Hello, friends. I'm Bagho from India. I have a doubt about HD video that I need to clear up.
The full HD Sony HDR CX-210 has a sensor of 1.3 megapixels effective resolution. To make the video footage full HD (2 megapixels,) the camera interpolates the 1.3 megapixels into 2 megapixels, thus making the picture fuzzy and it looks over-processed.
Now, the full HD Sony HDR CX-220 (which is cheaper than the 210,) has a sensor of 2.2 megapixels effective resolution. So it doesn't have to interpolate to make the footage full HD, and we expect the footage to be much sharper than the CX-210.
But, reviews say that CX-220's picture quality is much worse than the CX-210. Some say the 220's footages are very un-sharp. Why is this? The sensor produces more pixels than a full HD resolution.
Could you please explain?
Thank you.
Effective resolution of camcorder's sensor - does it matter?
Re: Effective resolution of camcorder's sensor - does it mat
Hi, yes they do need to interpolate because the sensor isn't the exact shape or size of Full HD spatial resolution (1920 x 1080 pixels), or even orientation (Exmor sensors are tilted 45 degrees).
There are so many factors that contribute to image quality, it's not just sensor resolution. I realize that they are much more expensive, but the 7xx series has outstanding picture quality.
There are so many factors that contribute to image quality, it's not just sensor resolution. I realize that they are much more expensive, but the 7xx series has outstanding picture quality.
Re: Effective resolution of camcorder's sensor - does it mat
The task of upscaling sensor resolution is much easier than downsampling.
That means that with the CX210, the image is oversampled to fit into the 1080 samples per line, but the resultant image presented to the AVCHD coder and the resultant stream shows little deterioration from the quality of the sensor itself when displayed on a monitor. The oversampling cannot improve it in real terms but also, it will not reduce the quality that much.
Downscaling is however a much more difficult task. The main problem is that fine detail that is picked up by the higher resolution sensor will create moire, aliasing and other picture artifacts if not filtered out before downscaling. This can of course be performed by using an optical low-pass filter in front of the sensor. With consumer cameras, manufacturers want to sell a still feature capability so reducing the sensor resolution would be detrimental to the feature set. So an alternative method is to process the raw data stream with a digital filter before subsampling the stream to a lower resolution. This required complex real-time processing using mathematical filters such as Lanczos or Spline, which means costly digital hardware and the power to run it. This is incompatible with manufacturer's objectives for low cost consumer hardware. So their solution is to used substandard filtering to reduce the artifacts produced by the subsampling, which also softens the wanted detail.
That means that with the CX210, the image is oversampled to fit into the 1080 samples per line, but the resultant image presented to the AVCHD coder and the resultant stream shows little deterioration from the quality of the sensor itself when displayed on a monitor. The oversampling cannot improve it in real terms but also, it will not reduce the quality that much.
Downscaling is however a much more difficult task. The main problem is that fine detail that is picked up by the higher resolution sensor will create moire, aliasing and other picture artifacts if not filtered out before downscaling. This can of course be performed by using an optical low-pass filter in front of the sensor. With consumer cameras, manufacturers want to sell a still feature capability so reducing the sensor resolution would be detrimental to the feature set. So an alternative method is to process the raw data stream with a digital filter before subsampling the stream to a lower resolution. This required complex real-time processing using mathematical filters such as Lanczos or Spline, which means costly digital hardware and the power to run it. This is incompatible with manufacturer's objectives for low cost consumer hardware. So their solution is to used substandard filtering to reduce the artifacts produced by the subsampling, which also softens the wanted detail.
Re: Effective resolution of camcorder's sensor - does it mat
That was an excellent and detailed summary by Steve.
Sensor design and the data processing to get from the raw sensor data to the final data stream is interesting stuff.
Sensor design and the data processing to get from the raw sensor data to the final data stream is interesting stuff.