How do you guys rate your camera's sensitivity?

[Will Edwick]

I ask because I recently tested the Blackmagic Ursa and my (possibly inaccurate) test revealed that the camera was shooting at around 50 ISO when set to 200 ISO. I know cameras often shoot slightly over or under what they say they do but this seemed a little too much and is making me think that my method of testing is completely wrong.

 

Essentially, I set up the camera about 2 metres away from my grey card and put on an 85mm so that the card filled the majority of the frame. I then lit the card to a flat stop and took a reading on my meter at 200 ISO and set my stop accordingly. Once in Resolve, I pulled up the waveform and the horizontal line of my grey card fell well below the mid-grey threshold. I re-did this test until my grey card was falling on the mid-grey line in the waveform, and by that point my meter was reading about 50 ISO.

 

For the record my lenses measure in T-stops, I wasn't using anything like a speedbooster and all the settings on my meter were the same as the camera.

 

To me this seems like a sensible way to test that sort of thing but I could be totally wrong and if I am, does anyone have a better/more accurate way of testing this?

[Adrian Sierkowski]

Were you in Film mode or video mode.

Are you sure your meter is properly calibrated.

Are you sure you're on a grey card

are you sure your lens is in T stops

 

There's too many variables to figure out-- but it's certainly an error in there somewhere-- not to mention the URSA 4K is rated, if memory serves, as a 400ASA camera, native.

 

Also RAW or ProRes?

 

 

As for rating a camera-- personally I go with manufacturer's suggested-- unless when i test it out that just seems wrong. Normally I will shoot over under tests from the base ISO with some kind of skin tone reference etc, macbeth chart, grey and white card, and then correct them all back to "normal." I'll look at this quick correction and whichever one looks best to my own eye is what i'll tend to err towards.

If, however, there is some funky stuff happening with post or production design then i'd start getting into more specific testing.

 

For myself, also I never change the ISO on the camera-- I keep it native, and if that is 800, then it means a fair amount of ND for day exts and even Ints. Else if that happens to be 200 that means pumping in more light to your scenes (and maybe trying to figure out who get you a digital bolex in the first place)

[Brian Drysdale]

It's usally a balance between highlight clippimg and the noise level you find acceptable.

[David Mullen ASC]

What format did you record to? Because if log, or converted raw to log, then middle grey can fall lower than if converting to Rec.709, though not two stops lower.

 

You should try shooting an 11-step grey scale so you can see where the white, black, and grey chips fall on a waveform.

 

Often you'll find that when looking at log on a waveform that the white chip is maybe two stops lower in IRE value than in Rec.709. This is because with log you are capturing more than 10 or 11 stops of dynamic range, so there are a few more stops of overexposure detail above what a white chip gives you on a chart.

[Adrian Sierkowski]

Also-- how did you meter? If you were in spot mode-- loose the grey card for a moment and try it in incident mode. If you were in Incident, get a spot reading off of the card.

[Dmitry Savinov_38080]

I prefer to rate at preset speed 800 for Alexa,even bright exteriors,to achieve plus or minus all the highlights and shadows,if you change your speed,you loose in one or other.

When shooting Red (rarely,to be honest),I also use the preset speed,remembering,that its 800 is 320 in general)))

[Will Edwick]

I shouldn't really have used the term 'rating' because I'm not looking to find the sweet spot of the sensor, I'm more trying to find out how to get mid-grey at mid-grey so that my LUT isn't splitting the image at 1/1.5 stops over if that makes sense?

 

Adrian - I was shooting ProRes in 'Film' mode. I'm pretty sure my meter is calibrated but if you could explain how to test that, it would be much appreciated. Yes I'm sure I was using a grey card and yes I'm sure my lenses measure in T-stops. I also took spot and incident readings which gave me the same stop.

 

David - I was shooting Log. If mid grey can fall lower in log than in Rec.709, would you recommend compensating for that with my meter if I don't want to convert to Rec.709 before putting on my LUT? It seems annoying that middle grey can fall anywhere other than middle grey. Also, what's the best budget 11-step grey-scale chart?

 

Cheers for the help guys!

[David Mullen ASC]

Watch this:

http://www.learningvideo.com/bmpcc-exposure-tutorial/

 

Read this:

http://www.rogerdeakins.com/forum2/viewtopic.php?f=22&t=3503

 

One of the points of log is extended highlight range like film negative has, so if you expose the image "to the right" as some people suggest to minimize noise, or to make whites look white and medium grey look medium grey, you will be sacrificing that extended highlight range. Log isn't meant to look normal on a Rec.709 gamma monitor.

 

So if you want to get the extra highlight range from recording log, then do what the video says, set the exposure in Rec.709 video mode, where your meter should match, and then switch to recording log where middle grey will drop down to 40 IRE or so. But of you are worried about noise then you'll want to open up. You could, for example, set the camera to 800 ISO but set your meter to 500 ISO, for example, but you may have to correct any log-to-red.709 dailies to look correct in terms of brightness.

[Will Edwick]

So as I understand it, different camera manufacturers rate their ISO's to varying IRE values and two different cameras at the same ISO can be quite different in terms of exposure when shooting Log? So to combat this, it seems like I should compensate with my meter, so that I can stick on my LUT without converting the footage to Rec.709, and get a similar exposure between multiple cameras?

[David Mullen ASC]

What do you mean "stick on the LUT"?  Isn't the LUT to convert the log image for viewing on a Rec.709 monitor?

 

Yes, different camera manufacturers have designed different log formats, partly because the cameras have different dynamic ranges that they have to cram into 8-bits, 10-bits, 12-bits, etc.  The cameras with less dynamic range recording at a lower bit depth tend to have very mild log versions, just a flatter, slightly darker version of their Rec.709 version -- while cameras with very wide dynamic ranges recording at higher bit depths can have flatter log curves where more information is in the middle and lower to hold more stops of overexposure.

 

And you may find through testing that you prefer to expose different than what they recommend.

 

Exposure is playing a game between noise vs. headroom/clipping.  Some are more concerned with one than the other.

[Will Edwick]

I'm not sure, it's a film emulation LUT that I like the look of and I'm just wanting to make sure that when I use it, it's not making mid grey darker. Or anything over mid grey for that matter

 

Thanks a lot for the help!

[David Mullen ASC]

Yes, but a film emulation LUT from what to what -- log to log, and then you need a LUT for log to Rec.709 gamma for viewing? If this was made for finishing in Rec.709, it probably is doing both a log to Rec.709 conversion but tweaked to resemble some film stock viewed in Rec.709 gamma like from a telecine transfer.

 

If it is a film emulation LUT made for D.I. work, then it would keep the gamma in a log format and you'd need a second LUT to view the image correctly on a Rec.709 monitor.

 

Or is it a LUT designed for Rec.709 in and out?

[Tyler Purcell]

The DaVinci film emulators really screw with things. I've been experimenting with them as well and honestly, you've gotta shoot in RAW to make them work. For some reason the REC709 Pro Res "Film" dynamic range, looks like crap when you use those emulators. The moment you give it RAW material, it comes to life and you can really make them work well.

 

In terms of measuring your cameras sensitivity, I've used a standard SMPTE television gray scale chart. This makes it a lot easier to see your camera's complete dynamic range and get a far more detailed look on the scopes. I've used the chart in light-controlled situations and also outside with direct sunlight. I've found the Blackmagic cameras have a lot of problems at lower ASA's with dynamic range. As Adrian pointed out, the cameras natural sensitivity rating delivers the most dynamic range and everything on the charts just line up in the scopes perfectly @ 800ASA. However, the lower you put the ASA, the more issues you have with clipping. The only solution I've found is to underexpose everything by at least a stop if you can and re-work the image in DaVinci later. 200ASA has marketably less grain/noise and delivers a much smoother image then 800 ASA.

 

Having shot on film and video for the decades prior to this Digital Cinema revolution, I have a general idea of what sensitivity, shutter speed and stop will work for given situation. Add a decent light meter to the party and you're gonna get spot on almost every time. I've found the Blackmagic cameras to be pretty accurate for their given ASA. Even when you crank up the shutter speed (lower shutter angle), at least the pocket camera is incapable of shooting in bright sunlight without an ND filter @ 200 ASA. In those very same situations, I'd be shooting 50 ASA on film and running the camera almost all the way closed, no less then F8. I recently did a 1600ASA test shoot using some christmas tree lights and it worked really well, I was really shocked how sensitive that little sensor is and the noise level wasn't horrid. So even the ultra-cheap cameras like the pocket, do a great job at delivering a decent image with little to no light.

[David Mullen ASC]

What's nice about getting an 11-step chip chart that has two rows of grey patches going in opposite directions in density is that on a waveform it forms an X shape with middle grey at the crossover point. The X will also show you any gamma curves or knee/shadow compressions being applied, and makes clear how much the signal is split between highlights and shadows.

[Will Edwick]

Yeah I'm pretty sure it's Log to Rec.709 then making that look more filmic, I think I'm understanding this a lot better now and I'm going to purchase an 11 step chart to help with accuracy in the future.

 

Thanks a lot Dave!

[John E Clark]

I ask because I recently tested the Blackmagic Ursa and my (possibly inaccurate) test revealed that the camera was shooting at around 50 ISO when set to 200 ISO. I know cameras often shoot slightly over or under what they say they do but this seemed a little too much and is making me think that my method of testing is completely wrong.

 

What it sounds like you are asking is... "How do I develop a personal Exposure Index for a given camera".

 

With film one had one ISO value recommended by the manufacturer, and that was based on certain densities resulting from given exposure values and processing procedure.

 

One could use a densitometer to measure the film that one had in hand, and derive a 'personal' ASA/ISO/Exposure index.

 

This is my procedure for the digital world, which pretty much follows what I did in the ancient past for my stills...

 

I have a chart with a grey step wedge (and it also has colors) but for the basic ISO rating using a 18% grey card, using 'RAW' mode, I do the following...

 

Set the exposure given by my light meter at the ISO setting for the camera. So if the camera has a 400 or 800 setting, I set the meter to that value.

 

Shot a short clip... 10 seconds or so.

 

Then I open up the lens one stop, and take another short clip, and I to that for several stops over and under the camera setting value.

 

I then import those clips, and using Resolve these days, look at were the 18% grey card value 'falls' on the waveform. For most camera, shooting in some sort of 'log' mode, 38-39-40 % IRE seems to be a 'favored' values. I find the clip that yields a waveform in that range.

 

So, if with a camera setting of 400, I find that I must open up 'one stop' then in reality given my lens, camera, meter, I really should be setting the meter at 200 despite the camera setting of 400. Some meters have a compensation that sort of 'hides' this offset 'trick'. Why would I come up with a ISO/EI value so 'whacked' from the camera setting... because it may be that I have a lens that isn't really calibrated, and so it could have some error, I have a meter that isn't really calibrated, so there could be error, etc.

 

But even with those 'absolute' errors, once I determine that for my camera/lens/meter, this ISO value yields the desired IRE value, it doesn't matter that those elements are not NBS traceable, this personal Exposure Index works for my set of equipment.

 

This of course requires that one test all one's lenses... and lenses that are calibrated with T-stops would allow one to make one determination, and not 'worry' when changing lenses...

 

With the step wedge, I can then see how many stops 'over' and 'under' that middle grey value I have for highlights or shadow detail.

 

Some manufacturers place the 'natural' ISO value for a camera when there are equal number of stops over and under from that middle grey. There are several options for determining the digital ISO for a camera, and so, it is somewhat 'unknown' exactly what criteria a manufacturer may have used in publishing their specs.

 

With the step wedge and color chart you can also see if the ISO setting you determine from the above, 'works' for color saturation or 'look'...

 

Where Rec 709 or other display spec comes into play, is after you find what the camera is capable of, you then have to 'fit' the camera's image into whatever curve the output media for display will be.

 

In the Film film days, this would be the paper print for stills, or the projection print for motion pictures. In any case, the output display can require another round of analysis to get the 'optimum' displayed image that the viewers will see.

Edited October 5, 2015 by John E Clark

[Will Edwick]

Great, cheers John! That was exactly what I was looking for.

 

Thanks again!

[Tom Yanowitz]

So a couple of things.

- An image sensor doesn't have a base sensitivity. (The only "real parameters" it has are the size/surface of its photosites and its bitrate)

- Middle grey is more or less randomly picked by the manufacturers and is pretty much useless in the digital world.

- Placing your optic fstop according to the incident reading is getting old (unless you really don't have any color grading planned and shoot 709, which can be considered as in-camera grading)

- I don't think "ISO" (as the group of people in charge) has done a thing yet for digital.

 

Some say the Alexa native ISO is 800, some say 200, some say 640 some say 1000.

Everybody's wrong, it would be too easy a thing to have a magic number that gives a better image thant the other.

But for every digital camera, if you expose the old way (incident reading, caucasian skins at +1 etc etc..) then the lower the ISO the cleaner the image.

Why ? because you allow more light to hit the sensor. The more light the better.

That said, the more you expose, the closer your highlights get to the clipping point of the sensor, so you have to check for blow out highlights more. It's a trade off, low and high ISO both have their good and bad.

So there's no one definitive answer, that's why it's better to choose your ISO/EI almost case by case depending on the scene, and it's more interesting that way (compared to lazily sticking 24/7 with what the manufacturer told you was good).

 

 

I hope I don't offend any of the great DP's out there but when's something is not well understood you might as well point it out.

Edited October 7, 2015 by Tom Yanowitz

[Tom Yanowitz]

One quick way to picturing the "trade off" is shooting an actor with some pratical lights in an overall very dark room.

Let's say you can compensate the fstop to get a constant +1 reading on the skin tones as you change the ISO.

ISO 200 : you'll get details in the shadows but the practcals will be burnt.

ISO 800 (for example) : you'll get less details in the shadows, but the practicals will have details.

The skin tones will be cleaner at 200 (not sure we can spot the difference though).

 

(I couldnt seem to edit my previous post anymore)

Edited October 7, 2015 by Tom Yanowitz

[John E Clark]

- I don't think "ISO" (as the group of people in charge) has done a thing yet for digital.

 

 

There is an ISO spec for digital still cameras, and since a 'motion picture' digital camera is actually a 'still' camera with the ability to record multiple stills at a given frame rate... by extension the spec can be used for determining a digital motion picture camera's ISO.

 

----

ISO 12232:2006 specifies the method for assigning and reporting ISO speed ratings, ISO speed latitude ratings, standard output sensitivity values, and recommended exposure index values, for digital still cameras. ISO 12232:2006 is applicable to both monochrome and colour digital still cameras.

----

 

The problem arises in that most digital cameras allow for a selection of the ISO setting, and as such behaves more like analog gain of video cameras, rather than the fixed ISO values for Film stock.

 

With that in mind, the use of the term 'native ISO' is the manufacturer's attempt at saying this camera will respond to exposure in a certain way that we (marketing, engineering, fiat...) have determined to be 'optimal/best'.

 

One way to state that is in terms of ISO, given a 'middle grey', the 'optimal/best' is to have as many stops above Middle Grey as below, in terms of recorded differentiation. So a camera may have ISO 800 as that value, and when the step wedge is 'shot' and displayed on a waveform display, one sees distinct steps, perhaps 5-6 steps above, and 5-6 steps below.

Film speed used densities, and was defined in terms of the 'base fog' due to developing, and an exposure that would yield a .1 log density on the film.

 

The second important characteristic of Film was the gamma of the characteristic curve. Development times could alter the curve so one could increase or decrease contrast to a limited degree.

 

In Digital systems using a threshold above the 'sensor black' noise, may be one way to define 'speed', but it doesn't work as well.

 

In any case, even in the olden days, photographers, and I presume motion picture photographers as well, would determine their own 'Exposure Index' for a given film and would use that for meter readings, etc.

 

In that regard I rated Tri-X at ASA/ISO 200, which Kodak gave a ASA/ISO rating of 400. I also cut development based on determining the density for 'white' as printed on the papers I was using...

 

For those who want to go into the way back machine... here's the Kodak brochure on various topics related to determining 'film speed' and 'characteristic curves'...

 

http://motion.kodak.com/motion/uploadedFiles/US_plugins_acrobat_en_motion_newsletters_filmEss_06_Characteristics_of_Film.pdf

Edited October 7, 2015 by John E Clark

[Albion Hockney]

it should be noted that the stops lost in the shadows from higher ISO...those stops are not going away...its just that now you are exposing at say 1600ISO on a camera that is 0DB at 400asa... and therefore you are underexposing the image and introducing more noise so those stops just get buried. but you gain highlight detail because the sensor itself is not clipping and you are just pushing the signal up after it has passed the sensor.

[Mark Kenfield]

I've always just gone off the manufacturers' stated values for middle grey, and used a waveform in conjunction with my meter to rate the sensor accordingly.

 

Over and underexposure tests then tell me how much over and underexposure headroom I have, and I expose accordingly.

 

Personally I find it much more reassuring lighting and exposing from my meter, because so long as I've done my tests, it takes all of the guesswork out of it.