What's the deal with that setting? Why isn't -3db=0db and 18db=21db? Lower latitude? Worse highlight-handling?

gain settings are really arbitrairy but the MINUS (-) settings are generally used in outdoor work OR EXTREMELY bright indoors (never seen one like that EVER but I suppose it could be)

Like I said though, gain seetings are different for every manufacturer, Sony, Canon, JVC-there is no way that a JVC at 0 will be exactly the same as Sony at 0. When I had my JVC5000 and used it with a Sony PD150 one would think the 1/2 chip would be all over the 1/3 chip. Nope! In unscientific test, I used the 2 cameras in the same conditions right next to each other. The 150 had it all over the JVC. Now part of that was the Canon 16X lens (the 19X was a killer compared to the 16 but) IIRC the Sony was at 1/60th at f/2 and 0 gain-the jvc was at 1/60th,f/2 but I had to go to +6db of gain to get a quality image. BTW this was done with 2 properly calibrated production monitors not LCDs. When I dropped the footage into my NLE you didn't have to be a genius to see the difference.
So long story short, -3 can be a very helpful setting in a very very bright and contrasty outdoor setting.
Don

gain settings are really arbitrairy but the MINUS (-) settings are generally used in outdoor work OR EXTREMELY bright indoors (never seen one like that EVER but I suppose it could be)
1/sqrt(2) isn't much help compared to a 1/8-ND. It's more like a water drop on a burning rock.
there is no way that a JVC at 0 will be exactly the same as Sony at 0.
Of course, because they can have different iso-ratings. But that wasn't my assumption/question.
Again:
Why do they call it -3db instead of 0db (and hence 18db instead of 21db)? Just for fun? No drawbacks?

Who knows? I would guess that it's simply that the first set (Z1U) HDV cameras didn't even have a decent ISO 100 or ISO 200. I'd like to see an ISO setting as well as 1/3 stop shutter speed settings.

I'd also like some more metering functions such as lost highlights and lost shadows, but DSLRs don't even have them. The live histogram is great, but there are so many useful metering tools that could be added.

I would guess that it's simply that the first set (Z1U) HDV cameras didn't even have a decent ISO 100 or ISO 200.
I don't understand. Do you mean a "EX1-0db-noise = Z1-0db-noise"-idea?
DSLRs
Apropos DSLRs: I know the Canon 5D which has a "L"-iso-setting which is a iso-50-setting with lower latitude than the other iso-settings. That's the source of one of my guesses.

The gain function increase (or decreases) sensitivity. 0dB gain means you are seeing what the image sensors are putting out, with all the other processing being done. +3dB means they have electronically increased the output with an amplifier. -3dB means they have electronically decreased the output.

Different manufacturers have different relative gain structures because of the different design of the cameras--you can't even compare cameras of the same manufacturer unless they use the exact same imaging device.

It is exactly like audio gain. Add it when you need more or subtract it when you need less. That includes the same penalty of gain--increased noise. However, audio negative gain usually doesn't decrease the noise the way video negative gain does.

The gain function increase (or decreases) sensitivity. 0dB gain means you are seeing what the image sensors are putting out, with all the other processing being done. +3dB means they have electronically increased the output with an amplifier. -3dB means they have electronically decreased the output.


The odd thing about the -3 dB setting is that it is apparently the optimal setting for low noise. A lot of DSLRs have a Base ISO (usually 100 or 200), but offer some settings below that that are digitally attenuted. We'll need a real answer from Sony for that. A -3dB digital attenuation may reduce the noise, but it would also reduce the headroom.

Looking at the SAW and WFM while switching from 0dB to -3dB will reveal that the latitude is slightly reduced at -3dB. Clipping happens around 93%, ILO 100%. You gain with a lower noise floor, however.

Bill,

Two acronyms I can't decypher in your post: SAW and ILO. Can you (or anyone else) shed some light on these?

George/

ILO...in liew of

SAW....I don't know what this stands for, however, it is a kind of standardized test image for displaying the gamma curve for a camera. In the EX1, the SAW curve is built into the "maintenance" menu. It consists of a luma only image gradient, extending from 0 to 100% IRE. Displaying this image, while looking at a WFM will give you the gamma curve used by the video camera. In the case of the EX1, at -3dB gain, the gamma curve abruptly clips around 90%.

I guess SAW is not an acronim, but it means what it means -- after the shape of the oscillo pattern.

ILO...in liew of

SAW....I don't know what this stands for, however, it is a kind of standardized test image for displaying the gamma curve for a camera. In the EX1, the SAW curve is built into the "maintenance" menu.

Hi Bill,

Thanks for pointing me to where to find the saw curve/gradient in the EX.

I think it was a case of early-morning acronym overload ;-)

George/

Well, as long as everybody is taking their shots at this, it's "lieu" and I think SAW is an abbreviated form of "sawtooth pattern" from o'scope usage.

Looking at the SAW and WFM while switching from 0dB to -3dB will reveal that the latitude is slightly reduced at -3dB. Clipping happens around 93%, ILO 100%. You gain with a lower noise floor, however.

Thank you for the measurements. So, 0dB is the native dynamic range. Big question answered.

I shoot everything that I can at -3db. A lot of the bigger broadcast cameras had a -3 db setting.

It's like shooting with a low ISO film (back when we used to shoot film).

Daniel Weber

works great as long as the dynamic range of the scene is fairly forgiving, The problem with most inexperienced shooters, is that they really don't get when it's NOT ok. the problem I've found with this forum is that there's LOT of inexperienced shooters....LOL

So -3db definitely isnt the no gain? I wondered if maybe they just decided that noise was so acceptable at with some gain that they called +3db 0db and 0db -3db. i guess i just wasnt sure if it was even possible to actually apply negative gain to sensor output. how sure are we that -3b,0db,+3db,+6db etc actually refer to what the hardware is actually doing in absolute terms, and not just names they decided on? like for all we know sony engineers actually know what we call 0db gain as +11.4db (or some arbitrary number) and -3db as +8.4db, or is it common practice for this actual numbers to carry over to the end user? if someone has insight on this id be very interested to hear it. And if we can confirm that -3db is actually what it says it is or that the the changes we see between +3/0 and 0/-3 are actually different (lower latitude etc).

so if -3db causes clipping earlier, is it possible that it is just not calibrated correctly? maybe working in -3db requires different gamma/knee adjustments but will actually give greater latitude if set up correctly?

Bill, That's a fascinating observation.

I've always assumed (and most people I know) that -3DB was a marketing gimmick. That the camera manufacturer in an effort to make the camera look faster than it really was arbitrarily picked a certain sensitivity with what they felt was an acceptable noise level and called it "0". Then to accommodate people who wanted less noise they invented -3DB which should have actually been "0". Thus most people i know always prefer to shoot at -3DB when possible. It also always been known tha one cameras noise at "0" might be the same as another's at -3 or another's at say +3DB.

However you are introducing an entirely different idea - that "0" might really be the ideal setting for a camera including both noise and dynamic range as well as other possible parameters. Interesting. I'd love to hear from a Sony engineer about this or perhaps Adam Wilt our resident tech guru.

The only exception I know of to the general rule to shoot at -3DB is the SDX900. All the techs I know soon discovered that the SDX lost dynamic range in the highlights at -3DB so we all stopped using it there. Hurt flesh tones in particular.

I guess I'll have to test for this.

Do you know of a place on the web that explains the SAW gamma patterns well? I'm used to judging gamma looking at a waveform of a chip chart but the SAW patterns are kind of greek to me.

Lenny Levy