CCD question
- Thread starter ernesto
- Start date
AuditoryVisuals
Well-known member
The HVX200 has 960x540 sensors. Which is 16:9 and one quarter the resolution of 1920x1080 then DVCPRO-HD puts it in 1280x1080. The XH A1/XL H1 has 1440x1080 sensors, which is the resolution of HDV. 1440x1080 is 3 quarters the resolution of 1920x1080, which means the XH A1 is 3x the resolution of the HVX200.
Elton
Veteran
The XL-H1 employs a relatively minor amount of pixel shift in order to internally generate a 1920x1080 image. (which is what is sent out the SDI spigot in square pixels) On the CCD the pixels are something like 1.2.1 PAR which gives it an effective resolution of over 1600 horizontal pixels, after which it performs horizontal pixel shift to achieve 1920.
It's legitimate resolution, and pretty amazing to achieve with a 1/3" sensor.
It's legitimate resolution, and pretty amazing to achieve with a 1/3" sensor.
Barry_Green
Moderator
HAH! Hahahahahahaha...AuditoryVisuals said:
No, not quite. 3X the pixels, yes. 3x the resolution? Not by a country mile. Not on this planet or any other.
Anyway, back on the original question: the pixels aren't square. The Canon cameras are using anamorphic pixels, wider than they are tall. Same thing the Sony does. The HVX uses square pixels. All three products use spatial offset to increase their resolved frame to 1920x1080.
Barry_Green said:
While Barry is right ( as usual) the XL-H1 is still perceptually higher res than the HVX200 by a decent amount. 3X...? Not that much. But you see a significantly sharper image than the HVX. While the amount of pixels isn't everything, it is still important.
I'd like to note that when viewing footage on a 1920x1080 LCD panel, everyone I had view footage in an A-B comparison between the H1 and the HVX found the H1 footage to look more "stunning" and "high def". The HVX didn't look bad mind you, just people were not blown away by the High Defness of it like the H1. These were non technical people who had no agenda either....they were just used to seeing HD like on Discovery and HDnet theater and felt the H1 looked closer to that than the HVX.
Personally I think the HVX looked more filmic and the colors were deadon.....but the res left me wanting. And upon playing with the H1 and using color correction I could get darn close the HVX look. However no amount of tweaking the HVX was going to give me the H1 look. And for me personally, P2 workflow was not a good option. I tried.
They each are great cameras and have their own strengths and weakness. It is a good time to be in the HD market.
Yas Kassana
Well-known member
What's the difference between resolution and pixels? I thought more pixels equals more resolution?
If you're talking about pixels as meaning "sampled points" (photosites on the ccds/cmos chip), then yes more pixels do equal more resolution. But from there on in, it gets more complicated!
A set of 1440x1080 chips can be 'spatially offset' (pixel shifted) to resolve more than 1440x1080. This does use interpolation, and the amount increase will vary (depending on the content of the image), but this increase is real and used in all cameras at this level.
scaling a 1440x1080 image to 1920x1080 has more pixels .. but the larger image doesn't have any more resolution than the original. Obviously you could take all of the outputs of all digicams and scale the result to 1920x1080-- same number of pixels, but vastly varying resolution.
There is also "perceived" resolution and "measurable" resolution. cameras can use sharpening to make an image "look" sharper, i.e. higher perceived resolution, but it will not resolve any more lines when tested with res charts. On the contrary, a camera can "resolve" a great deal of detail (in measured tests), while looking relatively "soft" to the eye, due to it's softness settings being low.
A set of 1440x1080 chips can be 'spatially offset' (pixel shifted) to resolve more than 1440x1080. This does use interpolation, and the amount increase will vary (depending on the content of the image), but this increase is real and used in all cameras at this level.
scaling a 1440x1080 image to 1920x1080 has more pixels .. but the larger image doesn't have any more resolution than the original. Obviously you could take all of the outputs of all digicams and scale the result to 1920x1080-- same number of pixels, but vastly varying resolution.
There is also "perceived" resolution and "measurable" resolution. cameras can use sharpening to make an image "look" sharper, i.e. higher perceived resolution, but it will not resolve any more lines when tested with res charts. On the contrary, a camera can "resolve" a great deal of detail (in measured tests), while looking relatively "soft" to the eye, due to it's softness settings being low.
I wrote the previous mail in a rush as I was leaving, so just going to add to it-
about the 3x more resolution thing!
this is fairly subjective, and you need to separate whether people are talking about perceived resolution or measured resolution.
Whether it's "3x" specifically-- how much more resolution does a 200x200 pixel image have over a 100x100 pixel image? It's 4x as many pixels (40k vs 10k), but only 2x as many "lines". So if we're talking in linear terms, I suppose you'd have to say that it has 3x the effective pixels and 1.73x the measurable linear resolution.
about the 3x more resolution thing!
this is fairly subjective, and you need to separate whether people are talking about perceived resolution or measured resolution.
Whether it's "3x" specifically-- how much more resolution does a 200x200 pixel image have over a 100x100 pixel image? It's 4x as many pixels (40k vs 10k), but only 2x as many "lines". So if we're talking in linear terms, I suppose you'd have to say that it has 3x the effective pixels and 1.73x the measurable linear resolution.
DavidBeier
Veteran
Barry also mentioned in a question of mine on the cinematography forums that a CCD pixel count never translates directly into the actual resolution. The CCD may have a 1440x1080 sensor but when the camera scans the image it won't neccesarily retain all that data.