Mozilla2:Image Encoding: Difference between revisions

From MozillaWiki
Jump to navigation Jump to search
mNo edit summary
 
(12 intermediate revisions by 3 users not shown)
Line 1: Line 1:
== Interface ==
== Interface ==
We need to add a new interface. Do we want to support passing in an output stream? When you read from the stream should we encode just that much?  Thoughts?
We need to add a new interface.  When you read from the stream should we encode just that much?  Thoughts?


Something like:
Something like:


  interface imgIEncoder : nsISupports
  interface imgIEncoder : nsIInputStream
  {
  {
   nsIOutputStream encode(imgIContainer image);
   /* input image data
    * result width
  /* Output width/height */
    * result height
   attribute unsigned long width;
    */
  attribute unsigned long height;
   void init(in imgIContainer input,
            in unsigned long width, in unsigned long height);
  /* scaling */
  ??
  };
  };


=== Scaling ===
Do we want to support different types of scaling?  I'd prefer to just do nice Bicubic resampling. [http://members.bellatlantic.net/~vze2vrva/design.html] [http://astronomy.swin.edu.au/~pbourke/colour/bicubic/]
=== Brendan's comments ===
A reader should get the bytes requested, if not greater than length of encoded image, and it's up to the encoder to buffer more if it encodes more bytes, reading from the buffer on next read.
This expresses the idea that the encoder is something you read from.
== Use ==
You would create one by doing:
You would create one by doing:
  createInstance("@mozilla.org/image/encoder;2?type=image/png");
  createInstance("@mozilla.org/image/encoder;2?type=image/png");

Latest revision as of 06:17, 8 March 2005

Interface

We need to add a new interface. When you read from the stream should we encode just that much? Thoughts?

Something like:

interface imgIEncoder : nsIInputStream
{
  /* input image data
   * result width
   * result height
   */
  void init(in imgIContainer input,
            in unsigned long width, in unsigned long height);
};

Scaling

Do we want to support different types of scaling? I'd prefer to just do nice Bicubic resampling. [1] [2]

Brendan's comments

A reader should get the bytes requested, if not greater than length of encoded image, and it's up to the encoder to buffer more if it encodes more bytes, reading from the buffer on next read.


This expresses the idea that the encoder is something you read from.

Use

You would create one by doing:

createInstance("@mozilla.org/image/encoder;2?type=image/png");