What Are P.A.P. Filters?

[THoff]

In a camera like the HPX250 an 8bit 100Mbit coedc would probably be better then a 10bit 100Mbit codec. This could be tested with an externall recorder or something...

You could use DVCProHD on the HPX250, that is 4:2:2, 8 bit, 100 Mbps. Of course the horizontal resolution is reduced and the compression is not wavelet-based.

[Mike Warren]

There is an argument that a 10bit codec doesn't really pay of in a small chip camcorder.

If each step represents the same analog voltage difference I'd be inclined to agree with you, but it may be (I hope) that what it represents is finer graduations.

[KINOKS]

You could use DVCProHD on the HPX250, that is 4:2:2, 8 bit, 100 Mbps. Of course the horizontal resolution is reduced and the compression is not wavelet-based.

If resolution isn't such a problem, then maaaybe. But AVCi is a better codec so it probably looks better in any case.

If each step represents the same analog voltage difference I'd be inclined to agree with you, but it may be (I hope) that what it represents is finer graduations.

Sorry but that's to techy for me Care to elaborate?

[Mike Warren]

Sorry but that's to techy for me Care to elaborate?

Noise is more apparent in the shadows. If adding extra bits just increased the voltage range covered then it wouldn't affect noise, but because a video signal is a specific voltage, what the extra bits do is produce finer graduations in the voltages. Because the final output is nearly always 8 bit, that means much more adjustment is possible during post before banding and posterization will appear.

Noise is an analog voltage between zero and some maximum. if the noise voltage at a particular pixel is more than half a bit in strength, that bit is set, so with finer graduations, some of the noise will translate to a lower level.

This all assumes the signal is being captured at the sensor at 10 bits or more.

[KINOKS]

Noise is an analog voltage between zero and some maximum. if the noise voltage at a particular pixel is more than half a bit in strength, that bit is set, so with finer graduations, some of the noise will translate to a lower level.

So with finer graduations you're referring to 10bit right (in contrast to 8bit)? And to a lower level you mean to a lower voltage?

[timbook2]

That's not necessarily PAP1. I was shooting the annular solar eclipse with pretty much the same settings. ND all the way, fast shutter speed and closed down the iris. Lots of noise in the dark areas. Backed off the ND one and opened up the iris some and noise was gone.
PAP was set to default.

Ok thanks, good to know. I have learned to switch PAP and make regular use of it for daylight and dark shots. The noise issue I had with 1/64 ND and iris @8 was due to the fact that PAP1 was on. With PAP2 I would have never been necessary to close down that far. I always prefer to slightly underexpose if possible, havent had time to see if blown whites can be corrected in post even if they are 10 bit:-) (I doubt it).

[KINOKS]

havent had time to see if blown whites can be corrected in post even if they are 10 bit:-) (I doubt it).

Clipped means clipped! nothing you can do

[Gabrobot]

I always prefer to slightly underexpose if possible, havent had time to see if blown whites can be corrected in post even if they are 10 bit:-) (I doubt it).

Typically I find it saves stuff above the clipping point, so if you're working in a program that works in 32bit floating point space, you can recover a certain amount of the clipped highlights. I've also found that thanks to the fidelity of the 10bit codec, you can bump your pedestal way up to like +20 so you can underexpose and get more highlight range.