clarify this

macgregor

Veteran
"Another plus is that the GH1 is doing a better job of scaling and processing data from the sensor than the 5D Mark II because it isn't line-skipping when scanning the sensor. Instead it bins individual pixels - just like when you downsize an image in Photoshop."

Ok, i´ve read this in several places now but there is something not making sense here.

If the GH1 was downresizing the full sensor to 1080p, there would be no trace of aliasing. And I see moire and chroma aliasing in many shots and in Barry´s ISO chart tests.

So where does this info come from? Can someone clarify?
 
It comes from people who don't know what they're talking about.

There is definitely aliasing in the GH1 image. It just doesn't have the same rainbow purple/orange aliasing that the Canons have. But it has about the same overall amount of luma aliasing. However, in 720p mode it's much better than the Canon's 720p mode.
 
Well, that info is posted everywhere, from provideo coallition to Bloom´s site.

It´s a serious mistake and nobody seems to notice.
 
Lots of horsecrap gets repeated around the net.

Facts are facts, and they're easily demonstratable. The truth is never dependent on "what people say" or whether people "believe" it... the truth is just the truth, no matter how unpopular it may be.
 
"Another plus is that the GH1 is doing a better job of scaling and processing data from the sensor than the 5D Mark II because it isn't line-skipping when scanning the sensor. Instead it bins individual pixels - just like when you downsize an image in Photoshop."

Erm. Makes sense to me. This is not saying the whole sensor is downscaled. The GH1 bins pixels rather than whole lines. That's why it has better (but still not perfect) moire / aliasing.
 
But photoshop doesn´t bin pixel packages, photoshop downresizes using all the pixels of an image.

So does pixel binning create less aliasing than line skiping? It looks like, but the whole paragraph is missleading because it sounds like you have the full sensor downresized as F35 or Alexa do. I belive that´s not really happening.
 
Last edited:
Pixel binning example
pixelbinningfigure1.jpg
 
So does pixel binning create less aliasing than line skiping? It looks like
Okay, let's go back to the source.

WHO SAYS THEY'RE PIXEL BINNING???

WHO SAYS THEY'RE LINE-SKIPPING???

None of that has ever been confirmed by the manufacturers. So when people authoritatively declare "Canon uses line skipping, Panasonic uses pixel binning" they're talking out their ***. They DON'T KNOW. They're guessing.

Nobody knows.

Some pretty well-reasoned discussions have taken place that come to the conclusion that the Canon is line-skipping; there's one on dpreview that looks compelling, but it's still speculation. And, furthermore, Canon has flatly DENIED line-skipping; they said that they instead use pixel binning.

So, here's the upshot -- nobody, outside of the sensor engineers inside Panasonic and Canon (or wherever they're buying their sensors from) knows. They're all guessing and speculating.

Here's the one and only thing we do know: the Panasonic's image looks a lot better than the Canon's. Whatever they're doing, they're doing it better. There's aliasing, yes, but the overall image (as evidenced by the GH13) is substantially sharper, and there isn't nearly the same level of chroma aliasing and purple/orange rainbow moire. So it is fair to say that Panasonic's processing is "better", absolutely yes it's fair to say that. But to attribute that difference to "pixel binning" vs. "line skipping" is unsubstantiated speculation. It may even be true, or it may be nonsense, we don't know. Only the sensor engineers really know.

, but the whole paragraph is missleading because it sounds like you have the full sensor downresized as F35 or Alexa do. I belive that´s not really happening.
We're all pretty sure that they're not using the full sensor downsized. That part I'll buy into. And I think that they must be using some sort of pixel binning or line skipping combination -- there really isn't anything else that makes a lot of sense. But it's still speculation and guessing, because we don't know. I don't know how they could get the full field of view, at the full frame rate, without some sort of skipping or binning scheme, or combination thereof. But which one bins, and which one skips? We don't know. Or do they both do it? We don't know. Or do they both exclusively skip, and ignore binning? We don't know. Or do they both bin, and ignore skipping? We don't know. If Canon is to be believed, they don't skip rows at all, they use a "six-pixel bin".
 
6 pixel bin is non-square no matter how you rotate it. So that's where the aliasing might come from in Canons. OTOH if Panasonic is using a square bin, so that the resolution of the sensor area it reads the data from is exactly N times 1920x1080 - there shouldn't be much of the aliasing. It's like resizing a picture to 50%, 25%, 12,5% etc which is exactly 2,4,8 times. It can be done with good results yet without sophisticated algorithms that tax processor with lots of calculations. Again, these are just mere speculations.
 
My GH1.4 has a photoshop mod in it and sends each 12 megapixel frame into photoshop to be resized before storing it in avchd, and it does this is real time while also driving me to theme parks. It makes a wicked whiskey sour too.
 
Back
Top