Richard Butler's Sense and Sensitivity Parts 1 & 2

[girod]

Hello Keith,

What is your take on Richard Butler's articles in dpreview: 1) Sense and Sensitivity Part 1 and 2) ISO and Sensitivity Part 2 (Behind the Scenes: Extended Highlights)?

Thanks,
jaime

[keithsnell]

Hello Keith,

What is your take on Richard Butler's articles in dpreview: 1) Sense and Sensitivity Part 1 and 2) ISO and Sensitivity Part 2 (Behind the Scenes: Extended Highlights)?

Thanks,
jaime

Hi Jaime,

Thank you for asking.  I think Richard makes a good attempt at explaining a very complex subject in a way that can be understood by the "average" DPreview reader.  When attempting something like this, it is always difficult to "simplify" a complex subject without becoming inaccurate.  

I believe there are a couple of areas where Richard "oversimplified," and therefore the conclusions that the average reader draws from his explanations will be slightly inaccurate. For example, he implies that Nikon is using the "Standard Output Specification" definition of ISO in establishing the sensitivity of the sensors in their new generation of cameras, and therefore middle gray should be exposed properly for an in-camera JPEG.  Those of us that have examined the data would say that isn't quite true.  Middle gray is actually rendered as 1/2 "overexposed" in the output from these cameras, based on the tone curve being applied.  This goes a long way toward explaining many of the "my D700 overexposes" threads on DPReview.

In the end, I think Richard's article is more about explaining the rationale for why DPReview describes their test results in relation to ISO the way they do, rather than an attempt to help readers understand the truly complex world of ISO designations on their cameras.  His primary point is that DPReview uses a consistent and logical (at least to them) method of incorporating ISO into their testing.  It is however, an oversimplification.  

It is good that Richard made the attempt.  It means the folks at DPReview are making a conscious effort to understand this complex subject, and the eventual result is that they will help us all understand it better.  I suspect that as their test results don't quite match their "simplified" explanation of ISO, they will have to evolve their explanation.

I choose to understand the subject in this way:  Camera makers use ISO to define a standard interface between subsystems in a camera.  The first "interface" is between the camera's exposure system and the sensor or film, and is used to define how much light is delivered to the film plane for a specific EV reading by the meter.  The camera's metering system senses the light intensity, then based on the ISO or stated sensitivity of the sensor or film, adjusts the aperture and shutter speed to deliver the required amount of light to the film/sensor plane for the stated sensitivity of the recording medium.  That's the first part of the equation, and is rather straightforward.  You will find that every Nikon camera is consistent in the amount of light it delivers to the film plane based on a stated ISO (within engineering tolerances).  In other words, if you were to take a Nikon F5 film camera from 1996 and measure the light delivered to the film plane at a specific ISO,and compare that to your D700, the result would be exactly the same.  Spot metering the same subject would result in the same aperture and shutter speed being selected for the same ISO on both cameras.

This begs the question of how the "stated" ISO is established for the sensor.  The ISO Standard specifies five different ways manufacturers can define the "sensitivity" of their sensor.    One way of establishing this sensitivity is based on the rendering of mid-tone gray, as specified in the Standard Output Specification (SOS) defined in the ISO standards.  Another way is "saturation based" which defines the sensitivity of the sensor based on its saturation point.  (Many speculate that the ISO of the sensor in the D3 and D700 is based on the "saturation based" standard.)  There are two methods that define the ISO based on the noise floor of the sensor and processing circuits, while a fifth method gives the manufacturers the ability to define a "recommended exposure index" which is essentially a way of defining the "best" sensitivity rating, taking into account the sensors saturation point, noise characteristics, and color response across the range of brightness the sensor is able to record.  

The SOS specification is the simplest, and appears to be the specification used in earlier generations of Nikon cameras.  However, SOS does not doesn't give the manufacturers as much latitude as the other methods to optimize the output based on sensor characteristics.  The most recent generation of cameras (Nikon, Canon and Sony) do not appear to be using the SOS specification to define the "sensitivity" of the sensor.  The ISO standard implies that manufacturers should use the method of establishing the sensitivity of the sensor that best reflects the "limiting factors" of that particular sensor design.  In other words, they shouldn't use the "saturation based" method if that method results in unacceptable noise in the image, and they shouldn't use the "noise based" method if that results in unacceptable saturation of the sensor.  If a sensor has a very large dynamic range, manufactures can use the "noise based" method of defining its sensitivity, thereby providing extended highlight headroom (like the 5D II).  Nikon appears to be be using the "saturation based" method of defining the "sensitivity" of the sensor in the D3 and D700, probably because of the "unique" saturation characteristics of the green channel in those cameras.  By specifically defining where this saturation point is, and taking this into consideration during tone mapping and rendering, the manufacturer can mitigate any issues (banding) seen at the top end of the tone curve, while still optimizing the noise characteristics of the sensor output.  It doesn't appear (on the surface) that Nikon is using the SOS method of determining sensitivity, since a mid-tone is actually rendered at +0.5 EV with the standard Nikon tone curve.  (There is some anecdotal evidence that they could still be applying the SOS method, but using the "brightness" adjustment introduced with the D3 and D700 to link the standard to the rendered tone curve; however, the bottom line is that the new generation of Nikon cameras renders mid-tone +0.5 EV "hotter" than middle gray when the manufactures recommended rendering intent is followed.)

The second area where ISO is used to define a standard interface is between the "raw" data recorded by the sensor and how that data is rendered.  When the data is "handed off" to either the in-camera processing or the raw processing software, it is tagged with an ISO value, which defines how the raw values will be mapped to brightness levels in the rendered output.  Without this ISO "tag" all images shot at higher ISOs would be rendered too dark.  (This is an oversimplification, since higher ISOs are in reality handled by both in-camera amplification of the signal, as well as amplification in the "tone mapping," however for our purposes it is OK to ignore the signal amplification.  The raw processing software has the ability to match the "tagged" ISO value to a defined rendering intent.)  

Part of this rendering is an attempt to map the linear data from the sensor in a way that matches our visual perception.  This is where the "tone curves" come into play, and where much of the confusion is introduced with respect to ISO vs "rendering."  The reality is that the linear data from the sensor is a very poor match for our visual perception; therefore, if they want pleasing out-of-camera results, manufacturers must manipulate the data in a way to best match our perception.  One complaint with earlier generation Nikon dSLRs was that raw images were too "flat" and required post processing in order to "look good."  Nikon addressed this in several ways, first by adopting a "punchier" more contrasty "standard" tone curve that has a more "film like" roll off of the highlights, and by boosting the mid-tones and shadows in order to better match our "perception."  The end result is that the tone curve is much more aggressive than most films ever were, and our ideas of "proper" placement of mid-tone should probably be adjusted accordingly.  The bottom line is that in most situations the aggressive tone curves provide pleasing out-of-camera results.  The notable exception is portrait photography, where the aggressive tone curve resulted in "blown" highlights in the skin when using the traditional exposure settings, which is why much of the "my D700 is overexposing" posts were initiated by photographers after attempting to shoot portraits with the "standard" tone curves.  That is also why Nikon responded by making available the "D2X" tone curves, which were much less aggressive, and could be loaded into your D3 or processing software to provide a better match to the "traditional" rendering of mid-tone in the output.

So, to summarize, camera makers use ISO to define a standard interface between subsystems in a camera.  The first defined interface determines how much light the camera's exposure system delivers to the film/sensor plane for a stated ISO.  The implementation of this "interface" has remained consistent since the 1960s, and it is this "standard" that enables you to use interchangeable lenses and external strobes with your camera.  The second interface is between the raw data and the "rendering intent" of the data.  This interface  has evolved, and based on the ISO method chosen, can result in mid-tones that are rendered brighter than "middle gray."

In the "old days" of film, pro photographers would test each new film on the market before using it for a critical project in order to determine if they needed to adjust the "rated" ISO of the film to produce the results they were after.  Most photographers that used Fujifilm's Velvia adjusted the ISO from a rating of 50 to a setting of 40 on their cameras, in order to produce a "punchier" result that would sell better to magazines, who were looking for eye-catching images that would "sell" the magazine.  In retrospect, this "derating" of the ISO was also "pushing up" the mid-tones and shadows, much like tone curves in the latest generation of digital cameras are doing today.  The old advice to "know your film" still applies.  Except now instead of "derating" the ISO on your slide film, you might choose to use exposure compensation to get the result you want, given the characteristics of your recording medium (and rendering software).

This was probably a much longer response than you were looking for, but I hope it helps your understanding of ISO, and the role it plays in our photography.

Keith

[girod]

Thank you very much Keith. As always, you can taylor your explanation in such a way that your targeted student(s) will understand even the most complex concept in photography and most importantly, be able to implement it.

Presently, I use ISO on my D700 the following ways:
1) ISO 200-800 is the cleanest and widest DR, so If I have to go above 800 I have to pre-conditioned my mind that I will have noise in my images; I have never tried below 200.

2) In handheld, I prefer to use higher ISO (upto 800) to have a shutter speed not lower than 1/60s.

3) When using the SB-900 (mostly TTL-BL-FP), I use ISO 800; manual exposure mode; rear curtain flash mode.

4) In outdoor actions with varying lighting every minute, I use autoISO (200-800); manual exposure mode - I adjust the aperture according to my DOF taste for that day and the shutter speed to freeze the moment; spotmeter (skin, greyish-equivalent shirts) mostly to 0.0EV, if there's a white shirt/spot that I can follow consistently - I'll spot meter to +2.0EV

Am I doing it right? What or how else should I use the ISO of my D700 to get the best IQ that it can deliver?

jaime

[keithsnell]

Presently, I use ISO on my D700 the following ways:
1) ISO 200-800 is the cleanest and widest DR, so If I have to go above 800 I have to pre-conditioned my mind that I will have noise in my images; I have never tried below 200.

2) In handheld, I prefer to use higher ISO (upto 800) to have a shutter speed not lower than 1/60s.

3) When using the SB-900 (mostly TTL-BL-FP), I use ISO 800; manual exposure mode; rear curtain flash mode.

4) In outdoor actions with varying lighting every minute, I use autoISO (200-800); manual exposure mode - I adjust the aperture according to my DOF taste for that day and the shutter speed to freeze the moment; spotmeter (skin, greyish-equivalent shirts) mostly to 0.0EV, if there's a white shirt/spot that I can follow consistently - I'll spot meter to +2.0EV

Am I doing it right? What or how else should I use the ISO of my D700 to get the best IQ that it can deliver?

jaime

Hi Jaime,

The ISO guidelines you have adapted are generally right on.  Here's how I might "tweak" them slightly:

1)  You are correct that ISO 200-800 provides the cleanest images, with the widest DR.  I have found that ISO 1600 provides perfectly acceptable images as well.  If I "pixel peep" I can find noise in the image, but nothing that is objectionable.  ISO 1600 often provides the margin I need for hand-held shots in low light.  I find myself using it often in those conditions.

2)  See #1 above.  I think you would find ISO 1600 acceptable as well in "marginal" light conditions.

3)  First, you should recognize that FP mode significantly reduces the power of your flash.  I often find that I need more power when attempting to fill flash in bright high-contrast (daylight) conditions.  Therefore I will use ISO 200 in order to get the shutter speed down to 1/250th or slower, which will allow me to use the full power of the flash.  The "sunny sixteen" (http://community.spiritofphotography.com/index.php?topic=37.msg137#msg137) guidelines for exposure (general guidelines for exposure that have been around since the early years of photography when many cameras did not have internal exposure meters) tell me that at ISO 200, I will need to use a shutter speed of 1/200th of a second (the inverse of the ISO) at an aperture of f16 in order to get "proper" exposures in bright sunlight.  If I were to use a higher ISO, my shutter speeds would need to be faster than 1/250th (in order to control the ambient exposure), which would put me in the "FP" range above 1/250th of a second where my flash power is significantly reduced, and I might not get enough flash on the subject to adequately fill the shadows.  For example, at an ISO of 200, aperture of f16, and lens zoom position of 35mm, and shutter speed of 1/250th, my flash has a range of 11 ft (at full power).  If I bump my shutter speed up to 1/320 of a second and use FP mode on the flash, my flash range is reduced to 4.6 ft!  The minimum range I would be able to get effective fill (-2.0 EV from full power on the flash) using this FP setting would be 9.3 ft.  Because the flash power in FP mode is limited by the length of time the shutter is open, and increasing the ISO will require a corresponding increase in shutter speed to balance ambient and fill, this essentially defines my limit for the range of my fill flash in sunny conditions using FP mode.  At a range of 9.3 ft from the subject, I will only be able to get light on the subject equal to -2.0 EV of full power, regardless of how much I bump up the ISO.  This is just enough power to barely begin filling in the shadows.  In contrast, I can get a good fill ratio of -1.0 EV from full flash power on the subject at a range of 15 ft if I stay at or below a synch speed of 1/250th of a second.  (To summarize, Although FP mode can be nice to have in some situations, I generally avoid it, since I find I can better optimize my flash power by staying at or below a sync speed of 1/250th.)

This might be a good time to talk about how you can estimate your flash power based on ISO and aperture.  I can calculate the approximate range of my flash (SB-800) by dividing the guide number of the flash by the aperture and adjusting for ISO.  The guide number for the SB-800 is 174 at an ISO of 200 and a flash zoom position of 35mm.  (You can find the guide number for your SB-900 in the specifications section of the manual.)  To calculate my range at these settings, I divide the guide number by the aperture, so 174/16 = 10.9 ft (shown as 11 ft on the flash LCD).  If I'm forced to use an aperture of f22 to reduce the ambient exposure, then my flash range is reduced by a corresponding amount, i.e., 174/22 = 8 ft.  Zooming the flash head will increase these ranges slightly.  For example zooming the flash from 35mm to 70mm will increase my range from 8 ft to 10 ft (at an aperture of f22 and ISO 200).  The guide number gives you a way to easily estimate what flash power you will be able to put on the subject given a specific aperture and ISO.  It is important to note that doubling the ISO (from 200 to 400) will increase the flash range by 1.4x, and that quadrupling the ISO (from 200 to 800) will double the flash range.  

I know I have covered several different concepts rather quickly in this answer, so if you have any questions, please don't hesitate to ask.

4)  You should realize that if you use auto ISO and spot meter, the camera will adjust the ISO to achieve an exposure of 0 EV on the area under the spot meter (assuming you don't have exposure compensation set).  So, if you meter and recompose, the camera will adjust the exposure (ISO) based on what it is now metering with the "spot."  I'm not sure the method you are using in #4 is the most effective method for varying light conditions and/or moving subjects, so let's talk about those techniques a bit in a follow-on post.

Hope this helps!

Keith

[girod]

I will then remove my fear and use ISO1600 more sparingly.

I have yet to really get a good handle/understanding of the flash concepts, in general and how to implement them specifically to the D700+SB900. Hence even though I understand your initial explanation, I really don't until I can implement them properly and comfortably.....so, I don't even know what to ask but at the same time I can feel the huge vacuum in me as to the proper use of flash. Right now my baseline is Mike Hagen's "The Nikon Crative Lighting System". Since I know close to nothing, I set my D700 to manual exposure at f/5.6-8 1/60s ISO 400-800, matrix-metering mainly to get iTTL-BL on the SB900. I thought that by setting the D700 to 1/320s FP, the FP function will only be activated if I use 1/320s and above on the D700.

I would very much appreciate it also if you could teach me the best method for varying light conditions and or moving subjects. I know that  the D700 will automatically expose for 0.0EV with a setting of varying ISO but fix aperture and shutter speed - so I try to spot-meter on middle gray equivalents (skin, green or near grey shirts) or on whites (I dial an EC of +2EV) or blacks (EC of -2EV). For focusing, I use the AF-ON button; AF-C, release priority; dynamic AF area; 51-pts + 3D-tracking; 8fps. I have not have the chance to test this method enough to even comfortably say that it's OK.

jaime

 

[keithsnell]

I set my D700 to manual exposure at f/5.6-8 1/60s ISO 400-800, matrix-metering mainly to get iTTL-BL on the SB900.

Hi Jaime,

Good flash technique is one of the hardest aspects of photography to get a handle on.  What makes it even more challenging is that there isn't one solution for every situation.  What works in one situation won't necessarily work in another.  I use different techniques for fill flash in daylight conditions than I would for fill flash indoors.  

As a general rule, I try to "control the variables" in my photography.  That means that I tend to use modes and settings where I am in control, and not rely on the camera to come up with the right exposure based on the limited logic programmed into its exposure algorithms.  One of the reasons I am saying this now is to help explain why I don't normally use iTTL-BL mode on the flash.  It introduces a variable that will change based on shooting conditions.  If the camera's exposure system decides that ambient is underexposed, it will reduce the flash exposure accordingly in order to "balance" the flash output.  But it only does this to a certain point.  If ambient is too underexposed, the camera will decide that flash is the primary light source, and will expose the flash accordingly.  So, using exposure compensation to control the amount of fill, in conjunction with the BL setting on the flash, will result in "variable" amounts of exposure compensation depending on the ambient exposure.  Since I want to control (eliminate) this variable, I typically use iTTL (not BL) and set the amount of negative exposure compensation I want to provide adequate fill.  For me, this give more consistent (and therefore more controllable) results.  BL works fine in many situations, it is just a bit too "variable" for my tastes.

I thought that by setting the D700 to 1/320s FP, the FP function will only be activated if I use 1/320s and above on the D700.

True.  You should realize however that (as mentioned on page 429 of your D700 manual) "flash range drops at speeds between 1/250 and 1/320."  Much like using FP, "you lose more than you gain" by going to a synch speed of 1/320.  I would only use this synch speed if I needed the 1/320 shutter speed.  I very rarely use FP mode, and would only use it if I needed to set a wide aperture and were able to do so without overexposing the ambient exposure.  

I would very much appreciate it also if you could teach me the best method for varying light conditions and or moving subjects. I know that  the D700 will automatically expose for 0.0EV with a setting of varying ISO but fix aperture and shutter speed - so I try to spot-meter on middle gray equivalents (skin, green or near grey shirts) or on whites (I dial an EC of +2EV) or blacks (EC of -2EV). For focusing, I use the AF-ON button; AF-C, release priority; dynamic AF area; 51-pts + 3D-tracking; 8fps. I have not have the chance to test this method enough to even comfortably say that it's OK.

There are a couple of things I'd like to address here, and so will answer in a follow-on post.

Keith

[keithsnell]

For focusing, I use the AF-ON button; AF-C, release priority; dynamic AF area; 51-pts + 3D-tracking; 8fps. I have not have the chance to test this method enough to even comfortably say that it's OK.

Jaime,

I'll address the focus settings first, and then come back to the exposure settings.  (Please understand that many of these recommendations are "personal preferences" and reflect what I am comfortable with.)  51 pt 3D tracking works remarkably well, especially for tracking objects with a strong color component when they are moving from side to side in the frame.  Still, If find that the placement of the active focus point often isn't as precise as I would like it to be.  The 3D tracking is a significant update to autofocus in that it uses color information to track the subject.  So, once you focus on a particular color, it will "track" that color across the frame.  However, the position of the focus point can shift within that region of color.  

I find that "manual" selection of the autofocus point can give me more precise placement of the focus point, which is important to me.  I also find that most of the time when I am using one of the "dynamic" focus modes, It is in a situation where I have a long lens on the camera.  The 9 point and 21 point dynamic focus settings work very well in this situation (long lens with large aperture and "defocused" background).  In general, you should select the lowest number of active focus points that will allow you to stay "on your subject."  If your subject is moving very erratically, you might have a difficult time keeping a single focus point on the subject, and selecting more active focus points will give you a "margin of error" around the subject.  Many people mistakenly believe that if they select 21 autofocus points (for example) they can place their subject anywhere within those 21 AF points and the subject will stay in focus.  This isn't quite the way the AF system was designed, and "reorienting" your thinking can result in much better results.  It is much better to think of these AF options as giving you a "selected" AF point, with a "margin of error" around that AF point.  

For Nikon D3, D700 and D300 cameras, selecting 9 active auto-focus points will give you a "selected" auto-focus point with a "margin" of one focus point all around the selected AF point.  That way if the subject moves out from under the "selected" auto-focus point, you will hopefully still have a focus point in the surrounding "margin" on your subject and can react to reposition the camera and retain focus.  Selecting 21 active auto-focus points gives you a margin of two focus points all around your selected auto-focus point, which allows for a more erratically moving subject (or slower reaction time  .  Selecting more active focus points than you actually need increases the chances that one of the active focus points will "steal focus" by focusing on a high-contrast element in the background, so again you should choose the minimum number of focus points needed in order to keep your active focus points on your primary subject.  ("Following" a moving subject is a skill that gets better with time and practice, so experienced photographers are usually more successful with fewer active focus points.)  And as I noted earlier, the 9 and 21 point dynamic AF settings work best with a long lens and large aperture, where the background will remain "defocused," so as not to "steal" focus from the subject.  For a normal to short telephoto, 51 point 3D focus probably works better for subjects with significant side-to-side motion in the frame, especially if you are photographing a subject with strong color components; however, I normally prefer the "control" of selecting my own AF point and precisely placing it where I want in the frame.  

I also prefer "focus priority" instead of release priority.  I don't see any sense in cluttering my computer with out-of-focus shots, and so largely eliminate those by selecting focus priority.  Yes, this might cause me to miss a few rare shots; however I would attribute that to my own ineptitude at selecting an "appropriate" location to place my AF point in order to obtain good focus.  (i.e., with experience, you learn where the camera will "grab" focus, and where it won't.)

I would very much appreciate it also if you could teach me the best method for varying light conditions and or moving subjects. I know that  the D700 will automatically expose for 0.0EV with a setting of varying ISO but fix aperture and shutter speed - so I try to spot-meter on middle gray equivalents (skin, green or near grey shirts) or on whites (I dial an EC of +2EV) or blacks (EC of -2EV).

I am a very strong proponent of using manual exposure whenever possible (back to my desire to "control the variables"); however, when I am attempting to photograph moving subjects in varying light conditions, I often find that I get better results by relying on the camera's autoexposure modes.  (Of course this depends on how "variable" the light is, and how much the subjects are actually moving.)  Assuming the subject is moving fast enough that I am using AF-C, and the light is truly variable, I will usually use Aperture Priority mode, with matrix metering, and set exposure compensation appropriately for the tonality of the area where I intend to place the active AF point.  Theoretically I might be able to get better results using manual exposure and spot metering; however, pragmatically I know that it is difficult to manage with moving subjects in variable light, so I let my expensive exposure system go to work for me.  Just be sure you are conscious of the tonality of the area where you are focusing, and make sure you adjust your exposure compensation accordingly if your focus area is significantly brighter or darker than mid-tone.

Again, many of these recommendations are based on my personal experiences, and I can only tell you what works for me.  Please be sure and ask questions if what I have said doesn't make sense.

Keith
  

[girod]

We just came back last night from a week-long vacation in Orlando and as you probably know better than me - with 3 and 5 year-old girls, it's either you're on the go or you're dead tired....nothing in between......but lots of practice in photography.

First off, thank you very much for the "Editor's Choice" on my "Unbearable Uncertainties". It's my first ever in my year-long willful photography....my first participation too outside my families and friends. It meant a lot to me and I'm so inspired.

I'll try your i-TTL (without BL) +/- EC. Do you spot meter for the background that you want to preserve and adjust (trial and error) for the amount of fill flash needed on the dark subject? I am assuming that you dial the needed EC in-flash and not in-camera. Do you have a ballpark guidelines for the right amount of fill flash to lessen trial and error?

It's only recently that I am using the 51-pts 3D tracking of my D700 and spot-metering technique (fixed Aperture and shutter speed; auto ISO; dial EC appropriately for the tone of the main subject) - to see if the combination of AF (phase-detection and contrast-detection) and the Scene Recognition System technologies can fill-in into my skill inadequacies in fast actions. If this does not - I'll go back to 9/21 pts, Aperture/Shutter priority, matrix-metering and really try harder to improve my focusing skills.

jaime

[keithsnell]

I'll try your i-TTL (without BL) +/- EC. Do you spot meter for the background that you want to preserve and adjust (trial and error) for the amount of fill flash needed on the dark subject? I am assuming that you dial the needed EC in-flash and not in-camera. Do you have a ballpark guidelines for the right amount of fill flash to lessen trial and error?

jaime

Hi Jaime, 

Welcome back.  It sounds like you had a great (if very busy) trip. 

My metering for the background varies based on the scene, but yes, typically for a high contrast scene I will use spot meter.  Yes, I dial in EC on the flash.  I've found that in general -0.7 EV is a good starting point, if I'm using TTL.  It is important to recognize that changes in subject size, position in the frame, and reflectivity will affect the camera's flash metering, so you might need to adjust the EC accordingly.

Hope this helps,

Keith 

[girod]

Thanks Keith.

jaime