The specs of the F3 have me drooling. However, I just finished perusing the BBC analysis for the EX3 and I noticed they rated the EX cameras at 44db. Much lower than Sony's stated 54db. Also, much has been written about the noise coming from the EX's. I love my EX3 and I have captured beautiful, clean pictures, but I've also captured a few noisy images as well. Anyone else think these figures may be just a bunch of marketing hooplah? I would love to see an independent test.

As soon as someone has a production release it will be put through it's paces for sure. Mid-Feb to late-Feb the camera will be in the hands of the first buyers.

Dear Friends,

The footage from an F3, as recorded on a quality extrernal HD-SDI recorder looks great.

That wouldn't be a 10 bit recorder, would it Dan? :)

Dear James,

Yes, it would.

I attended a F3 Presentation over a month ago at Sony Pictures Studios.

We were shown footage from the camera itself, and from an external high-end recorder.

I am not trying to be coy, I just do not remember which recorder was used but I think it was a Sony HDCam HD deck. But it could have been their upcomming SR-R1 recorder.

In any case, the footage was just outstanding and to me it showcased the camera's low-noise capabilities.

Also, a friend, Alister Chapman, has tested the camera at various gain levels, including +18 dB gain, which showed a very low amount of gain noise.

Dan,
Not sure if the footage you're referring to is the same demo that I've seen, but much of the externally recorded footage utilized an HDcamSR deck.
There was also some footage at the demo from a different piece that was recorded with the NanoFlash at 50Mbps LongGOP 4:2:2 (The Ice skating wide shots, as I recall), which also looked excellent.
Since at this point there's quite a bit of demo footage out there, it does get a bit confusing...

I'm hoping the April and NAB will shed some light on the future of external recording options.

I don't expect Sony's deck to be competitive with Nanoflash or Ki Pro pricing, but I'm hoping that it will be competitive with the Cinedeck Extreme. I'm aware that the Cinedeck has some issues - but it does set a baseline price for portable 4:4:4 recording.

Dan has elsewhere made compelling arguments for why the Nanoflash is a strong option against the Ki Pro Mini, and just as soon as the Nanoflash is available in a 10bit flavor, I think I'll agree wholeheartedly.

I realize Convergent Designs, like many companies, doesn't want to tip their hand on upcoming product development, but I'd have to assume they're watching the market trends and adjusting accordingly.

I'm holding off on a external recorder purchase until at least April. By then we should have some field reports on the Ki Pro Mini (and Atomos Ninja for that matter), and hopefully NAB will bring some fresh announcements.

The F3 is one of the quietest HD cameras I've ever seen. It certainly has a lot less noise than an EX1, and I do mean a lot less.

its accurate :)

I noticed they rated the EX cameras at 44db. Much lower than Sony's stated 54db.
From what I understand, the S/N of a camera will differ depending on whether it's measured in highlight or lowlight areas, and I believe it may be due to S/N being linear off the chip, but then affected by the gamma of the camera. Hardly surprisingly, manufacturers tend to choose what suits their marketing people best.

It may not matter which measurement method is used - but if you're comparing cameras it needs to be consistent from model to model.

The F3 is odd because of what Sony claims its "natural" ISO is vs What the menu states as +0dB.

They claim 63dB S/N but is that on the 800ISO ( +6dB) claim of base sensitivity? or the +0dB which is 400 ISO in the camera. Mind you at ISO800 its grain less. At ISO400 the image is eerie its so spotless.

Just for reference ISO 800, looks like ISO 160 on my 5DmkII. ISO 3200 visually looks like ISO 640 on my 5D.

Does the F3 have an ISO indication option? Everything I've read says 800ISO 63db SNR. I always assumed that was at 0db. It's also indicated that in S-Log it's 1600ISO and 57db.

Where did you get 400ISO from?

In the menu you can select the camera to show "Gain" in ISO values. When you select your gain presets, LMH, -3 = ISO280, 0= ISO400, +3 = ISO580, +6=ISO800, +9= ISO1100, +12=ISO1600, +18=ISO3200

Essentially, 0dB = ISO400. Seems odd that they didnt make 0dB 800, but i suppose there is no "-9dB" for ISO280 in the broadcast world.

Mitch Gross just posted this on the other forum. "On the F3, 18db is indeed 6400 ISO. We have tested this and it tracks accurately."

There is something odd with yours if 0db is 400 in the menus?

Well i mean, if you select "L" on the gain switch -3dB and put it here, the camera says "ISO280". If you set 0dB to "L","M" or "H", it says ISO400.... i mean if 18dB is ISO6400 .,.... thats the best ISO6400 i ve ever seen....

Wouldn't this depend on wether you are using an interlaced or progressive mode? 1080i gives you one stop over 1080p. I think I recall reading that the F3 is ISO800 at 0 dB in interlaced mode. Hence ISO6400 @ +18dB would be correct. Right?

So I did some tests in the standard REC709 gamma mode. And here is how it rates. Amazingly clean through the range. This is the test I showed Mitch.

-3 db - 640 ISO
0 db - 800 ISO
3 db - 1250 ISO
6 db - 1600 ISO
9 db - 2000 ISO
12 db - 3200 ISO
18 db - 6400 ISO

Tests where completed with DSCLabs chart, Sekonic Meter, and Leader scopes. I preformed the test under both tungsten, and daylight balanced lights for accuracy. Camera was at 1080 23.98, 180 degree shutter, Gamma = REC 709, Auto Knee OFF, Black Gamma = 0, Black Level = 0 (edit).

Using the Cine gamma modes (Cine 1,2,3) lowers the gray point effectively lowering the ISO value, but for comparing to any DSLR these numbers are accurate.

I'm going to post my findings on our blog in a couple days with pictures.

Andy

Ola's right. It would be more helpful if the tests would include the video format. Personally, I will be using 1080 24p for most of my shooting.

Good point. I'll add that above.

Thanks Andy.

A couple stills from my test.

I grabbed this out of Clip Browser (note the settings):

http://blog.abelcine.com/wp-content/uploads/2011/02/F3_6400.jpg

Frame grab (off EX recording):

http://blog.abelcine.com/wp-content/uploads/2011/02/F3_6400_2.jpg

Jeez. astounding.


I was testing in cinegamma.

Lemme reup, and match your settings.

(edit)

The only way i got the camera to Read 6400 ISO was in 59.94 Interlace mode. Gotta bust out the light meter.

I didn't test the camera in ISO mode vs dB gain mode. So it could possible be limited in someway. But I don't see why shooting at a higher frame rate (59.94i) would effect ISO options. I'll give a more detailed report tomorrow.

Funny enough Sony's manual even suggest that 18 db is the same as 6400 ISO. Excerpt from F3 manual:

http://blog.abelcine.com/wp-content/uploads/2011/02/ISOChange.jpg

well it seems like your readings are based on actual exposure, not what the camera "says". I trust a light meter. If 18bB is 6400 ISO, and it looks like that! its astounding.

Andy,

Shooting at a higher frame rate may not affect ISO settings but shutter speed/angle definitely will. Also, as Ola stated previously, interlaced appears to be more sensitive than progressive.

Thanks,
Daniel

Thanks Daniel

Shutter speed is a limitation in exposure length, so it would not effect ISO. As I decrease my shutter angle I am reducing exposure length and my image gets darker, this is not a change in sensitivity just exposure. Increasing my frame rate has the same effect, regardless of if the camera is capturing an interlaced or progressive image it will always have a reduced length of exposure.

In 59.94 mode it is possible they are trying to compensate for the shorter exposure time by adding additional gain, but that does not seem to make sense. I'll do some more testing, and if anything it should only be dependent on gamma mode.

Another note.. The F3 appear to always capture images progressively, if you turn on dual link output when in 59.94i you will get a 59.94 progressive out of the dual link connector. So the interlaced recording is just pulling odd and even fields from the progressive stream.