@@ -444,7 +444,7 @@ The behavior of the PS and Agg backends was DPI dependent, thus::
444444 mpl.rcParams['hatch.linewidth'] = 1.0 / dpi # previous ps and Agg hatch linewidth
445445
446446
447- There is no API level control of the hacth linewidth.
447+ There is no API level control of the hatch linewidth.
448448
449449
450450.. _default_changes_font :
@@ -460,9 +460,9 @@ Sans". "DejaVu Sans" is an improvement on "Bistream Vera Sans" that
460460adds more international and math characters, but otherwise has the
461461same appearance. Latin, Greek, Cyrillic, Armenian, Georgian, Hebrew,
462462and Arabic are `all supported
463- <http://dejavu-fonts.org/wiki/Main_Page> `__ (but right-to-left render
463+ <http://dejavu-fonts.org/wiki/Main_Page> `__ (but right-to-left rendering
464464is still not handled by matplotlib). In addition , DejaVu contains a
465- sub-set of emoji symbols
465+ sub-set of emoji symbols.
466466
467467.. plot ::
468468
@@ -506,7 +506,7 @@ TeX backend is used (i.e. ``text.usetex`` is ``True``).
506506
507507 fig, ax = plt.subplots(tight_layout=True, figsize=(3, 3))
508508
509- ax.plot(range(15), label='int: $15 \i nt_0^\i nfty dx$')
509+ ax.plot(range(15), label=r 'int: $15 \i nt_0^\i nfty dx$')
510510 ax.legend()
511511 ax.set_title('classic')
512512
@@ -518,7 +518,7 @@ TeX backend is used (i.e. ``text.usetex`` is ``True``).
518518
519519 fig, ax = plt.subplots(tight_layout=True, figsize=(3, 3))
520520
521- ax.plot(range(15), label='int: $15 \i nt_0^\i nfty dx$')
521+ ax.plot(range(15), label=r 'int: $15 \i nt_0^\i nfty dx$')
522522 ax.legend()
523523 ax.set_title('v2.0')
524524
@@ -531,7 +531,7 @@ To revert to the old behavior set the::
531531
532532or by setting::
533533
534- mathetxt .fontset: cm
534+ mathtext .fontset: cm
535535 mathtext.rm : serif
536536
537537
@@ -556,7 +556,6 @@ Legends
556556.. plot ::
557557
558558 import matplotlib as mpl
559- import numpy as np
560559 import matplotlib.pyplot as plt
561560 import numpy as np
562561
@@ -567,8 +566,8 @@ Legends
567566 x = range(N)
568567 y = np.cumsum(np.random.randn(N) )
569568 ln, = ax.plot(x, y, marker='s',
570- linestyle='-', label='plot')
571- ax.fill_between(x, y, 0, label='fill', alpha=.5, color=ln.get_color())
569+ linestyle='-', label='plot')
570+ ax.fill_between(x, y, 0, label='fill', alpha=.5, color=ln.get_color())
572571 ax.scatter(N*np.random.rand(N), np.random.rand(N), label='scatter')
573572 ax.set_title(title)
574573 ax.legend()
@@ -605,9 +604,9 @@ or by setting::
605604 legend.loc : upper right
606605 legend.numpoints : 2 # the number of points in the legend line
607606 legend.fontsize : large
608- legend.framealpha : None # opacity of of legend frame
607+ legend.framealpha : None # opacity of legend frame
609608 legend.scatterpoints : 3 # number of scatter points
610- legend.edgecolor : inherit # legend edge color (when None inherits from axes.facecolor )
609+ legend.edgecolor : inherit # legend edge color (when 'inherit' uses axes.edgecolor )
611610
612611in your :file: `matplotlibrc ` file.
613612
@@ -617,9 +616,9 @@ Image
617616Interpolation
618617-------------
619618
620- The default interpolation method for `~matplotlib.axes.Axes.imshow ` is now
621- ``'nearest' `` and by default resamples the data to both up and down sample
622- then input before color mapping.
619+ The default interpolation method for `~matplotlib.axes.Axes.imshow ` is
620+ now ``'nearest' `` and by default resamples the data ( both up and down
621+ sampling) before color mapping.
623622
624623
625624.. plot ::
@@ -645,21 +644,6 @@ then input before color mapping.
645644 demo(ax1, classic_rcparams, 'classic')
646645 demo(ax2, {}, 'v2.0')
647646
648- The default value for the rcParam ``image.resample `` is now ``True ``.
649- This will apply interpolation for both upsampling and downsampling
650- of an image. ``image.interpolation ``, is now ``nearest ``.
651-
652- Previously, the input data was normalized, then color mapped, and the
653- resampled to the resolution required for the screen. This means that
654- the final resampling was being done in color space. Because the color
655- maps are not always linear in RGB space this can result in colors not
656- in the color map appearing in the final image. This bug was addressed
657- by an almost complete overhaul of how the image handling code works.
658- The input data is now normalized, then resampled to the correct
659- resolution (in scaled dataspace), and then finally color mapped to
660- RGBA space which prevents any colors not in the color map from
661- appearing in the final image (if your viewer subsequently resamples
662- the image the artifact may reappear).
663647
664648To restore the previous behavior set::
665649
@@ -673,6 +657,24 @@ or set::
673657
674658in your :file: `matplotlibrc ` file.
675659
660+ Colormapping pipeline
661+ ---------------------
662+
663+ Previously, the input data was normalized, then color mapped, and then
664+ resampled to the resolution required for the screen. This meant that
665+ the final resampling was being done in color space. Because the color
666+ maps are not generally linear in RGB space colors not in the color map
667+ may appear in the final image. This bug was addressed by an almost
668+ complete overhaul of how the image handling code works.
669+
670+ The input data is now normalized, then resampled to the correct
671+ resolution (in normalized dataspace), and then finally color mapped to
672+ RGB space. This ensures only colors actually in the color map appear
673+ in the final image (if your viewer subsequently resamples the image
674+ the artifact may reappear).
675+
676+ The previous behavior can not be restored.
677+
676678
677679Shading
678680-------
0 commit comments