a poster, and it costs a good amount of money to print one of these suckers, I'd like to make sure that I'm going to get what I expect first. Does anyone have any idea of what is going on? I'd really appreciate any light that can be shed on the matter, thanks!
Rick Spillane wrote:
I posted to the list before about this, but I guess I wasn't specific enough. However has anyone else but me noticed that Export to Bitmap doesn't really work? It seems to particularly break when you use gradients with additional interpolation points at arbitrary offsets.
If you suspect that you have found a bug in the program you can better file it in the bug tracker (of course, first asking here or at the developers' mailing list whether it is already a know issue is appreciated, but you have already done so)
Include a small/simple example SVG that shows the problem. Maybe a screenshot of what you see in Inkscape and the exported Png as well. Point-out the differences you have found.
However has anyone else but me noticed that Export to Bitmap doesn't really work?
Nope, I export my SVGs to PNG quite often (damn browsers that don't support SVG) and it works just fine for me, even with quite complex gradients. I'm on the latest version in debian/sid (0.44-1), but 0.43 worked for me too; if you're on an SVN snapshot, perhaps you should expect it to be broken sometimes.
But, of course, do file a bug report in the tracker if it's not working for you.
I only ask this because I am making a poster, and it costs a good amount of money to print one of these suckers, I'd like to make sure that I'm going to get what I expect first.
I would hope that you check the output file before sending it off to the printers in any case.
Rick Spillane wrote:
I posted to the list before about this, but I guess I wasn't specific enough. However has anyone else but me noticed that Export to Bitmap doesn't really work? ...
On a related note, I'm trying to improve Inkscape's rendering quality a bit (nothing drastic, but the current about should look a bit better) and I've noticed that apparently Export to Bitmap doesn't use the exact same procedure as normal rendering (to screen). (I noticed because at one point I apparently broke something that didn't affect rendering to screen but caused export to bitmap to give a very obviously wrong output.) Could someone explain what's going on here?
On 7/11/06, Jasper van de Gronde <th.v.d.gronde@...226...> wrote:
Could someone explain what's going on here?
No, unless you provide a sample file and detailed explanations, best of all in the form of a bug report.
bulia byak wrote:
On 7/11/06, Jasper van de Gronde <th.v.d.gronde@...226...> wrote:
Could someone explain what's going on here?
No, unless you provide a sample file and detailed explanations, best of all in the form of a bug report.
Well, I had hoped someone would simply say "yeah, sure, of course the export to bitmap function has a slightly different output, ...", but I guess that's not the case. But before actually filing a bug report I would like to have it confirmed that it actually is a bug, as I can think of reasons why export to bitmap might produce different outputs (I could imagine some sort of caching is used for on-screen display for example, which might change the order in which the different objects are composited). Anyway, have a look at the following two images:
http://home.hccnet.nl/th.v.d.gronde/gradient-test-new.png http://home.hccnet.nl/th.v.d.gronde/gradient-test-screenshot.png
They both come from the same version of Inkscape (the modified one I just posted about in another e-mail, but 0.44 has the exact same issue) and show the same file on the same scale, yet the banding and coloring are slightly different.
On Tue, Jul 11, 2006 at 07:29:13PM +0200, Jasper van de Gronde wrote:
bulia byak wrote:
On 7/11/06, Jasper van de Gronde <th.v.d.gronde@...226...> wrote:
Could someone explain what's going on here?
No, unless you provide a sample file and detailed explanations, best of all in the form of a bug report.
Well, I had hoped someone would simply say "yeah, sure, of course the export to bitmap function has a slightly different output, ...", but I guess that's not the case. But before actually filing a bug report I would like to have it confirmed that it actually is a bug, as I can think of reasons why export to bitmap might produce different outputs (I could imagine some sort of caching is used for on-screen display for example, which might change the order in which the different objects are composited). Anyway, have a look at the following two images:
http://home.hccnet.nl/th.v.d.gronde/gradient-test-new.png http://home.hccnet.nl/th.v.d.gronde/gradient-test-screenshot.png
They both come from the same version of Inkscape (the modified one I just posted about in another e-mail, but 0.44 has the exact same issue) and show the same file on the same scale, yet the banding and coloring are slightly different.
Oh, I think I see what's going on. Inkscape has a defined "canvas area", as indicated by the document outline thingee. Anything overlapping or extending beyond that box is clipped when you export. In your screenshot it's clear you can see there is portions of the document extending beyond this boundary; it looks like inside the boundary everything is identical for both images.
Possibly you disagree with this approach, but it is only the default. Inkscape has some controls to let you disable/alter this behavior:
* In Document Preferences -> Page, adjust Show canvas border, show page shadow, etc. as suits you.
* In export bitmap, experiment with exporting page vs. drawing vs. selection vs. custom. This will allow you to export stuff that extends beyond the normal page boundary, or to clip it down to specific objects or specific dimensions of the drawing, or whatever you desire.
Hope this helps, thanks for posting the gfx. Bryce
Bryce Harrington wrote:
On Tue, Jul 11, 2006 at 07:29:13PM +0200, Jasper van de Gronde wrote:
bulia byak wrote:
On 7/11/06, Jasper van de Gronde <th.v.d.gronde@...226...> wrote:
... http://home.hccnet.nl/th.v.d.gronde/gradient-test-new.png http://home.hccnet.nl/th.v.d.gronde/gradient-test-screenshot.png
They both come from the same version of Inkscape (the modified one I just posted about in another e-mail, but 0.44 has the exact same issue) and show the same file on the same scale, yet the banding and coloring are slightly different.
Oh, I think I see what's going on. Inkscape has a defined "canvas area", as indicated by the document outline thingee. Anything overlapping or extending beyond that box is clipped when you export. In your screenshot it's clear you can see there is portions of the document extending beyond this boundary; it looks like inside the boundary everything is identical for both images.
This is weird! When looking at the images with Firefox they indeed appear to be (virtually) identical (apart from the border of course, I just clipped a bit rough), but when I look at images with IrfanView (or Corel PhotoPaint, I didn't try any other programs) they seem to be different... (In fact, to make sure I wasn't going crazy I checked the histograms, after cropping properly naturally, the mean value is off by about 10.) Any gamma values or something similar being written to the PNG file?
I'm using a color profile for my monitor btw, but I don't think that should matter in this case (should affect both images in the same way when displaying them and shouldn't be affecting the RGB values reported for the image).
On Tue, Jul 11, 2006 at 11:31:56PM +0200, Jasper van de Gronde wrote:
Bryce Harrington wrote:
On Tue, Jul 11, 2006 at 07:29:13PM +0200, Jasper van de Gronde wrote:
bulia byak wrote:
On 7/11/06, Jasper van de Gronde <th.v.d.gronde@...226...> wrote:
... http://home.hccnet.nl/th.v.d.gronde/gradient-test-new.png http://home.hccnet.nl/th.v.d.gronde/gradient-test-screenshot.png
They both come from the same version of Inkscape (the modified one I just posted about in another e-mail, but 0.44 has the exact same issue) and show the same file on the same scale, yet the banding and coloring are slightly different.
Oh, I think I see what's going on. Inkscape has a defined "canvas area", as indicated by the document outline thingee. Anything overlapping or extending beyond that box is clipped when you export. In your screenshot it's clear you can see there is portions of the document extending beyond this boundary; it looks like inside the boundary everything is identical for both images.
This is weird! When looking at the images with Firefox they indeed appear to be (virtually) identical (apart from the border of course, I just clipped a bit rough), but when I look at images with IrfanView (or Corel PhotoPaint, I didn't try any other programs) they seem to be different... (In fact, to make sure I wasn't going crazy I checked the histograms, after cropping properly naturally, the mean value is off by about 10.) Any gamma values or something similar being written to the PNG file?
I'm using a color profile for my monitor btw, but I don't think that should matter in this case (should affect both images in the same way when displaying them and shouldn't be affecting the RGB values reported for the image).
Hmm, well perhaps the tool you used to create the screen capture is introducing the variance? Not really sure.
Bryce
Jasper van de Gronde wrote:
This is weird! When looking at the images with Firefox they indeed appear to be (virtually) identical (apart from the border of course, I just clipped a bit rough), but when I look at images with IrfanView (or Corel PhotoPaint, I didn't try any other programs) they seem to be different...
To (hopefully) illustrate what I mean I've uploaded a screenshot of gradient-test-new shown from within IrfanView as well as Firefox (so to be absolutely clear, you're looking at the exact same image twice, one that was generated with export to bitmap):
http://home.hccnet.nl/th.v.d.gronde/gradient-test-screenshot2.png
Jasper van de Gronde wrote:
Bryce Harrington wrote:
On Tue, Jul 11, 2006 at 07:29:13PM +0200, Jasper van de Gronde wrote:
bulia byak wrote:
On 7/11/06, Jasper van de Gronde <th.v.d.gronde@...226...> wrote:
... http://home.hccnet.nl/th.v.d.gronde/gradient-test-new.png http://home.hccnet.nl/th.v.d.gronde/gradient-test-screenshot.png
They both come from the same version of Inkscape (the modified one I just posted about in another e-mail, but 0.44 has the exact same issue) and show the same file on the same scale, yet the banding and coloring are slightly different.
Oh, I think I see what's going on. Inkscape has a defined "canvas area", as indicated by the document outline thingee. Anything overlapping or extending beyond that box is clipped when you export. In your screenshot it's clear you can see there is portions of the document extending beyond this boundary; it looks like inside the boundary everything is identical for both images.
This is weird! When looking at the images with Firefox they indeed appear to be (virtually) identical (apart from the border of course, I just clipped a bit rough), but when I look at images with IrfanView (or Corel PhotoPaint, I didn't try any other programs) they seem to be different... (In fact, to make sure I wasn't going crazy I checked the histograms, after cropping properly naturally, the mean value is off by about 10.) Any gamma values or something similar being written to the PNG file?
Wait a second... it looks like there's transparency involved in the object the gradient is applied to and the canvas is most likely transparent as well... is this possibly the issue with the PNG export?
In Inkscape it will render on the white (transparent) background, but, the PNG would just have transparency w/o that "artificial" white. I guess the quick test is to make the background opaque in the document properties and test exporting to PNG again.
-Josh
On Tue, Jul 11, 2006 at 11:40:34PM +0200, Jasper van de Gronde wrote:
Jasper van de Gronde wrote:
This is weird! When looking at the images with Firefox they indeed appear to be (virtually) identical (apart from the border of course, I just clipped a bit rough), but when I look at images with IrfanView (or Corel PhotoPaint, I didn't try any other programs) they seem to be different...
To (hopefully) illustrate what I mean I've uploaded a screenshot of gradient-test-new shown from within IrfanView as well as Firefox (so to be absolutely clear, you're looking at the exact same image twice, one that was generated with export to bitmap):
http://home.hccnet.nl/th.v.d.gronde/gradient-test-screenshot2.png
I've no idea what causes that. I would suggest posting this with a subject line more appropriate than just "Major Issue", and hope someone who understands the color stuff will answer. Otherwise, send in a bug report.
It's not clear to me whether the issue is actually with Inkscape, or with one of these other viewing programs.
Bryce
On Tue, Jul 11, 2006 at 02:42:31PM -0700, Joshua A. Andler wrote:
Wait a second... it looks like there's transparency involved in the object the gradient is applied to and the canvas is most likely transparent as well... is this possibly the issue with the PNG export?
In Inkscape it will render on the white (transparent) background, but, the PNG would just have transparency w/o that "artificial" white. I guess the quick test is to make the background opaque in the document properties and test exporting to PNG again.
It sounds like Joshua may have your answer. I do my drawing on a partially transparent pink background so I can pick out weird errors like that right away :)
Jeff
On Tue, July 11, 2006 10:31 pm, Jasper van de Gronde wrote:
Bryce Harrington wrote:
On Tue, Jul 11, 2006 at 07:29:13PM +0200, Jasper van de Gronde wrote:
bulia byak wrote:
On 7/11/06, Jasper van de Gronde <th.v.d.gronde@...226...> wrote:
... http://home.hccnet.nl/th.v.d.gronde/gradient-test-new.png http://home.hccnet.nl/th.v.d.gronde/gradient-test-screenshot.png
They both come from the same version of Inkscape (the modified one I just posted about in another e-mail, but 0.44 has the exact same issue) and show the same file on the same scale, yet the banding and coloring are slightly different.
Oh, I think I see what's going on. Inkscape has a defined "canvas area", as indicated by the document outline thingee. Anything overlapping or extending beyond that box is clipped when you export. In your screenshot it's clear you can see there is portions of the document extending beyond this boundary; it looks like inside the boundary everything is identical for both images.
This is weird! When looking at the images with Firefox they indeed appear to be (virtually) identical (apart from the border of course, I just clipped a bit rough), but when I look at images with IrfanView (or Corel PhotoPaint, I didn't try any other programs) they seem to be different... (In fact, to make sure I wasn't going crazy I checked the histograms, after cropping properly naturally, the mean value is off by about 10.) Any gamma values or something similar being written to the PNG file?
Yes, PNG contains gamma information. And not all applications use (write or read) it, which is probably why you obtain different results with different applications.
Hope this helps.
Fred
Joshua A. Andler wrote:
... Wait a second... it looks like there's transparency involved in the object the gradient is applied to and the canvas is most likely transparent as well... is this possibly the issue with the PNG export?
In Inkscape it will render on the white (transparent) background, but, the PNG would just have transparency w/o that "artificial" white. I guess the quick test is to make the background opaque in the document properties and test exporting to PNG again.
That was it. After making the background opaque the exported bitmap displayed exactly the same in IrfanView as the original in Inkscape.
The svg file that originally had this "problem" when rendering to a png will indeed not be viewable from IrfanView, but be viewable from Firefox. As for the svg file itself, it still isn't viewable on other svg viewers, even Batik. This probably isn't a problem so much with Inkscape as it is with other svg "compliant" software. However if Inkscape is supposed to produce svg which is viewable on other viewers, then this may be a problem. If I should submit a bug report, I would be happy to do so, and include the svg file causing the problem.
On 7/11/06, Jasper van de Gronde <th.v.d.gronde@...226...> wrote:
Joshua A. Andler wrote:
... Wait a second... it looks like there's transparency involved in the object the gradient is applied to and the canvas is most likely transparent as well... is this possibly the issue with the PNG export?
In Inkscape it will render on the white (transparent) background, but, the PNG would just have transparency w/o that "artificial" white. I guess the quick test is to make the background opaque in the document properties and test exporting to PNG again.
That was it. After making the background opaque the exported bitmap displayed exactly the same in IrfanView as the original in Inkscape.
Using Tomcat but need to do more? Need to support web services, security? Get stuff done quickly with pre-integrated technology to make your job easier Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&da... _______________________________________________ Inkscape-user mailing list Inkscape-user@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/inkscape-user
Rick Spillane wrote:
The svg file that originally had this "problem" when rendering to a png will indeed not be viewable from IrfanView, but be viewable from Firefox. As for the svg file itself, it still isn't viewable on other svg viewers, even Batik. This probably isn't a problem so much with Inkscape as it is with other svg "compliant" software. However if Inkscape is supposed to produce svg which is viewable on other viewers, then this may be a problem. If I should submit a bug report, I would be happy to do so, and include the svg file causing the problem.
Outside of Inkscape I was just viewing a PNG file. The "problem" was that Inkscape treats the transparency of the background slightly different when exporting then when rendering to screen (and with good cause I might add). On-screen a background necessarily has some opaque color (well, theoretically Inkscape might be able to make the background transparent, but I doubt anyone would appreciate that), when exporting this doesn't have to be the case. And apparently IrfanView simply discards the alpha channel or something, so if anything it's probably a problem with IrfanView and not Inkscape.
On Wed, Jul 12, 2006 at 09:59:49AM +0200, Jasper van de Gronde wrote:
Outside of Inkscape I was just viewing a PNG file. The "problem" was that Inkscape treats the transparency of the background slightly different when exporting then when rendering to screen (and with good cause I might add). On-screen a background necessarily has some opaque color (well, theoretically Inkscape might be able to make the background transparent, but I doubt anyone would appreciate that), when exporting this doesn't have to be the case. And apparently IrfanView simply discards the alpha channel or something, so if anything it's probably a problem with IrfanView and not Inkscape.
I personally would like the grid option from GIMP for the canvas.
At 09.59 12/07/2006, Jasper van de Gronde wrote:
Rick Spillane wrote:
The svg file that originally had this "problem" . . . [ cut ]
. . . [ cut ] apparently IrfanView simply discards the alpha channel or something, so if anything it's probably a problem with IrfanView and not Inkscape.
Yea! It's true ! i.e. if you load your png with the Ms Image viewer you can see its own trasparency . The problem is twice 'cause Irfanview lost this effect also while printing.
We need to remember this and maybe drop two lines to Irfan Skiljan :-)
R.
participants (10)
-
Bryce Harrington -
bulia byak -
Daniel Hulme -
Fred Labrosse -
Jasper van de Gronde -
Jeffrey Brent McBeth -
Joshua A. Andler -
Rick Spillane -
RoSourceforge -
Ulf Erikson