What's -3dB technically ?
 

Hi :)

Years ago I always used the -3dB setting on my first DV-Camera. But to be honest - not really knowing why. Someone had suggested it I guess or had given me a reasonable explanation. Whatever ...

Now I found the -3dB setting option again in the EX1.
So what's happening there ?

Of course the Video-Level seems to be lowered by 3dB. But what for ?
Noise reduction ? And is it really a noticable effect ? - besides getting darker :)

Thanks!

Peter

Wow I always thought it was just a sneaky way for manufacturers to claim low light performance (by saying "this is the image you get with 0db," when it was actually +3db, kinda). Didn't have time to read the article, just looked at the pics ; ). Plan on checking it out when I have more time.

JS

Thanks!

Very interesting.

Peter

I'm not sure I agree with that article.

A lot may depend on the camera, but it is possible with cameras (I've just tried it with a DSR500) to use -3dB and have no problem at all with reaching 100% white as the camera is further opened up. If people are experiencing an issue with the SDX900 whereby they can never reach 100%, I suspect it's a bug with that camera, rather than an issue intrinsic to negative gain. It's conceivably due to a line up issue, and lessening the amount of "knee" would make the problem go away.

Assuming what the article describes as "Sensor Dynamic Range" shows the chip output for increasing light level, then for most modern cameras at 0dB, the point at which it reaches a limit should be about 6x the amount it needs to make a nominal peak white. It's this headroom that gets exploited via a knee to retain highlight detail. Hence, in the model as used in the article, the left hand edge of the box should be near the left hand side of the "Sensor Dynamic Range" window, the right hand side about 1/6 of the way towards the right hand edge.

Altering gain then should be seen as moving the right hand edge of the box, whilst keeping the left hand stationary - not keeping the box the same size whilst moving both edges. The latter would be equivalent to altering the black lift level. Making the box bigger is equivalent to a reduction in gain, making it smaller an increase.

Not sure? Just think what happens to noise levels. Lets say noise on the chip "jitters" 0.1% of the total dynamic range of the sensor, and the window for a given gain setting extends for 1/10 of the total dynamic range. The output noise will be 1% of the width of the window. Keep the box the same size - just move it horizontally - and nothing will change. Make the box half the size (increase gain) and the output noise is doubled to now be 2% of the new window size. Which is what happens in real life.

It's true that negative gain will affect a cameras ability to handle highlights, but generally it would need to be far more than -3dB before the chip saturates before reaching 100% output.

It's even less clear with the EX as if you use Cinegammas 1, 3 or 4 at 0db the camera records up to 109%. At -3db it records to 104%. So in both cases it is still over 100% so after grading/legalising the recorded dynamic ranges will be very similar. So similar that for me the reduction in noise is well worth the tiny trade off in dynamic range, if there even is one.

As for reducing the ability to "see into the shadows". I don't see how reducing the gain affects this as in most normal situations if you reduce the gain you would open the iris up to compensate.

I'd missed that on first reading, but Alister is absolutely right.

Reduce gain and levels of shadow detail will indeed go lower (along with the rest of the picture), but open the iris to restore correct exposure and the level of shadow detail will come back to what it was before.