Sony F5 Native ISO Tests

There continues to be some considerable debate around the native sensitivity of Sony's F5 (and the FS7 which supposedly shares the same sensor), so I figured I'd share my tests with everyone in the hope that it clears things up for people.

For these tests I left the camera set to its default setting of 2000 EI and shot an 18% grey card in SLOG2 rating the camera at 1000 ISO, 1250 ISO and 2000 ISO as well as using the onboard false colour monitoring to achieve perfect green over the grey card.

I recorded both 4k raw, and 2k XAVC. But since the waveforms were identical between the two formats, I'll just post the waveforms from the raw files for the sake of clarity. 

Now before we get started, here's the breakdown of 0% black, 18% grey and 90% white values for Sony's three SLOG gamma curves: 

My tests were recorded in SLOG2, so when perfectly exposed, my 18% grey card should sit at 32% IRE (Code Value: 347) on the waveform.

Now on Resolve’s waveform display, we are given Code Values (CV) between 0 and 1023 broken down into 32 CV portions. So the values between the 256 and 384 markings on the waveform (the relevant section for us, since this is where our 18% grey card is placed) are as follows:

256, 288, 320, 352, 384

Sony tells us that 18% grey should fall at 347 CV, so we’d expect to see (on a perfectly exposed image) the meat of the line that defines our grey card on the waveform, hovering just under the first marker below 384 on the Resolve waveform scale.

So with this in mind, when we look at the waveform for the 1000 ISO image, we see that we’re pretty close to that, the centre of our line denoting the grey card, sits a little above that 347 point:

1000 ISO:

Moving on to our 1250 ISO image, we see a similarly close result, the meat of grey card line is sitting a little bit below that 347 point:

1250 ISO:

Moving on to the camera’s supposedly native sensitivity of 2000 ISO, we see that it’s not even close. The meat of the grey card line is sitting between 256 and 288. Well underexposed. 2000 ISO cannot be the camera’s native sensitivity, since it cannot provide the appropriate values on the gamma curve.

2000 ISO:

Now here’s where things get interesting. When we use the camera’s built-in false colour monitoring to set exposure, we finally get a perfect exposure. The F5’s false colour monitoring shows green between the ranges of 30.3-34.3% IRE. Slap-bang in the middle of that range yields us a perfect green over the entire surface of the grey card in the camera’s EVF. And the waveform bares this out, with the meat of our grey card line sitting right where it should, around the 347 CV point on the waveform scale:

Perfect Green with False Colour:

What’s interesting is that if we compare our various waveforms, we see that our perfect exposure (the false colour derived one), sits exactly between our exposures for 1000 ISO and 1250 ISO, 2/10 of a stop between them.

What does this mean? It means that the native sensitivity of the camera - that’s the sensitivity the sensor must be rated at in order to achieve correct exposure values for its SLOG gamma curves (I won’t speak for hypergammas/cinegammas since I’ve never used them) is halfway between 1000 and 1250 ISO. The camera’s native ISO is therefore approximately 1125.

Now this doesn’t effect exposure choices. People get themselves overly worked up about these things, and it’s silly. If you want to rate the camera at 2000 ISO, that’s fine, it will yield you an extra 4/5ths of stop of latitude in your highlights (at the cost of an equivalent amount of additional noise in your shadows). The reason it’s important to know the ACTUAL native sensitivity of the camera, is to know when you’re pushing the camera beyond it, and by how much. It simply gives you a clearer idea of how the camera will respond to various exposure choices. 

Personally, I rate the camera at 1000 ISO almost entirely (since my meter won’t go to 1125 ISO), only stepping up to 2000 ISO when I either have an extremely high key scene (and therefore want to retain more highlight detail), or when I simply run out of light on a night exterior.

Now what I haven’t had the chance to test yet, is how this ACTUAL sensitivity affects the distribution of the camera’s dynamic range above and below middle grey. Since Sony claims six stops of latitude above middle grey at 2000 ISO, the question arises as to whether this is six stops when the camera is actually rated at 2000 ISO (and is therefore 4/5ths of stop underexposed)? Or when the camera is correctly exposed with middle grey at 32% IRE (at 1125 ISO)?

At any rate, these are the test results, I’ve heard several people claim that they’ve seen (or heard of) tests confirming the camera’s 2000 ISO sensitivity, so I’d certainly be keen to see those tests (if indeed they exist). I’ve compared my meter to half a dozen others in order to confirm its accuracy and it appears to ring true. So I have little cause to doubt these results.

Hopefully this info helps clear things up for people.

Cheers,

Mark