GEGL Gaussian Blur (Gimp/gegl git)
This discussion is connected to the gimp-developer-list.gnome.org mailing list which is provided by the GIMP developers and not related to gimpusers.com.
This is a read-only list on gimpusers.com so this discussion thread is read-only, too.
GEGL Gaussian Blur (Gimp/gegl git) | Partha Bagchi | 12 Oct 04:21 |
GEGL Gaussian Blur (Gimp/gegl git) | Partha Bagchi | 12 Oct 11:47 |
GEGL Gaussian Blur (Gimp/gegl git) | Partha Bagchi | 27 Oct 21:44 |
GEGL Gaussian Blur (Gimp/gegl git) | Victor Oliveira | 28 Oct 03:56 |
GEGL Gaussian Blur (Gimp/gegl git)
I am finding that one possible show stopper is the gegl gaussian blur.
Here is a simple example. Open a 16bit, say 12 MP tiff image. Duplicate Layer, and run gegl gaussian blur with x=y=100. On my Macbook Pro with 8G RAM, either I will get vertical lines on the layer (at best) or crash my Mac.
Is there a way to speed up gaussian blur or is this a Mac issue and never happens on Linux?
Thanks,
Partha
GEGL Gaussian Blur (Gimp/gegl git)
Further information to my post. I built on Windows 7 64-bit and here I have 16G of RAM with dedicated Nvidia graphics. It crashes my video driver (which is able to recover) and then Gimp dies with this following. This is after graphics card recovery and then opening c2g.
Here is the error message:
bps: 16
load_contiguous
bytes_per_pixel: 6, format: 6
file-tiff-load.exe is updating the progress too often
Parsing
'C:\Users\partha\AppData\Roaming\GIMP\2.9\filters\GimpGegl-gegl-gaussian-blur-config.settings'
(gimp-2.9.exe:9400): GEGL-gaussian-blur.c-WARNING **: Error in gaussian-blur.c:519@cl_gaussian_blur - mem object allocation failure
Writing
'C:\Users\partha\AppData\Roaming\GIMP\2.9\filters\GimpGegl-gegl-gaussian-blur-config.settings'
Parsing
'C:\Users\partha\AppData\Roaming\GIMP\2.9\filters\GimpGegl-gegl-c2g-config.settings'
(script-fu.exe:6928): LibGimpBase-WARNING **: script-fu.exe: gimp_wire_read(): error
On Sat, Oct 12, 2013 at 12:21 AM, Partha Bagchi wrote:
I am finding that one possible show stopper is the gegl gaussian blur.
Here is a simple example. Open a 16bit, say 12 MP tiff image. Duplicate Layer, and run gegl gaussian blur with x=y=100. On my Macbook Pro with 8G RAM, either I will get vertical lines on the layer (at best) or crash my Mac.
Is there a way to speed up gaussian blur or is this a Mac issue and never happens on Linux?
Thanks,
Partha
GEGL Gaussian Blur (Gimp/gegl git)
Victor must have done some magic since he sent me an email saying that he was looking into the OpenCL bug. I can now say that this is gone and in fact GEGL is considerably more stable. I can actually work on 16-bit tiffs without crashing! :)
I was so excited about this that I rebuilt both Windows and Mac and posted on my website.
Thanks devs!
Partha
On Sat, Oct 12, 2013 at 7:47 AM, Partha Bagchi wrote:
Further information to my post. I built on Windows 7 64-bit and here I have 16G of RAM with dedicated Nvidia graphics. It crashes my video driver (which is able to recover) and then Gimp dies with this following. This is after graphics card recovery and then opening c2g.
Here is the error message:
bps: 16 load_contiguous
bytes_per_pixel: 6, format: 6
file-tiff-load.exe is updating the progress too often Parsing
'C:\Users\partha\AppData\Roaming\GIMP\2.9\filters\GimpGegl-gegl-gaussian-blur-config.settings'(gimp-2.9.exe:9400): GEGL-gaussian-blur.c-WARNING **: Error in gaussian-blur.c:519@cl_gaussian_blur - mem object allocation failure
Writing 'C:\Users\partha\AppData\Roaming\GIMP\2.9\filters\GimpGegl-gegl-gaussian-blur-config.settings' Parsing
'C:\Users\partha\AppData\Roaming\GIMP\2.9\filters\GimpGegl-gegl-c2g-config.settings'(script-fu.exe:6928): LibGimpBase-WARNING **: script-fu.exe: gimp_wire_read(): error
On Sat, Oct 12, 2013 at 12:21 AM, Partha Bagchi wrote:
I am finding that one possible show stopper is the gegl gaussian blur.
Here is a simple example. Open a 16bit, say 12 MP tiff image. Duplicate Layer, and run gegl gaussian blur with x=y=100. On my Macbook Pro with 8G RAM, either I will get vertical lines on the layer (at best) or crash my Mac.
Is there a way to speed up gaussian blur or is this a Mac issue and never happens on Linux?
Thanks,
Partha
GEGL Gaussian Blur (Gimp/gegl git)
I actually was looking at it when I realized Daniel solved the problem :)
On Sun, Oct 27, 2013 at 2:44 PM, Partha Bagchi wrote:
Victor must have done some magic since he sent me an email saying that he was looking into the OpenCL bug. I can now say that this is gone and in fact GEGL is considerably more stable. I can actually work on 16-bit tiffs without crashing! :)
I was so excited about this that I rebuilt both Windows and Mac and posted on my website.
Thanks devs!
Partha
On Sat, Oct 12, 2013 at 7:47 AM, Partha Bagchi wrote:
Further information to my post. I built on Windows 7 64-bit and here I have 16G of RAM with dedicated Nvidia graphics. It crashes my video driver (which is able to recover) and then Gimp dies with this following. This is after graphics card recovery and then opening c2g.
Here is the error message:
bps: 16 load_contiguous
bytes_per_pixel: 6, format: 6
file-tiff-load.exe is updating the progress too often Parsing
'C:\Users\partha\AppData\Roaming\GIMP\2.9\filters\GimpGegl-gegl-gaussian-blur-config.settings'(gimp-2.9.exe:9400): GEGL-gaussian-blur.c-WARNING **: Error in gaussian-blur.c:519@cl_gaussian_blur - mem object allocation failure
Writing 'C:\Users\partha\AppData\Roaming\GIMP\2.9\filters\GimpGegl-gegl-gaussian-blur-config.settings' Parsing
'C:\Users\partha\AppData\Roaming\GIMP\2.9\filters\GimpGegl-gegl-c2g-config.settings'(script-fu.exe:6928): LibGimpBase-WARNING **: script-fu.exe: gimp_wire_read(): error
On Sat, Oct 12, 2013 at 12:21 AM, Partha Bagchi wrote:
I am finding that one possible show stopper is the gegl gaussian blur.
Here is a simple example. Open a 16bit, say 12 MP tiff image. Duplicate Layer, and run gegl gaussian blur with x=y=100. On my Macbook Pro with 8G RAM, either I will get vertical lines on the layer (at best) or crash my Mac.
Is there a way to speed up gaussian blur or is this a Mac issue and never happens on Linux?
Thanks, Partha