Re: Issue 7508 in chromium: We don't anti-alias <canvas> drawImage
chro...@googlecode.com
Comment #8 on issue 7508 by mri...@google.com: We don't anti-alias <canvas>
drawImage
http://code.google.com/p/chromium/issues/detail?id=7508
Please don't fix this unless you provide a way to maintain the current non
anti-aliased behavior.
--
You received this message because you are listed in the owner
or CC fields of this issue, or because you starred this issue.
You may adjust your issue notification preferences at:
http://code.google.com/hosting/settings
chro...@googlecode.com
Comment #9 on issue 7508 by cagret: We don't anti-alias <canvas> drawImage
http://code.google.com/p/chromium/issues/detail?id=7508
Rounded corners for images look like crap in Chrome, they are "pixelated" a
lot. See
the attachments.
rounded-chrome.jpg - how it looks in Chrome because there is no
anti-aliasing
rounded-other-browsers - this is how it looks in: Firefox, Opera, Internet
Explorer
(using vml)
Attachments:
rounded-chrome.jpg 8.8 KB
rounded-other-browsers.jpg 8.7 KB
--
Automated mail from issue updates at http://crbug.com/
Subscription options: http://groups.google.com/a/chromium.org/group/chromium-bugs
chro...@googlecode.com
Comment #11 on issue 7508 by ricardo.cabello: We don't anti-alias <canvas>
drawImage
http://code.google.com/p/chromium/issues/detail?id=7508
This sample doesn't look bad in Safari:
http://mrdoob.com/projects/three.js/examples/materials_video.html
I agree with Dean, Chrome shouldn't be different from the rest of the
browsers and should antialias clip().
Too bad that mr. Hickson was against having a antialias boolean. Using
<svg> you can specify whether a shape is renderer with antialias or
without, a shame you can't do the same with <canvas>.
chro...@googlecode.com
Comment #15 on issue 7508 by downchuck: We don't anti-alias <canvas>
drawImage
http://code.google.com/p/chromium/issues/detail?id=7508
IanODea: Have you considered using the CSS transform attributes? I'm not
100% sure, but I believe that they will show an anti-aliased canvas. It's
not a solution, but it's a work around for some use cases. The WHATWG has
stated without any doubt, that they will not be altering the canvas spec in
relation to anti-aliasing any time soon. Their general attitude seems to be
that SVG should be used in cases where Canvas can't be.
I did try to push this on the WHATWG quite a long time ago, I understand
the issues it brings, and they have been outlined here in this thread.
Using WebGL may be another possible work-around (haven't tried it).
A special HTML / CSS attribute could be added as a means to address this
issue, and keep current functionality.
Would either of these be acceptable to Chromium developers?:
[canvas x-chromium-antialias]
[canvas style="-chromium-antialias: true"]
chro...@googlecode.com
Comment #17 on issue 7508 by megazzt: We don't anti-alias <canvas> drawImage
http://code.google.com/p/chromium/issues/detail?id=7508
Just discovered this while trying to figure out why my canvas-drawn browser
action was much more jaggy than the default image I assigned it. It would
definitely be nice to be able to anti-alias.
chro...@googlecode.com
Comment #18 on issue 7508 by downchuck: We don't anti-alias <canvas>
drawImage
http://code.google.com/p/chromium/issues/detail?id=7508
I've a proposal, for chrome behavior. When a shadow is set, drawImage
should be anti-aliased: This would allow it to keep its current non-aliased
behavior, when a shadow is not set, but would use resampled / aliased
behavior when the shadow is set.
Currently, shadow behavior on drawImage is not defined/implemented, as
stated in issue #11153
chro...@googlecode.com
Comment #19 on issue 7508 by jaffathecake: We don't anti-alias <canvas>
drawImage
http://code.google.com/p/chromium/issues/detail?id=7508
Interesting way of having both behaviours, but it it sounds a bit hacky. As
a user I wouldn't expect shadow to change the anti-aliasing of drawImage.
Personally, I don't think there needs to be an option for both, but if we
need one it should be an option on the context rather than the canvas, else
we'll end up with settings that might not make sense (or won't be
implemented) for future contexts.
Eg:
context.antialias = 'nearestneighbour';
// drawing won't be antialiased...
context.antialias = 'normal';
// now it will be (normal should be the default)
This would allow for other kinds of anti-aliasing to be applied, eg
advanced anti-aliasing for text which looks great for static text, but when
animated looks weird as parts of the glyph try to snap to pixels.
(http://help.adobe.com/en_US/FlashPlatform/reference/actionscript/3/flash/text/AntiAliasType.html)
chro...@googlecode.com
Comment #21 on issue 7508 by jdrinkwater: We don't anti-alias <canvas>
drawImage
http://code.google.com/p/chromium/issues/detail?id=7508
Something like Gecko’s CanvasContext.mozImageSmoothingEnabled would be nice
to have.
I’m working on a project where I *don’t* want the drawImage() call to
smooth/anti‐alias the resulting scaled image.
chro...@googlecode.com
Comment #23 on issue 7508 by av...@shadypixel.com: We don't anti-alias
<canvas> drawImage
http://code.google.com/p/chromium/issues/detail?id=7508
Is ever going to be fixed in chrome? I'm basically maintaining two
different drawing code paths because everyone else antialiases drawImage
and chrome doesn't (and requires significant work to even get it looking
half way decent).
One thing to note is that if images are antialiases by default and you
don't want antialiasing you can always floor/round/ceil your non integer
coordinates prior to calling drawImage. The reverse is not possible or at
the very least not possible to do efficiently.
chro...@googlecode.com
Comment #24 on issue 7508 by antipl...@gmail.com: We don't anti-alias
<canvas> drawImage
http://code.google.com/p/chromium/issues/detail?id=7508
"I'm basically maintaining two different drawing code paths because
everyone else antialiases drawImage and chrome doesn't (and requires
significant work to even get it looking half way decent)."
how can you improve quality of image? i'm storing rotated and clipped
images in separate canvas in memory, and then use drawImage() method.
result is very poor in chrome.
chro...@googlecode.com
Comment #26 on issue 7508 by chris.pi...@gmail.com: We don't anti-alias
<canvas> drawImage
http://code.google.com/p/chromium/issues/detail?id=7508
It would be really great to get Chrome in line with the other browsers in
this.
chro...@googlecode.com
Comment #27 on issue 7508 by aleksand...@gmail.com: We don't anti-alias
<canvas> drawImage
http://code.google.com/p/chromium/issues/detail?id=7508
It would be awesome
chro...@googlecode.com
Comment #29 on issue 7508 by wiltz...@chromium.org: We don't anti-alias
<canvas> drawImage
http://code.google.com/p/chromium/issues/detail?id=7508
I recall some other discussion of anti-aliased images inside of canvas on
another bug but now can't find it. bsalomon@ might be able to provide more
color on what the current state of affairs is and whether we're considering
changing anything in the near future.
chro...@googlecode.com
Comment #36 on issue 7508 by wiltz...@chromium.org: We don't anti-alias
<canvas> drawImage
http://code.google.com/p/chromium/issues/detail?id=7508
Hey Brian, has this bubbled up at all? I realize its probably behind a
bunch of performance improvements, just curious if anything's changed.
chro...@googlecode.com
Cc: robertph...@google.com
Comment #37 on issue 7508 by bsalo...@google.com: We don't anti-alias
<canvas> drawImage
http://code.google.com/p/chromium/issues/detail?id=7508
Rob Phillips is starting to make some improvements in how we handle mask
textures. I think a next step for him may be to add support for
soft-clipping using mask textures.
I just want to note that image filtering (or smoothing) is really about how
how images are scaled. The difference we're seeing here between chrome and
FF or IE is how clips are applied. Today we aren't antialiasing the edges
of the clip.
chro...@googlecode.com
Comment #38 on issue 7508 by wiltz...@chromium.org: We don't anti-alias
<canvas> drawImage
http://code.google.com/p/chromium/issues/detail?id=7508
@bsalomon, will Rob be doing any Chrome plumbing of those eventual changes?
Just wondering if I should assign him this bug and how it's going. Thanks,
chro...@googlecode.com
Comment #39 on issue 7508 by tza...@gmail.com: We don't anti-alias <canvas>
drawImage
http://code.google.com/p/chromium/issues/detail?id=7508
Hey guys.
It is very important issue for my application, and its a bit disappointing
that other browsers do antialias images in canvas, but my favorite doesn't.
I understand that skiping antialias in images on canvas gives performance
boost, but having a switch for enabling/disabling antialias, would allow
the app to chose what kind of compromise to do would be great.
From what I noticed, currently not only rotated images are
having 'staircase' edges, but also scaled down images look very badly.
We are currently in beta, with lots of bugs, but we are having kind of PPT
clone done with canvas. You can try dropping some images from websearch on
the page, and then use handles to rotate and resize.
http://www.remby.com/try/
The absence of antialias makes these resized and rotated images loose a lot
of their initial quality
chro...@googlecode.com
Comment #40 on issue 7508 by bsalo...@google.com: We don't anti-alias
<canvas> drawImage
http://code.google.com/p/chromium/issues/detail?id=7508
wiltzius@, when we come out next week we want to talk to people there about
how best to make it possible to have smooth/aa clipping in canvas2d without
regressing performance in cases where it isn't necessary.
chro...@googlecode.com
Comment #41 on issue 7508 by bernhard...@gmail.com: We don't anti-alias
<canvas> drawImage
http://code.google.com/p/chromium/issues/detail?id=7508
regarding performance - IE9 does perfect anti-aliasing on transformed
images and the performance is as good as it is in Chrome, even better ...
chro...@googlecode.com
Comment #42 on issue 7508 by downch...@gmail.com: We don't anti-alias
<canvas> drawImage
http://code.google.com/p/chromium/issues/detail?id=7508
An image smoothing flag seems to have consensus among vendors. The idea is
to pass imageSmoothing = false to scale up pixels without blurring them. I
suggest the default behavior of Chrome be bound to that behavior. So,
people who need those jagged edges for rendering techniques will get them
when they disable image smoothing. The rest of us will get nicely
transformed images.
chro...@googlecode.com
Cc: r...@chromium.org tomhud...@chromium.org
Comment #43 on issue 7508 by bsalo...@google.com: We don't anti-alias
<canvas> drawImage
http://code.google.com/p/chromium/issues/detail?id=7508
Issue 124756 has been merged into this issue.
chro...@googlecode.com
Comment #44 on issue 7508 by wiltz...@chromium.org: We don't anti-alias
<canvas> drawImage
http://code.google.com/p/chromium/issues/detail?id=7508
Brian did you get a chance to talk to folks out here about this? Sorry I
totally neglected to bring it up.
chro...@googlecode.com
Comment #45 on issue 7508 by sa...@chixor.net: We don't anti-alias <canvas>
drawImage
http://code.google.com/p/chromium/issues/detail?id=7508
Looking forward to anti-aliasing drawImage scaling in chrome. Currently my
html5 canvas app looks the worst in chrome over any other browser.
I'm surprised this has been on your radar for 3yrs and there's been no
progress.
chro...@googlecode.com
Comment #47 on issue 7508 by iva...@gmail.com: We don't anti-alias <canvas>
drawImage
http://code.google.com/p/chromium/issues/detail?id=7508
This is really strange; Google seems to be pushing new technologies in
every other release of Chrome like the File API and the Audio PUT but is
not interested in fixing this issue nor the issue with the font rendering
of (fancy-looking) fonts.
In this age where beautiful apps seems to be so strongly desired by the
general public and therefore a key-point in many products, this issue
(which was created 3 years ago) is unacceptable.
chro...@googlecode.com
Comment #48 on issue 7508 by ogle....@gmail.com: We don't anti-alias
<canvas> drawImage
http://code.google.com/p/chromium/issues/detail?id=7508
Judging by the comment by robert in
http://code.google.com/p/chromium/issues/detail?id=132442 , it seems maybe
someone is paying attention to it.
Is there some kind of estimation for this? Even realistic ballpark: 1 week?
1 year? never? chrome 22?
The clipping mask issue totally kills my canvas app. I'm considering
rendering everything at 2x and scaling down with css as a workaround. But I
really really really dont want to.
chro...@googlecode.com
Comment #49 on issue 7508 by robertph...@google.com: We don't anti-alias
<canvas> drawImage
http://code.google.com/p/chromium/issues/detail?id=7508
We have anti-aliased clipping already working inside Skia (both HW & SW)
but a lot of work remains to make it visible from canvas in a way that (if
possible) works when perfect seaming is required (e.g. the tulrich.com link
above). Additionally there is still a lot of optimization needed before
we're happy with it. I can't give an exact timeline but certainly more than
1 week and way less than 1 year.
chro...@googlecode.com
Comment #51 on issue 7508 by hypothet...@gmail.com: We don't anti-alias
<canvas> drawImage
http://code.google.com/p/chromium/issues/detail?id=7508
Just chiming in for a flag instead of a default antialias. Not having this
available would severely limit many designers and artists online who prefer
jaggy edges, such as the pixel art scene and people working with small
fonts. Thanks!
chro...@googlecode.com
Comment #52 on issue 7508 by opusc...@gmail.com: We don't anti-alias
<canvas> drawImage
http://code.google.com/p/chromium/issues/detail?id=7508
the anti-aliasing flag exists as "imageSmoothingEnabled" on a 2d canvas
context.
see
http://www.whatwg.org/specs/web-apps/current-work/multipage/the-canvas-element.html#dfnReturnLink-4
it isn't implemented yet in chrome, but works and is named
mozImageSmoothingEnabled in firefox.
chro...@googlecode.com
Comment #53 on issue 7508 by get...@gmail.com: We don't anti-alias <canvas>
drawImage
http://code.google.com/p/chromium/issues/detail?id=7508
The spec doesn't indicate if that flag is also supposed to control
anti-aliasing when you do a clearRect() call. Will Chrome's implementation
of this flag affect clearRect()?
Is this a short-coming of the spec that it doesn't mention how/if
anti-aliasing should happen with clears?
chro...@googlecode.com
Cc: ke...@chromium.org
Comment #54 on issue 7508 by bsalo...@google.com: We don't anti-alias
<canvas> drawImage
http://code.google.com/p/chromium/issues/detail?id=7508
Support for imageSmoothingEnabled is implemented in Chrome (though not the
shipping version). However, it does not affect the clip edges, only the
filtering of source bitmap pixels.
chro...@googlecode.com
Comment #55 on issue 7508 by get...@gmail.com: We don't anti-alias <canvas>
drawImage
http://code.google.com/p/chromium/issues/detail?id=7508
(in reply to comment #54):
This comment http://code.google.com/p/chromium/issues/detail?id=132442#c4
indicates that that bug (anti-aliasing on clipping) is related to this one,
in that this bug should fix that one.
Is it planned that `imageSmoothingEnabled` would eventually control the
anti-aliasing on clip edges, or is it planned that anti-aliasing of clip
will always be on and won't be controllable?
chro...@googlecode.com
Comment #56 on issue 7508 by bsalo...@google.com: We don't anti-alias
<canvas> drawImage
http://code.google.com/p/chromium/issues/detail?id=7508
We are not planning to tie anti-aliased clipping to imageSmoothingEnabled.
We think these are orthogonal. Anti-aliased alipping will either always be
enabled or we will have a different way to toggle it.
chro...@googlecode.com
Status: Assigned
Owner: robertph...@chromium.org
Comment #57 on issue 7508 by senorbla...@chromium.org: We don't anti-alias
<canvas> drawImage
http://code.google.com/p/chromium/issues/detail?id=7508
Robert, could you take this one on? If you don't plan to be working on it
soon, feel free to un-own it.
chro...@googlecode.com
Comment #61 on issue 7508 by kevco...@gmail.com: We don't anti-alias
<canvas> drawImage
http://code.google.com/p/chromium/issues/detail?id=7508
At InMobi we'd rather use canvas clip() in our rich media ad-building
products, but because of this bug output looks horrible on Android devices.
One workaround we use is compositing with source-atop but it isn't ideal.
Please fix this so we don't have to maintain separate code paths for iOS
and Android devices!
--
You received this message because this project is configured to send all
issue notifications to this address.
You may adjust your notification preferences at:
https://code.google.com/hosting/settings
chro...@googlecode.com
Comment #62 on issue 7508 by m...@wilmslowastro.com: We don't anti-alias
<canvas> drawImage
http://code.google.com/p/chromium/issues/detail?id=7508
I think this is the issue I am seeing with my canvas based gauges on Chrome
Android 30? They render fine on the other Android browsers (Built-in, FF,
Dolphin) and Chrome desktop, and I'm pretty sure they rendered fine on
older versions of Chrome for Android too. I draw to multiple off screen
canvas's and then drawImage() the layers into the on screen canvas.
http://www.wilmslowastro.com/steelseries/demoRadial.html
chro...@googlecode.com
Comment #63 on issue 7508 by senorbla...@chromium.org: We don't anti-alias
<canvas> drawImage
http://code.google.com/p/chromium/issues/detail?id=7508
ma...@wilmslowastro.com: what you're seeing there was an intentional
Android-specific change in v30; see
https://code.google.com/p/chromium/issues/detail?id=285066 for details.
chro...@googlecode.com
Comment #64 on issue 7508 by myquas...@gmail.com: We don't anti-alias
<canvas> drawImage
http://code.google.com/p/chromium/issues/detail?id=7508
Will this ever get fixed? It is really a shame that the clip() function in
Chrome turns out useless due the jaggy edges, even in the latest version.
What does canvas help without proper clipping?
chro...@googlecode.com
Comment #65 on issue 7508 by ryanack...@gmail.com: We don't anti-alias
<canvas> drawImage
http://code.google.com/p/chromium/issues/detail?id=7508
I think the clipping issue is a different bug. Do a bug search or Google
search to find it.
chro...@googlecode.com
Comment #67 on issue 7508 by simon.do...@gmail.com: We don't anti-alias
<canvas> drawImage
http://code.google.com/p/chromium/issues/detail?id=7508
The clipping mask issue is a big problem.. Will this ever get fixed?
chro...@googlecode.com
Comment #71 on issue 7508 by tra...@snapsapp.com: We don't anti-alias
<canvas> drawImage
http://code.google.com/p/chromium/issues/detail?id=7508
Any update from someone on the Chromium team would be awesome. It's
really, really sad that this still exists, and it shouldn't even be that
difficult of a bug in the first place compared with other new feature work.
chro...@googlecode.com
Comment #72 on issue 7508 by michael....@gmail.com: We don't anti-alias
<canvas> drawImage
http://code.google.com/p/chromium/issues/detail?id=7508
This bug causes rendering artifacts with PDF.js while rendering PDFs. See
below for more details:
https://github.com/mozilla/pdf.js/issues/5196
chro...@googlecode.com
Comment #73 on issue 7508 by kottena...@gmail.com: We don't anti-alias
<canvas> drawImage
https://code.google.com/p/chromium/issues/detail?id=7508
Chromium, why?
Clipping is very common operation. How could you still don't fix it?
Just compare next page in Chromium and Firefox:
http://kottenator.github.io/jquery-circle-progress/ - there is only a
simple clipping and it's awful in Chromium 36
chro...@googlecode.com
Comment #74 on issue 7508 by mich...@mknwebsolutions.com: We don't
Awaiting update on this too. I noticed Firefox and IE render smoothly. I
think I also noticed that Windows Chrome renders the image smoothly, but
Mac Chrome does not.
chro...@googlecode.com
Cc: ju...@chromium.org
Comment #75 on issue 7508 by robertph...@chromium.org: We don't anti-alias
<canvas> drawImage
https://code.google.com/p/chromium/issues/detail?id=7508
(No comment was entered for this change.)
chro...@googlecode.com
Comment #80 on issue 7508 by tra...@snapsapp.com: We don't anti-alias
<canvas> drawImage
https://code.google.com/p/chromium/issues/detail?id=7508
Hey Robert -- really appreciate the update!
Just to be clear, antialiasing clipping is a different topic than the
original focus of this issue, which is antialiasing drawImage calls,
right? The latter seems to me a much more common and important issue.
chro...@googlecode.com
Comment #82 on issue 7508 by tra...@snapsapp.com: We don't anti-alias
<canvas> drawImage
https://code.google.com/p/chromium/issues/detail?id=7508
Awesome -- thanks for the clarification, Robert!
chro...@googlecode.com
Comment #83 on issue 7508 by Jarda.Ko...@gmail.com: We don't anti-alias
<canvas> drawImage
https://code.google.com/p/chromium/issues/detail?id=7508
Interestingly antialiasing for drawImage seems to be working ok with Chrome
41. I have noticed difference on OS X and Win8.1when updating from Chrome
40 to 41. So is this still the issue?
chro...@googlecode.com
Owner: ju...@chromium.org
Comment #84 on issue 7508 by ju...@chromium.org: We don't anti-alias
<canvas> drawImage
https://code.google.com/p/chromium/issues/detail?id=7508
(No comment was entered for this change.)
chro...@googlecode.com
Comment #87 on issue 7508 by tobiaskl...@gmail.com: We don't anti-alias
<canvas> drawImage
https://code.google.com/p/chromium/issues/detail?id=7508
I'm on chrome Version 47.0.2526.106 (64-bit) and I am trying to slowly move
an image in canvas. This still looks horrible in chrome but good in other
browsers.