Bug 71962 – Simplify the way that images are passed into a filter chain

• Status: RESOLVED
• Resolution: INVALID
• Component: SVG

[Simon Fraser (smfr)]

Looking at the patch in bug 68472, I think the way that data gets into and out of filter chains could be improved. The code currently has to do:

m_filter->setSourceImageRect(size);
m_filter->setSourceImage(ImageBuffer::create(layoutSize));
GraphicsContext* sourceGraphicsContext = m_filter->sourceImage()->context();
// paint stuff

m_filter->effect()->apply();
p->drawImageBuffer(m_filter->effect()->asImageBuffer(), ...)


It seems like it would be much clearer as:

1. Make an ImageBuffer
2. Paint into the ImageBuffer
3. Hand the ImageBuffer to the filter.
4. Get the filtered result as an image buffer
5. Paint that image buffer.

[Radar WebKit Bug Importer]

<rdar://problem/10435248>

[Dirk Schulze]

(In reply to comment #0)
> Looking at the patch in bug 68472, I think the way that data gets into and out of filter chains could be improved. The code currently has to do:
> 
> m_filter->setSourceImageRect(size);
> m_filter->setSourceImage(ImageBuffer::create(layoutSize));
> GraphicsContext* sourceGraphicsContext = m_filter->sourceImage()->context();
> // paint stuff
> 
> m_filter->effect()->apply();
> p->drawImageBuffer(m_filter->effect()->asImageBuffer(), ...)
> 
> 
> It seems like it would be much clearer as:
> 
> 1. Make an ImageBuffer
> 2. Paint into the ImageBuffer
> 3. Hand the ImageBuffer to the filter.
> 4. Get the filtered result as an image buffer
> 5. Paint that image buffer.

Well, that seems what we are doing right now. What is your suggestion?

[Simon Fraser (smfr)]

True dat.