4 ms·

Regarding hardware-accelerated drawing of text: I thought that IE just like any other Windows program is using Windows core APIs like GDI+ to display text. And

Regarding hardware-accelerated drawing of text: I thought that IE just like any other Windows program is using Windows core APIs like GDI+ to display text. And those APIs, in turn, are hardware-accelerated as much as possible. These days the entire desktop is essentially a "game" running on top of DirectX, no? Even back in 1995 most of GDI 2D functions were hardware-accelerated, assuming you had a good card with good drivers. What am I missing? (I haven't done any Windows work in 5 years).

drcode 17y ago
The part of the post about the font rendering is BS anyway. The picture is supposed to show "sub-pixel positioning" and it only shows antialiasing (not even sub-pixel antialiasing) which as you've pointed out isn't handled by a browser anyway. "Sub-pixel positioning" on the other hand would indeed need to be handled inside of the browser, so that the engine knows how much space something takes up so it can position the next item- But the picture accompanying the post is wrong/irrelevant.
- herf 17y ago
  Not sure I agree - most fonts are cached (as pixels) at a few offsets, so the ability to place a character is constrained by 1-pixel increments (or 1/3-pixel increments in the case of ClearType). A true vector placement of each character gives you the ability to show type at any subpixel offset and then anti-alias, and I think there's a good chance they're doing this. See this Loop/Blinn paper for more details: http://research.microsoft.com/en-us/um/people/cloop/loopblinn05.pdf http://research.microsoft.com/en-us/um/people/cloop/loopblin...
wmf 17y ago
As I understand it, GDI+ does software rendering into a window buffer which then gets hardware-composited by DirectX. Direct2D now does hardware-accelerated rendering and compositing. http://msdn.microsoft.com/en-us/magazine/dd861344.aspx http://msdn.microsoft.com/en-us/magazine/dd861344.aspx http://blogs.msdn.com/directx/archive/2009/09/29/comparing-direct2d-and-gdi.aspx http://blogs.msdn.com/directx/archive/2009/09/29/comparing-d...

drcode 17y ago

The part of the post about the font rendering is BS anyway. The picture is supposed to show "sub-pixel positioning" and it only shows antialiasing (not even sub-pixel antialiasing) which as you've pointed out isn't handled by a browser anyway. "Sub-pixel positioning" on the other hand would indeed need to be handled inside of the browser, so that the engine knows how much space something takes up so it can position the next item- But the picture accompanying the post is wrong/irrelevant.

herf 17y ago
Not sure I agree - most fonts are cached (as pixels) at a few offsets, so the ability to place a character is constrained by 1-pixel increments (or 1/3-pixel increments in the case of ClearType). A true vector placement of each character gives you the ability to show type at any subpixel offset and then anti-alias, and I think there's a good chance they're doing this. See this Loop/Blinn paper for more details: http://research.microsoft.com/en-us/um/people/cloop/loopblinn05.pdf http://research.microsoft.com/en-us/um/people/cloop/loopblin...

wmf 17y ago

As I understand it, GDI+ does software rendering into a window buffer which then gets hardware-composited by DirectX. Direct2D now does hardware-accelerated rendering and compositing. http://msdn.microsoft.com/en-us/magazine/dd861344.aspx http://msdn.microsoft.com/en-us/magazine/dd861344.aspx http://blogs.msdn.com/directx/archive/2009/09/29/comparing-direct2d-and-gdi.aspx http://blogs.msdn.com/directx/archive/2009/09/29/comparing-d...