F5: Beware, the camera's not really 2000 ISO...

Grug

Grug

Veteran
Hi guys,

I've been rating my F5 at it's native 2000 ISO from the get go, and never really questioned it. Between exposing via a combination of my light meter and (primarily) the camera's false colour mode, everything's graded up nicely in Resolve. However on a couple of recent jobs where I rented out the camera to production and then just Gaffed the shoots, I noticed that the results the DoPs from those shoots got with the camera were a little underexposed. Which struck me as odd, as I've compared my Sekonic meter alongside several other meters on stills shoots, and it's always returned matching exposure readings - so I've always felt confident in the accuracy of its metering.

To get to the bottom of this I've run tests on the last couple of shoots I've had, comparing my meter readings to the false colour values on the camera, and it appears my F5 (set to it's native 2000 ISO in SLOG2) is actually a whole stop less sensitive it's supposed to be - by my measurements it comes up as just 1000 ISO.

Now if you're shooting with the camera and exposing via the onboard false colour metering, this shouldn't be a problem as you just expose for middle grey, and so long as middle grey shows up green in the camera's false colour, you're properly exposed. However if you're exposing by a meter and rating the camera at 2000 ISO, you'll end up a full stop underexposed.

I've found this quite surprising as my F3 metered accurately at 800 ISO, so I certainly wasn't expecting this inaccuracy on the F5.

So I'm wondering, has anyone else tried calibrating their meter to their F5(5) yet? And if so, what rating did you come up with for your camera?

Cheers,

Mark
 
ISO is a rather random figure to be interpreted by you the operator - especially with wide DR log profiles.

Say a 10stop camera you could rate it at X and have 5stops of shadow and 5stops of highlight or rate it at Y and have 4 of shadow and 6 of highlight.

This is exactly what sony have done because it does two things.. give you loads of highlight headroom, give a headline grabbing fast base ISO (and gives you noisy files but that is not in the headlines!).

Another reason they could have come up with 2000 is that it is a mid point - aka the cam works between 1000 and 4000,

I use my FS7 at 800-1000 where possible and it is clearly cleaner.

You lose some highlight headroom - or have to lift the shadows using a thing called lighting - it works.

S
 
It's not random in the slightest. Exposure choices may be subjective decisions, but the base sensitivity of the sensor certainly isn't. Sony's own literature states quite clearly that when shooting SLOG2, correctly exposed middle grey is 32 IRE and white is 59 IRE.

And at 25fps and with a 180 degree shutter, the camera (well, mine at least) has to be rated at 1000 ISO to expose middle grey at 32 IRE and white at 59 IRE.

This doesn't affect the distribution of highlight/shadow latitude, merely the base sensitivity.

What would be interesting to see is whether other people's F5(5)s respond similarly (i.e. Their sensors are a stop less sensitive than they are officially rated to be).
 
I also rated my F55 one stop lower than it actually is. My test was scientific, yet it was refuted by other so called "experts" on the Internet who didn't test it for themselves but rather decided to only say that "Sony can't be wrong".

Same story as Mark's, the test I did was also replicated with the Sony F3 which proved to have an accurate rating. So even if I was "wrong" there is no way my testing setup can be "right" for one camera and not the other.

My findings were that the only way to get the actual ISO performance as rated by Sony is to set the "shutter" setting to "off". Which makes 100% perfect sense. Without any shutter implemented on a motion picture film camera it is called 360 degree shutter because imagine a mechanical shutter rotating completely around before the frame is even exposed which is letting more light in. 180 degrees is half of 360 degrees so it chops the 1/24th exposure in half to 1/48th. Hence this is why setting the camera's shutter speed to 180 is the same as selecting 1/48th shutter speed. (Sony F3,F5,F55 have this as an option).

No idea why Sony decided to do this. It was either a mistake or they simply did not want to market the F55 and F5 cameras as 625 ISO and 1000 ISO respectively. So it's highly plausible that Sony marketing (and/or engineer) geniuses used the "shutter off" setting to set the ISO. "

My facts, arguments and defense documented here over a year ago; : http://community.sony.com/t5/F5-F55/How-are-the-ISOs-derived/m-p/240751#M9069


[Edit] Ok, ok, in Sony's defense along with other camera manufacturers, in the world of digital cameras and according to ISO 12232:2006 there are in fact a variety of methods to determine ISO ratings which can explain why they are "different" than some of us might expect - BUT that is EVEN more reason if you are going to use light meters for gauging your exposure why you should be performing your own rating of equipment in a controlled environment before using that old trust light meter. It's imperative.
 
Last edited:
I suspect it's probably (infuriatingly) the marketing angle that's to blame. Dennis, do you put the camera into 625 EI when shooting with it? Or do you leave it at 1250 EI and simply rate your meter at 625 ISO?
 
I could use 650EI but for some reason I like to leave the camera in native ISO and just know that I'm 1 stop over on my meter if set to 1250, or bang on if meter set to 650. I still double check stuff on WFM before rolling but I am still a fan of setting up lights and stuff before the camera comes out.
 
Nice read, guys.
I am a run n gun ruff shooter and never use light meters or false colour, only waveform. I come from the DSLR ETTR method, and basically expose just to protect the highlights (always in RAW)

My point I'm tryna make is in line with what u guys are saying (well maybe not), just that I have noticed when I grade in Resolve, When I use 800 ISO, and a LUT, resluts are super clean amazing footage, and I don't see any highlight or skintone issues either.

Before that, I used to always keep the footage at native 2000 ISO, and grade down if need be, and the noise bugged me, even crushing blacks back down didn't make the footage clean (like Alister claims it does)
Low 800 ISO looks amazing and clean with F5 grading, and I even go down to 500 ISO sometimes when I can, with no highlight issues at all. It makes everything clean, not just shadows, but the mids as well.
Well thats just MO.

Oh, try Sgamut3.Slog3.cine+709 LUT in sony raw viewer after setting RAW ISO to 800/640, then tweak CSL to taste and tell me how wonderful it looks ;) (This is RAW Im talking bout, don't know much of XAVC, as I never use it )
 
Last edited:
Damn, sorry guys if my post was off the topic, but I suppose I just wanted to say that I think the native 2000 ISO of the F5 (or most so called native ISO of current cinema cameras, for that matter) are more marketing hype than reality.
And that Im supper stoked that Ive started getting much better results out my F5 with lower ISO's in grading. I was excited.

Take Red for example, many good DPs rate the Epic/Dragon at around 320 ISO as standard practice, and even good colourists beg DPs to rate Red at 320 ISO too.

But you guys are really talking about the discrepancies of claimed nativity of ISO from sony, using light metering, thats not lining up correctly, right?

Excuse my ignorance, I have no experience at all with light meters, nor do I ever use grey cards or expose for middle grey. All I do is light by eye/experience, and make sure my shadows are way above 20 IRE, and highlights are not near clipped.
 
Jusgorilla said:
Damn, sorry guys if my post was off the topic, but I suppose I just wanted to say that I think the native 2000 ISO of the F5 (or most so called native ISO of current cinema cameras, for that matter) are more marketing hype than reality.
And that Im supper stoked that Ive started getting much better results out my F5 with lower ISO's in grading. I was excited.

Take Red for example, many good DPs rate the Epic/Dragon at around 320 ISO as standard practice, and even good colourists beg DPs to rate Red at 320 ISO too.

But you guys are really talking about the discrepancies of claimed nativity of ISO from sony, using light metering, thats not lining up correctly, right?

Excuse my ignorance, I have no experience at all with light meters, nor do I ever use grey cards or expose for middle grey. All I do is light by eye/experience, and make sure my shadows are way above 20 IRE, and highlights are not near clipped.
Click to expand...

Yep, we're talking specifically about the inaccuracy of Sony's claimed ISOs for the cameras (which is what our meters have confirmed).

The important distinction I'd like to make, is that I'm exposing (and would recommend others expose) the camera for it's actual native sensitivity (1000 ISO), I'm not setting the camera to 1000 EI (which is actually 500 ISO) and overexposing it by a stop (and therefore losing highlight latitude).
 
The main thing if using a light meter combined with the camera really is not to worry that the ISO's match but rather that knowing a certain amount of light as indicated on your meter will be what you expect it to be in terms of exposure or levels on the camera.

If you have to set your meter to 1000 ISO in order to get what would be a middle grey card at 32IRE when the camera is set to 2000 ISO, then really that's all that matters. Where the difference comes from can be a number or things.

And as you stated, its really important to understand the difference between actually "rating the camera differently" versus exposing for the cameras native base ISO. The former will result in losing your latitude.
 
Since the debate seems to continue on this matter (and for the FS7 too, which supposedly shares the same sensor), I've decided to post test findings up on my blog. It's a pretty comprehensive breakdown, so hopefully will help clear things up for people, you can see the results here: http://dreamsmiths.com.au/blog/
 
How much do you trust your 18% target? I tested multiple "18%" cards side by side and discrepancies between official brands were over 1 stop!
 
Given that the white markers on the card (and the whiteboard too, if a little messily) ring true at 59% IRE on the waveform (the level SLOG2 says they should be at). I trust it completely.

The sensor (well mine at least) is approximately 1125 ISO. Exposed accordingly it behaves exactly as you'd expect, whether you're putting things two stops under or two stops over.
 
Mark I had the same conclusions as you on my F55, but I want to go back and redo the tests at some point because some grey cards are 12% not 18% and some light meters (which is the method I used) are calibrated to 12% not 18%, so a stop here or there is easy to be off by.

Mostly what I do before shooting is look at where something white falls on a waveform monitor, and something mid-tone usually around grey. With that I can take meter reads and change the ISO of the meter so that my f-stops match and from there on in as I change setups or tweak setups "ISO" really becomes irrelevant.

My long term suspicion of the F series ISO rating is that they did it without the shutter turned on. But again it's subject to further testing, even different glass will give different results so this mystery 1 stop is really nothing I obsess over anymore.
 
Grug said:
Since the debate seems to continue on this matter (and for the FS7 too, which supposedly shares the same sensor), I've decided to post test findings up on my blog. It's a pretty comprehensive breakdown, so hopefully will help clear things up for people, you can see the results here: http://dreamsmiths.com.au/blog/
Click to expand...

The test does not say much about camera's native sensitivity, only about sensitivity when using this particular gamma (and even that is open to interpretation because Sony's values are just a recommendation). It only applies to S-Log 2, coincidently to S-Log 3, possibly to some other gammas too but it is not certain (in fact, if in your writeup you say that you can't speak for the hypergammas, then you understand that specific gammas affects how you'd rate camera's ISO). The test does prove that when shooting S-Log 2 it is recommended to rate it at 1125 ISO in order to hit the Sonys IRE numbers. My own findings are close to it (ISO 1250 for S-Log 3, BTW I usually rate the camera at 1000 ISO). BUT, S-Log 2 is just one of the curves, remember it is not a linear RAW signal. Your findings don't talk about camera's native sensitivity. I'm sure it is completely possible to create a custom curve which will yield greater sensitivity in the middle of the curve (imagine a curve with a "bulge" around mid point), hence it will be possible to achieve higher IRE numbers for the mid point. If you based your measurements on such gamma's mid point , would you all in a sudden say that the native sensitivity of the camera is greater? No, because this is specific only to this particular gamma and the real native sensitivity of the camera is the same no matter what curve you use, not to mention that such increase of sensitivity in mids does not say much about what happens with shadows and highlights. Camera's ISO rating for specific gamma and native sensitivity are two different things. Sure, in practice, especially if you shoot S-Log primarily, who cares what the "real" native sensitivity is vs. the one that dictates how much light we need, the practical one...
 
Last edited:
broughtonfilm said:
The test does not say much about camera's native sensitivity, only about sensitivity when using this particular gamma (and even that is open to interpretation because Sony's values are just a recommendation). It only applies to S-Log 2, coincidently to S-Log 3, possibly to some other gammas too but it is not certain (in fact, if in your writeup you say that you can't speak for the hypergammas, then you understand that specific gammas affects how you'd rate camera's ISO). The test does prove that when shooting S-Log 2 it is recommended to rate it at 1125 ISO in order to hit the Sonys IRE numbers. My own findings are close to it (ISO 1250 for S-Log 3, BTW I usually rate the camera at 1000 ISO). BUT, S-Log 2 is just one of the curves, remember it is not a linear RAW signal. Your findings don't talk about camera's native sensitivity. I'm sure it is completely possible to create a custom curve which will yield greater sensitivity in the middle of the curve (imagine a curve with a "bulge" around mid point), hence it will be possible to achieve higher IRE numbers for the mid point. If you based your measurements on such gamma's mid point , would you all in a sudden say that the native sensitivity of the camera is greater? No, because this is specific only to this particular gamma and the real native sensitivity of the camera is the same no matter what curve you use, not to mention that such increase of sensitivity in mids does not say much about what happens with shadows and highlights. Camera's ISO rating for specific gamma and native sensitivity are two different things. Sure, in practice, especially if you shoot S-Log primarily, who cares what the "real" native sensitivity is vs. the one that dictates how much light we need, the practical one...
Click to expand...

Every gamma curve has a known IRE mapping for middle grey and 90% white reflectance including slog2/3 and the standard and hypergammas.

What you're almost suggesting is that different gammas have different ISO's.
 
starcentral said:
Every gamma curve has a known IRE mapping for middle grey and 90% white reflectance including slog2/3 and the standard and hypergammas.

What you're almost suggesting is that different gammas have different ISO's.
Click to expand...

Yes, you are right, Dennis! My post is almost entirely wrong! I still ask myself if exposing based on those mapped values means exposing at the camera's native ISO or are they recommendations which are purposely offset by Sony somewhat arbitrarily to give best results (all mapped values offset uniformly for all gammas, not individually per each gamma as I wrongly suggested before). In other words, for example, even though the theoretical native sensitivity is in fact 2000 ISO, is it possible that the recommended values are meant to influence us to shoot at lower ISO (approx. 1125 - 1250 )? Therefore using those values to determine camera's native ISO would not give definite answer. I guess that's where I was going and got lost...
 
starcentral said:
My long term suspicion of the F series ISO rating is that they did it without the shutter turned on. But again it's subject to further testing, even different glass will give different results so this mystery 1 stop is really nothing I obsess over anymore.
Click to expand...

That would be only possible if they in fact had the shutter turned off but forgot to reflect that in the shutter speed setting in the light meter (or whatever device used as a reference). If you measure correctly then it does not matter what the shutter angle / speed is, it leads to the same conclusion. Based on how many bugs their cameras have I can totally believe they simply made such error :)
 
Last edited:
Okay, I'll admit right from the start that I have not bothered to read this whole thread but the title of it got me wondering whether the displayed ISO is correct or not for S-LOG. I already had my charts and scopes out for something else so I decided to check my F55 and FS7. Sorry, I don't have an F5. My testing was based on the assumption that a 90% white target with S-LOG2 should be exposed at 59%.

I used a Cine-Xenar 95mm prime, a Leader LV5330 waveform monitor, a dimmable Litepanels Fresnel, the white side of a genuine Kodak gray card, and my Sekonic L-758C lighter meter to check the sensitivity at several different f/stops, here's my results:

F55: Displayed ISO = 1250 Actual ISO = 1250
FS7: Displayed ISO = 2000 Actual ISO = 1600

I am absolutely certain that these numbers are correct for my cameras.
 
You must log in or register to reply here.
Back
Top