These extra outputs were caused by recent changes to
push_codec_srcs.
https://codereview.chromium.org/1327433003/
*** First, I would argue that we do not want to test
"native" modes (ex: codec, scanline, etc) to scales
that require sampling (ex: 0.1, 0.2, etc). Right now,
we are trying to scale jpegs to 0.1, settling for 0.125
as the closest option, and then trying to compare the
0.125 scaled image to the actual 0.1 scaled image in
Gold.
*** Second, I messed up and caused our test setup to
try to decode to kIndex8 and kGray8 "always" instead
of only when it is recommended. The bad effect of this
happens because we can decode jpegs to kGray8 even if
they are color images. Right now in Gold, we have a
bunch of untriaged gray versions of color images.
The second issue would have been caught if we signaled
a fatal failure for invalid conversions. Maybe we should
look into this now that 565 is supported everywhere?
BUG=skia:
Review URL: https://codereview.chromium.org/1314163007
*** Add CodecMode and ScaledCodecMode (in place of
NormalMode), so now we test SkCodec's getPixels() and
SkScaledCodec's getPixels()
*** Don't attempt to test scanline and codec modes using
the dimensions that were recommended for SkScaledCodec.
*** Change tags so that each scale gets its own output
folder.
TODO: Make ScanlineMode and ScanlineSubsetMode support
kOutOfOrder etc. I think this belongs with the gif CL -
I don't want to add test modes that we don't run yet.
BUG=skia:4202
BUG=skia:4238
Review URL: https://codereview.chromium.org/1327433003
Redesigns SkScanlineDecoder.h to indicate the ordering
in which the scanlines are provided
Refactors SkSwizzler::Fill() to include the zeroInit check
and to actually be correct.
BUG=skia:3257
BUG=skia:4198
Review URL: https://codereview.chromium.org/1287423002
Unfortunately, immintrin.h (which is also included by SkTypes)
includes xmmintrin.h which includes mm_malloc.h which includes
stdlib.h for malloc even though, from the implementation, it is
difficult to see why.
Fortunately, arm_neon.h does not seem to be involved in such
shenanigans, so building for Android will keep things sane.
TBR=reed@google.com
Doesn't change Skia API, just moves an include.
Review URL: https://codereview.chromium.org/1313203003
I suspect we might be doing some funky /n -> /r/n translations without 'b'.
This kills the PNG.
BUG=skia:
NOTREECHECKS=true
Review URL: https://codereview.chromium.org/1303063008
Reason for revert:
derek + Test-Win8-MSVC-ShuttleB-CPU-AVX2-x86_64-Debug
Original issue's description:
> Have DM manually encode its .png outputs.
>
> This eliminates some variability on various axes: different PNG encoders, different libpng versions, different formats (RGB, indexed), different unpremultiplication, different sRGB tags.
>
> BUG=skia:
>
> Committed: https://skia.googlesource.com/skia/+/3cc0dfffb70c0bd08ed8899efcd2e98da86a6ec7TBR=stephana@google.com,msarett@google.com
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=skia:
Review URL: https://codereview.chromium.org/1297383002
This eliminates some variability on various axes: different PNG encoders, different libpng versions, different formats (RGB, indexed), different unpremultiplication, different sRGB tags.
BUG=skia:
Review URL: https://codereview.chromium.org/1304443002
We don't want to test small images on Gold because they are
not interested to look at. Instead, I wrote a unit test to
verify that scaling small images does not cause crashes.
BUG=skia:
Review URL: https://codereview.chromium.org/1287863004
Add more swizzling functions for swizzling to 565. Much of this
code was revived from crrev.com/1055743003 (for BMP). Also added
swizzling functions for WBMP.
Consolidate the static function conversion_possible.
In SkCodec::getPixels, check that the alphatype corresponds to the
colorType. This prevents requesting 565 + non-opaque.
In SkIcoCodec, report that the image is unpremul (instead of
whatever the largest embedded codec thinks), but modify the
requested info to have the alpha type expected/required by the
embedded codec.
Add tests for decoding to 565.
BUG=skia:3257
BUG=skia:3683
Review URL: https://codereview.chromium.org/1277213002
libwebp has a fix for [1]. Update to the commit that contains the fix.
Update libwebp.gypi, corresponding to libwebp's latest makefile.
Turn back on DM testing for scaled webp, now that it should no longer
use uninitialized memory.
Fix a warning in config.h
[1] https://code.google.com/p/webp/issues/detail?id=254
BUG=skia:4038
Review URL: https://codereview.chromium.org/1280073002
Prior to this CL, if a client wanted to decode scanlines, they had to
create an SkCodec in order to get an SkScanlineDecoder. This introduces
complications if input data is not easily shared between the two
objects.
Instead, add methods to SkScanlineDecoder for creating a new one from
input data, and remove the creation functions from SkCodec.
Update DM and tests.
Review URL: https://codereview.chromium.org/1267583002
The decoding tests can now veto indirect sinks like pipe-8888.
This moves Sink type detection from automatic to explicit; I can't think of any
way to automatically differentiate pipe-8888 from 8888 based only on the
output. (They should ideally be identical, after all.)
BUG=skia:4138
Review URL: https://codereview.chromium.org/1263113002
In CodecSrc's scanline_subset test, we decode a subset of an image to a
bitmap, draw it to the canvas, and then repeat. This is fine for most
backends, but not for pipe. Pipe sees the same generation ID, so it
assumes it is the same bitmap it saw before, and just draws the
original one.
Call notifyPixelsChanged, so the bitmap will get a new generation ID,
fixing pipe.
BUG=skia:4138
Review URL: https://codereview.chromium.org/1265983004
This breaks Sinks down into three auto-detected types:
- GPU: anything that requests to be run in the GPU enclave
- Vector: anything that writes to the stream instead of the bitmap
- Raster: everything else
Some examples: gpu -> GPU, msaa16 -> GPU, 8888 -> raster, pdf -> vector,
svg -> vector, pipe-8888 -> raster, tiles_rt-gpu -> GPU
This lets image decoding sinks veto non-raster backends explicitly,
and can let particular GMs veto GPU or non-GPU sinks as they like.
BUG=skia:
Review URL: https://codereview.chromium.org/1239953004
This allows codecs that support subsets natively (i.e. WEBP) to do so.
Add a field on SkCodec::Options representing the subset.
Add a method on SkCodec to find a valid subset which approximately
matches a desired subset.
Implement subset decodes in SkWebpCodec.
Add a test in DM for decoding subsets.
Notice that we only start on even boundaries. This is due to the
way libwebp's API works. SkWEBPImageDecoder does not take this into
account, which results in visual artifacts.
FIXME: Subsets with scaling are not pixel identical, but close. (This
may be fine, though - they are not perceptually different. We'll just
need to mark another set of images in gold as valid, once
https://skbug.com/4038 is fixed, so we can tests scaled webp without
generating new images on each run.)
Review URL: https://codereview.chromium.org/1240143002
Scaling webp ends triggers warnings on our valgrind bot. It also results in
generating many images in Skia Gold that look mostly the same except
for a few pixels along the right edge.
BUG=skia:4038
Review URL: https://codereview.chromium.org/1227843005
Make getScanlineDecoder return a new object each time, which is
owned by the caller, and independent from any existing scanline
decoders and the SkCodec itself.
Since the SkCodec already contains the entire state machine, and it
is used by the scanline decoders, simply create a new SkCodec which
is now owned by the scanline decoder.
Move code that cleans up after using a scanline decoder into its
destructor
One side effect is that creating the first scanline decoder requires
a duplication of the stream and re-reading the header. (With some
more complexity/changes, we could pass the state machine to the
scanline decoder and make the SkCodec recreate its own state machine
instead.) The typical client of the scanline decoder (region decoder)
uses an SkMemoryStream, so the duplication is cheap, although we
should consider the extra time to reread the header/recreate the state
machine. (If/when we use the scanline decoder for other purposes,
where the stream may not be cheaply duplicated, we should consider
passing the state machine.)
One (intended) result of this change is that a client can create a
new scanline decoder in a new thread, and decode different pieces of
the image simultaneously.
In SkPngCodec::decodePalette, use fBitDepth rather than a parameter.
Review URL: https://codereview.chromium.org/1230033004
SkImageGenerator makes some assumptions that are not necessarily valid
for SkCodec. For example, SkCodec does not assume that it can always be
rewound.
We also have an ongoing question of what an SkCodec should report as
its default settings (i.e. the return from getInfo). It makes sense for
an SkCodec to report that its pixels are unpremultiplied, if that is
the case for the underlying data, but if a client of SkImageGenerator
uses the default settings (as many do), they will receive
unpremultiplied pixels which cannot (currently) be drawn with Skia. We
may ultimately decide to revisit SkCodec reporting an SkImageInfo, but
I have left it unchanged for now.
Import features of SkImageGenerator used by SkCodec into SkCodec.
I have left SkImageGenerator unchanged for now, but it no longer needs
Result or Options. This will require changes to Chromium.
Manually handle the lifetime of fScanlineDecoder, so SkScanlineDecoder.h
can include SkCodec.h (where Result is), and SkCodec.h does not need
to include it (to delete fScanlineDecoder).
In many places, make the following simple changes:
- Now include SkScanlineDecoder.h, which is no longer included by
SkCodec.h
- Use the enums in SkCodec, rather than SkImageGenerator
- Stop including SkImageGenerator.h where no longer needed
Review URL: https://codereview.chromium.org/1220733013
We name our .pngs by pixel hashes for gold. For 8888 images, we're hashing
SkPMColors, which have platform-dependent order: BGRA on Linux and Windows,
RGBA otherwise. This means we can end up with pixel-identical pngs with
different hashes, which is confusing.
This CL standardizes on RGBA for 8888 configs, arbitrarily chosen so that
Android ends up a no-op. Long-term, this should eliminate most of the
0-pixel-diff problems we see on gold.skia.org. There are other ways to end up
with the same .png from different SkBitmaps (think, red 565 square vs. red 8888
square) but they're rather less common / likely.
This will temporarily create a giant 0-pixel-diff problem on gold.skia.org.
Any Linux or Windows images which are not already pixel-identical to a Mac or
Android image should show up as untriaged hashes that are pixel-identical to
their version just before landing (we're only changing the hash, not the .png).
This means anything vaguely platform dependent (fonts, GPUs) will probably show
up as needing a triage but with a zero diff from a previous image.
If this goes well, we might do the same for 565. Just want to leave them out
for now to cut down on the triaging I need to do in one go.
BUG=skia:
Review URL: https://codereview.chromium.org/1226933005
