Attachment 4634 Details for Bug 3387 – patch that should fix the problem

[patch] patch that should fix the problem

imask.patch (text/plain), 2.02 KB, created by Jeff Muizelaar on 2006-02-17 04:29:18 UTC

(hide)

Description:

Filename:

MIME Type:

Creator: Jeff Muizelaar

Created: 2006-02-17 04:29:18 UTC

Size: 2.02 KB

patch

obsolete

>Index: poppler/CairoOutputDev.cc
>===================================================================
>RCS file: /cvs/poppler/poppler/poppler/CairoOutputDev.cc,v
>retrieving revision 1.27
>diff -u -r1.27 CairoOutputDev.cc
>--- poppler/CairoOutputDev.cc	26 Jan 2006 19:25:07 -0000	1.27
>+++ poppler/CairoOutputDev.cc	16 Feb 2006 17:19:04 -0000
>@@ -481,6 +481,26 @@
>   int invert_bit;
>   int row_stride;
> 
>+  ctm = state->getCTM();
>+  LOG (printf ("drawImageMask %dx%d, matrix: %f, %f, %f, %f, %f, %f\n",
>+	       width, height, ctm[0], ctm[1], ctm[2], ctm[3], ctm[4], ctm[5]));
>+  matrix.xx = ctm[0] / width;
>+  matrix.xy = -ctm[2] / height;
>+  matrix.yx = ctm[1] / width;
>+  matrix.yy = -ctm[3] / height;
>+  matrix.x0 = ctm[2] + ctm[4];
>+  matrix.y0 = ctm[3] + ctm[5];
>+
>+  /* work around a cairo bug when scaling 1x1 surfaces */
>+  if (width == 1 && height == 1) {
>+    cairo_save (cairo);
>+    cairo_set_matrix (cairo, &matrix);
>+    cairo_rectangle (cairo, 0., 0., 1., 1.);
>+    cairo_fill (cairo);
>+    cairo_restore (cairo);
>+    return;
>+  }
>+
>   row_stride = (width + 3) & ~3;
>   buffer = (unsigned char *) malloc (height * row_stride);
>   if (buffer == NULL) {
>@@ -514,19 +534,14 @@
>   if (pattern == NULL)
>     return;
> 
>-  ctm = state->getCTM();
>-  LOG (printf ("drawImageMask %dx%d, matrix: %f, %f, %f, %f, %f, %f\n",
>-	       width, height, ctm[0], ctm[1], ctm[2], ctm[3], ctm[4], ctm[5]));
>-  matrix.xx = ctm[0] / width;
>-  matrix.xy = -ctm[2] / height;
>-  matrix.yx = ctm[1] / width;
>-  matrix.yy = -ctm[3] / height;
>-  matrix.x0 = ctm[2] + ctm[4];
>-  matrix.y0 = ctm[3] + ctm[5];
>   cairo_matrix_invert (&matrix);
>   cairo_pattern_set_matrix (pattern, &matrix);
> 
>+  /* we should actually be using CAIRO_FILTER_NEAREST here. However,
>+   * cairo doesn't yet do minifaction filtering causing scaled down
>+   * images with CAIRO_FILTER_NEAREST to look really bad */
>   cairo_pattern_set_filter (pattern, CAIRO_FILTER_BEST);
>+
>   /* FIXME: Doesn't the image mask support any colorspace? */
>   cairo_set_source (cairo, fill_pattern);
>   cairo_mask (cairo, pattern);

Actions: View | Splinter Review

Attachments on bug 3387: 2761 | 4061 | 4634

Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.