Re: [whatwg] Canvas element image scaling

2010-12-27 Thread Ian Hickson
On Sat, 18 Sep 2010, Rob Evans wrote:

 Has the canvas spec changed the way the drawImage function scales before 
 drawing to canvas?

On Sat, 18 Sep 2010, Tab Atkins Jr. wrote:
 
 It doesn't specify how the function scales at all: Note: This 
 specification does not define the algorithm to use when scaling the 
 image, if necessary..

Indeed.


On Sun, 19 Sep 2010, Rob Evans wrote:
 
 Thanks for the reply. I'm already using high resolution images and I 
 cannot cut sections of the image off because I am not using canvas to 
 create thumbnails, I'm using it to create a game engine. I guess this is 
 a question I will have to take up with the Firefox and Chrome developers 
 directly, unless there are any devs here that know why I'm seeing a 
 dramatic difference in scaling quality?

This sounds like a quality-of-implementation issue.


On Fri, 24 Sep 2010, Ruben Rodriguez II wrote:

 Forgive me as I have not kept up with all the discussions on this list, 
 but have there been any suggestions with regards to NOT using a filter 
 for rescaling? That would be ideal for 'old-skool' graphics where 
 pixellation is actually desired, not smoothing.

It's on the list of things to look at in future versions. In the meantime, 
browsers are experimenting in this area.

-- 
Ian Hickson   U+1047E)\._.,--,'``.fL
http://ln.hixie.ch/   U+263A/,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'


Re: [whatwg] Canvas element image scaling

2010-09-24 Thread Gregg Tavares (wrk)
As others have pointed out, canvas scaling algorithm is not specified and is
different in each browser.

http://greggman.com/downloads/examples/canvas-test/test-01/canvas-test-01-results.html

http://greggman.com/downloads/examples/canvas-test/test-01/canvas-test-01.html


On Sat, Sep 18, 2010 at 7:51 PM, Rob Evans r...@mtn-i.com wrote:

 Thanks I'll give that a go in the morning!

 All the best,

 Rob

 On 19 Sep 2010 03:42, Boris Zbarsky bzbar...@mit.edu wrote:

 On 9/18/10 9:57 PM, Rob Evans wrote:
 
  Thanks for the reply. I’m already using high resolution ima...

 Gecko will scale canvas images in one of two ways: either using a
 nearest-neighbor algorithm or using a more complicated (bilinear, bicubic,
 may depend on other details) algorithm which is slower but usually gives
 better results.  You can control which is happening by setting
 mozImageSmoothingEnabled on the canvas 2d context (set to false to get
 nearest-neighbor and set to true to get the other).

 The default value there is true.  Does setting it to false give you the
 Chrome 6 behavior, perchance?  I'd be a little surprised if it does, but
 worth trying.

 -Boris




Re: [whatwg] Canvas element image scaling

2010-09-24 Thread Boris Zbarsky

On 9/24/10 10:04 PM, Ruben Rodriguez II wrote:

Forgive me as I have not kept up with all the discussions on this list,
but have there been any suggestions with regards to NOT using a filter
for rescaling?


The mails you quoted mention how to do this in Gecko.

-Boris


Re: [whatwg] Canvas element image scaling

2010-09-18 Thread Tab Atkins Jr.
On Sat, Sep 18, 2010 at 2:25 PM, Rob Evans r...@mtn-i.com wrote:
 Has the canvas spec changed the way the drawImage function scales before
 drawing to canvas?

It doesn't specify how the function scales at all: Note: This
specification does not define the algorithm to use when scaling the
image, if necessary..

~TJ


Re: [whatwg] Canvas element image scaling

2010-09-18 Thread Yp C

It depends on the graphic lib of the browser. I think you can use a higher 
resolution picture which will not be shown, then you draw part of the higher 
resolution pic as the  thumbnail.

From: r...@mtn-i.com
To: whatwg@lists.whatwg.org
Date: Sat, 18 Sep 2010 22:25:12 +0100
Subject: [whatwg] Canvas element image scaling



Has the canvas spec changed the way the drawImage function scales before 
drawing to canvas? I’ve been testing using Chrome 6 and now with Chromium 7 and 
Firefox 4 Beta 6 the image scaling looks significantly different to the way 
Chrome 6 scales with drawImage. Just interested to know if I can get the later 
browser test versions to render with drawImage the same way as the current 
release of Chrome, if the new scaling I’m seeing is expected behaviour or just 
because they are test versions. If it is expected behaviour, it seems like a 
step backwards for image quality! You can see the images for yourself here: 
http://irrelon.com/test/ Thanks! Rob Evans  
   

Re: [whatwg] Canvas element image scaling

2010-09-18 Thread Rob Evans
Hey ya,

 

Thanks for the reply. I'm already using high resolution
images and I cannot cut sections of the image off because I am not using
canvas to create thumbnails, I'm using it to create a game engine. I guess
this is a question I will have to take up with the Firefox and Chrome
developers directly, unless there are any devs here that know why I'm seeing
a dramatic difference in scaling quality?

 

Again, you can see the images and their differing scaling
quality here: http://www.irrelon.com/test/

 

Thanks!

 

Rob

 

From: Yp C [mailto:tccy...@hotmail.com] 
Sent: 19 September 2010 02:51
To: r...@mtn-i.com
Cc: whatwg@lists.whatwg.org
Subject: RE: [whatwg] Canvas element image scaling

 

It depends on the graphic lib of the browser.

 I think you can use a higher resolution picture which will not be shown,
then you draw part of the higher resolution pic as the  thumbnail.

  _  

From: r...@mtn-i.com
To: whatwg@lists.whatwg.org
Date: Sat, 18 Sep 2010 22:25:12 +0100
Subject: [whatwg] Canvas element image scaling

Has the canvas spec changed the way the drawImage function scales before
drawing to canvas?

 

I've been testing using Chrome 6 and now with Chromium 7 and Firefox 4 Beta
6 the image scaling looks significantly different to the way Chrome 6 scales
with drawImage.

 

Just interested to know if I can get the later browser test versions to
render with drawImage the same way as the current release of Chrome, if the
new scaling I'm seeing is expected behaviour or just because they are test
versions. If it is expected behaviour, it seems like a step backwards for
image quality!

 

You can see the images for yourself here: http://irrelon.com/test/

 

Thanks!

 

Rob Evans



Re: [whatwg] Canvas element image scaling

2010-09-18 Thread Boris Zbarsky

On 9/18/10 9:57 PM, Rob Evans wrote:

Thanks for the reply. I’m already using high resolution images and I
cannot cut sections of the image off because I am not using canvas to
create thumbnails, I’m using it to create a game engine. I guess this is
a question I will have to take up with the Firefox and Chrome developers
directly, unless there are any devs here that know why I’m seeing a
dramatic difference in scaling quality?


Gecko will scale canvas images in one of two ways: either using a 
nearest-neighbor algorithm or using a more complicated (bilinear, 
bicubic, may depend on other details) algorithm which is slower but 
usually gives better results.  You can control which is happening by 
setting mozImageSmoothingEnabled on the canvas 2d context (set to false 
to get nearest-neighbor and set to true to get the other).


The default value there is true.  Does setting it to false give you the 
Chrome 6 behavior, perchance?  I'd be a little surprised if it does, but 
worth trying.


-Boris


Re: [whatwg] Canvas element image scaling

2010-09-18 Thread Rob Evans
Thanks I'll give that a go in the morning!

All the best,

Rob

On 19 Sep 2010 03:42, Boris Zbarsky bzbar...@mit.edu wrote:

On 9/18/10 9:57 PM, Rob Evans wrote:

 Thanks for the reply. I’m already using high resolution ima...
Gecko will scale canvas images in one of two ways: either using a
nearest-neighbor algorithm or using a more complicated (bilinear, bicubic,
may depend on other details) algorithm which is slower but usually gives
better results.  You can control which is happening by setting
mozImageSmoothingEnabled on the canvas 2d context (set to false to get
nearest-neighbor and set to true to get the other).

The default value there is true.  Does setting it to false give you the
Chrome 6 behavior, perchance?  I'd be a little surprised if it does, but
worth trying.

-Boris