Truth Behind the ClearType/OpenSUSE FUD

Kennon writes "Steven J. Vaughan-Nichols over at Linux Watch clears up the FUD around Tuesday's Slashdot discussion concerning OpenSUSE, ClearType, and patent deals with Microsoft."

Comments

Re:This hurts my head

[10scjed]

According to Microsoft's attorney, software patents are invalid in the US as well:

MR. OLSON [For Microsoft]: The '580 patent is a program, as I understand it, that's married to a computer, has to be married to a computer in order to be patented.
JUSTICE SCALIA: You can't patent, you know, on-off, on-off code in the abstract, can you? MR. OLSON: That's correct, Justice Scalia.
JUSTICE SCALIA: There needs to be a device.
MR. OLSON: An idea or a principle, two plus two equals four can't be patented. It has to be put together with a machine and made into a usable device.

Re:This hurts my head

[squiggleslash]

There seems to be a confusion of terminology in the above, but I admit I may be misreading it.

Anti-aliasing is not the same thing as sub-pixel rendering, which is orthogonal to anti-aliasing and can be (and almost always is) combined with it.

Anti-aliasing is merely the use of different shades to adjust the sharpness of object boundaries, where the shade is based upon the amount of a pixel the objects covering that pixel would intersect. While this sounds like something that would be describable by the term "sub-pixel rendering" if, for a moment, you assume you would divide the pixel into smaller virtual pixels to calculate the end result, that's not what sub-pixel rendering refers to. The term "sub-pixel" is not being used to describe these smaller "virtual pixels".

In an LCD a pixel is made up of three "sub-pixels": real, discrete, lighting elements that together illuminate one complete pixel. The sub-pixels are the three primary colours and are almost always mounted side by side as three thin strips. Sub-pixel rendering is the technique of using the separate red, green, and blue sub-pixels of an LCD "pixel" in isolation to improve the sharpness of object boundaries. When used, the screen effectively has an increased horizonal resolution of 3x the regular resolution, so a 1400x1050 screen effectively becomes 4200x1050.

It is usually, if not always, used in conjunction with regular anti-aliasing (though technically it doesn't need to be.)

Microsoft's patents, as I understand it, cover the latter, and in particular focus on preventing "colour fringing" that is otherwise a major downside of using sub-pixel rendering.

Re:ClearType draws from Apple II, says developer

[JavaBear]

Gibson have a "Who Did It First?" text regarding sub-pixel rendering as well:

http://www.grc.com/ctwho.htm

Re:Changed Before the Microvell Deal

[kripkenstein]

This change was last summer, pre-microvell, so the news actually would have been if OpenSUSE was enabling it and taking advantage of MS' patent covenant for Novell customers and OpenSUSE contributors while other distros couldn't.

Yes. I submitted the previous story about this matter, and I stand corrected. I didn't know everything about the issue; I relied on the sources reporting on it, and mainly, the whole matter seemed suspicious, and I thought posting it to Slashdot would shine some light on it. Not 100% sure if doing so had an influence, but what matters in the end is that things are now clarified.

From my standpoint, the interesting issue that remains is what I mentioned in the little comment at the end of my submission for the previous story: ok, assuming there are MS patents on this technology, isn't Novell licensed to use them now (even if it "isn't a patent license", but it just acts like one)? Apparently the Microsoft-Novell deal doesn't help openSUSE out much with regard to MS patents. Is the same true for SUSE?

Re:OS X's subpixel rendering? Adobe's "CoolType"?

[TheRaven64]

You can still do sub-pixel AA on a CRT, you just can't use the same algorithm as on an LCD. The trick behind sub-pixel AA is to realise that any adjacent group of red green and blue emitters can be regarded as a pixel, not just those that are exposed as a pixel by the hardware. On an LCD, it's easy because you have a nice regular RGBRGBRGB pattern. You can trea a GBR, BRG or RGB run as a pixel, and just alter the colours for the hardware pixels to turn on the individual emitters as required. For CRTs, it's a little bit more difficult, but it's still possible.