Sekonic - L358 Calibration

[WayneF]

In writing? No, but I've spoken to them many times. MAC Group knows me quite well. I both use extensively and am involved in beta testing for several products they distribute. Again, i'm just acting as a messenger, in case you want me to show you the math. :wink:

Be careful of the message, some messengers were shot. :smile: Forgive me, I know your other work is much better, but this one message seems nonsense. I know this story is widespread, and you are just repeating what you have heard. I know many serious pros have heard it long ago (before we really understood much), and assumed it had some merit - after all, they did say "18% was middle gray", right? But it is nonsense nevertheless. It is not a large numeric error, but still wrong thinking, it simply cannot be true.

I suppose there is no telling what some individual might say to you, and I can believe he did. But I was rather confident Sekonic (the company) never mentioned calibrating from a gray card. There is no clear way to do it, but Sekonic echoes the ISO specification's words, to "make compensation after a sufficient number of test in actual photographic conditions". Sufficient number excludes one test result, and actual photographic conditions excludes gray cards or other test targets. Certainly Sekonic never mentions calibrating with gray cards and histograms. If it were that easy, Sekonic would be all over it. And they are not.

I'm just trying to tell it as I understand it. So what about some of the facts I've pointed out? Please refute away. I don't think I misunderstand, but I would certainly be interested in your explanation how and why 18% ought to come out at 128 ? How can that possibly be? Magic maybe, like a fairy tale? :smile:


IMO, imagining any relationship between a 18% card and the center of the histogram simply simply ignores all the clear facts that say these cannot possibly be related. Simply thinking about it a minute obviously refutes 128. It turns out to be a small error numerically, but it is a huge blunder conceptually.

1. When metering from such a card, the reflected meter simply gives a reading which depends on
a. the reflectivity of the card (reads higher for greater reflectivity)
b. the total incident light on the card (reads higher for a brighter light)

Regardless of that result, regardless of a or b, all such corresponding exposure readings will cause the camera to expose that metered card (specifically a frame consisting only of the card view) to the same mid-gray result in all cases, regardless of a or b above.

Clearly the 18% source has absolutely nothing to do with it. The same will be true if we use a sheet of white or black paper. The meter's 18% number is not about the scene, it is about this resulting manipulation of any scene, except that it is closer to 12%. True of any scene in front of the camera, but a blank sheet of paper is easier to see - any shade of card you choose, 18% is not a factor.

Trivially easy to try it and see it and know.

2. But there are still problems. This result is highly variable anyway, depending on what the camera firmware may be doing in regard to contrast S-curves, or White Balance, or Brightness, or Saturation, or any of other ways the camera manipulates things. When the camera makes tonal changes, it changes things. The camera is a confusion factor. This one part is indeed hard to predict, or to count on, but all the rest is obvious.

3. If the photo goal is to reproduce the card, reason says 18% ought to come out 18%. It is not 50% of anything. This may be the weakest argument, but still obvious nevertheless.

4. Any histogram we can see contains data which has been gamma encoded. Our RGB standards require gamma 2.2, in all images. Without question, this means the image data values are modified to be different numbers. Ideally and theoretically, in the perfect world, the gamma formula says 18% ought to come out 117 at 46%, and center 128 ought to come out 187 at 73%. But these numbers will vary due to camera manipulations, maybe ballpark but not precise. A gray card is often a little lower than midpoint, but it may be higher. Position is not related to midpoint, it has absolutely nothing to do with the middle of anything. Numeric position is related to gamma. To recalibrate the meter because the gray card comes out a little below center seems a stupid thing to do. Keep what you've got.

5. We know that when we stop down one stop, we expect 50% of the light. So if we carefully expose something white so that it just touches the 255 right end, and then intentionally underexpose exactly one stop, we expect half the light, and expect the 255 end of the data to fall to midpoint at 128. These are definitions, including definition of midpoint. Midpoint is where that 50% goes.

And it would do exactly that, 128, in linear RAW data at the sensor.

But we never see RAW, and this also confuses the troops. What we see is after RGB conversion and gamma encoded data, and we see that in fact, this one stop test only falls to about 3/4 scale, as gamma computes that it should. Point is, while one stop down is clearly 50% of the light, it is nowhere near center of the histogram. Or rather, that midpoint has simply been shifted up to about 73%. Why isn't this pesky 18% card up there with it? :smile:

Trivially easy to do this, and see it, and know. Clearly obvious, just look at it.

Theories should encompass the known facts, those we can see actually happen. The histogram simply does not show anything like you must imagine that it does. To change Sekonic's light meter calibration based on this seems real shame. I trust Sekonic's calibration, instead of those incorrect magical notions of gray cards at 128 somehow. :smile:

 

[Weebles]

I trust Sekonic's calibration, instead of those incorrect magical notions of gray cards at 128 somehow

I agree, as I said, we assume the Sekonic meters indicated ambient exposure is correct. It is the camera that is either over or underexposing when using the meters indicated exposure settings.

Thus, one must be able to correct the cameras exposure so that it matches what the meter indicates. And you have to be able to validate any compensation you made to correct it when compared to the meter.

Although one can look at an image and say "That looks good. I like that exposure", that is not what I'm referring to. And that is also not what the OP asked about. He asked how to calibrate his meter.

Remember, this is an ambient light exposure. Therefore, neither my L758DR or my cameras reflected meters should have any bearing on this discussion for the moment.

There is a reason Sekonic meters have a calibration feature that is user selectable. And it's not to calibrate the meter if you determine the factory calibration is wrong. It is to calibrate the meter so that it matches ones camera/s. Up to three camera settings/calibrations can be stored.

Lastly, when performing this calibration I'm only talking about a couple of tenths, not full stops. If the latter is the case then one may need to send their camera in for repair and/or go back to school. :wink:

Edit: To my knowledge Sekonic has never detailed, in writing, how to calibrate anything using their meters. The only mention of calibration is what is shown in the user manuals, which I posted, verbatim. However, that does not mean there is no method of doing so.

 

[TonyW]

......Remember, this is an ambient light exposure. Therefore, neither my L758DR or my cameras reflected meters should have any bearing on this discussion for the moment...

Gentlemen and OP. I have an apology. I really should have read and checked the original post before putting in my 2 cents. :redface: For some reason I had read this post as if the discussion centred around reflectance meters and metering I mistook the L358 for the L508 which offers both options :redface:.

Having been part of the discussion (which I found both interesting and informative) I do wonder if it has strayed a little too far from the OP's original intent - he has not been back to comment for a couple of days

 

[WayneF]

In the current line, the Sekonic L308S and L358 and L758 offer both choices, incident or reflective metering (at least those do, and likely other models do too.)

The rest is not really addressed to you Tony, as I feel certain you already know, but maybe others may be interested.

If we meter "from a gray card", that necessarily implies reflection from that card. In contrast, incident meters of course are aimed at the camera lens, or sometimes at the light source, but never at the subject (and therefore are not affected by the subjects reflectivity).

Incident meters measure the light source directly, and reflected meters measure the light reflected from the subject. Black or white subjects reflect very differently, but the incident light is the same either way. This incident reading is a GREAT result, but it is awkward to perform, because incident meters must be used at the subject, to meter the same light actually incident on the subject. In the portrait studio, we might hold the incident meter under the subjects chin, to catch the same light hitting the subjects face. Camera meters cannot work that way, it must read reflected light finding the path into the lens, so camera meters are reflective meters.

And this is a really huge difference, which those owning only reflective meters may not realize (the camera's internal meter is necessarily a reflective meter).

If we use a reflected meter, no matter what we aim it at will come out a middle gray tone. That includes 18% cards, white cards, black cards, anything at all will come out the same middle gray result (assuming the photograph area only includes that metered card - the area actually metered).

If that photo area includes a beach scene with sky, water, and pretty girls, in the same way, the overall average for that metered scene will be made to come out middle gray. The reflective meters goal in life is to make all scenes average out to a middle gray average. A general scene has all manner of things in it, and it likely does average out about middle gray, and this expected average is the only factor in this system's design. But a reflective beach or snow scene, or a white wall background, (more reflective than "average" scenes) may come out underexposed a bit (same as the white card).

This is the only "smart" that the reflective meter has. The meter is the dumbest computer with absolutely no clue what the subject is, or how it "ought" to come out. It will come out middle gray average. If it is an average or typical subject, it is probably close to correct. If it varies much from average, then probably not. If we expect better intelligence from the system, we will be very disappointed. We must learn to use the tools we have. The photographer provides the intelligence. :smile:

If we use an incident meter, and from the subject's position, properly aim it at the camera lens (so that the light source is incident on its dome), then it meters the light itself, and not the subject. Meaning, black subject things will come out about black, and white subject things will come out about white. This is great, but sometimes awkward to meter at the subject. One exception, outdoors in full bright sun, we can use the incident meter from the camera position, the sun is very distant and sunlight is all the same everywhere then (if in full bright sun).

Or if we meter a card to obtain a reflected exposure setting from it...
that card itself (18%, white, black, whatever, as the sole subject content) will come out the same middle gray. A white card will be underexposed to be middle gray, and a black card will be overexposed to be middle gray. A middle gray card will also be made to come out middle gray.

But if we use that card's reflected metered exposure setting as the exposure settings to then photograph some OTHER general subject (not the card this time - but the meter aimed at the card for reading, and camera aimed at another subject IN THE SAME LIGHT), then this method becomes more like the incident meter. We sort of measured the light, instead of the subject.

To avoid the 18% number for a second, if the card was say 42% gray, then with this method, all other 42% things will come out the same middle gray, and darker things will come out darker, and lighter things will come out lighter.

This same idea of course works better if we use a 18% card, since this better matches the expected average range of general typical subjects. However, Kodak does say to open 1/2 stop from metering their 18% card, which simulates metering a 12% card, which matches the meter's 12.5 constant better.

But this 18% or 12% is merely a numerical average, and all subject scenes will vary. It is a good average, but scenes vary, and we need to pay attention. The main point here is that it is good to realize how the reflected meter works.

 

[paradiddle]

I am here. Reading trying to digest what is being said.

I appreciate all the input. I dont think the thread is hijacked. Exactly what I wanted.

It just takes me time to read and understand. Keep it up. It may take a week or so for me to get the time and confidence to adjust my meter.

Subscribe to see EXIF info for this image (if available)

Here is an example picture of what I am talking about. Kind of blowing out on the detail between the nose and check on the lights side.

I am thinking either the meter is off or the model is moving closer and I need to pay more attention.

 

[Weebles]

Again, not necessarily the meter, but rather your camera when using the handheld meters exposure. Even assuming the handheld meter is correct, different camera and lens combinations can vary when using the same exposure. Even a couple tenths of a stop can make a difference with some subjects.

This is one of the reasons I like the L758DR for critical work since it can be calibrated to graphically display the dynamic range and clipping points of an individual camera and lens.

And like you said, models can move since they're human, not statues. This is why using a mark for them to stand on can be helpful. Professional models are more aware of the importance of this because they've been yelled at a lot. :wink:

 

[TonyW]

Looking at this image my impression is that there is very little wrong with your exposure and I would guess that neither your meter or camera or your interpretation of exposure is incorrect

What I think you may have done is to try and maintain the nice looking shadow areas for hair and flesh tones by compensating somewhat in post processing.
In short the contrast ratio between the light and dark areas was perhaps a little too much for your taste and you have generally lightened the whole.

The reason I think this as a possibility is that the information is all there in the original file without any signs of clipping.
I have taken the liberty (and if I am out of order will remove) of darkening your image way OTT but it does I think help to prove the theory - in darkening the skin values on the image left the seperation lacking in the original is back (with the exception of reflection around the nose). Of course this has caused the shadow areas to go to deep

While it can be fixed post processing it is not the ideal solution. So in this case it may have been better to add extra fill on the shadow side to balance and lighten

Subscribe to see EXIF info for this image (if available)

 

[Weebles]

Just to be clear, I'm not suggesting the exposure is incorrect in this case. I was merely stating what can potentially lead to exposure errors in some cases since the metered exposure or the models position was brought up.

In addition, a makeup artist is very helpful for both noticing and addressing hot spots and reflections for example, which are not necessarily a result of exposure errors or lighting choices. Although the latter can certainly be tailored to suit the model and situation.

 

[TonyW]

...In addition, a makeup artist is very helpful for both noticing and addressing hot spots and reflections for example, which are not necessarily a result of exposure errors or lighting choices. Although the latter can certainly be tailored to suit the model and situation.

I for one did not take your post to mean you were suggesting incorrect exposure. It does look like a makeup artist would have been helpful on this shot as well. It looks like at least 2 lights were employed one from above left and one from below left either one or both of them have combined to highlight a particular area of shiny skin. Funny thing is just a slight turn or tilt of the models head and it may not have been as apparent.

 

[paradiddle]

You guys are way more technical than I.

There was a little clipping in the RAW file. I believe I adjusted to a -.5 exposure in ACR.

Lighting. Camera Left - AB800 with Large softbox about 1 - 1.5 feet away from the model. The model is positioned towards the rear of the light. Camera right is a 4X8 White foam core book end. Maybe 2 feet away.

I don't mind that you edited the photo. Thanks for spending the time.

There was no make up artist.

 

[WayneF]

Lighting. Camera Left - AB800 with Large softbox about 1 - 1.5 feet away from the model.

It is a very nice picture. If you left the modeling lights on, it would close the pupil of her eyes much more (if desired). f/9 1/200 second will keep the modeling light out. One test picture with sync cord pulled out (without flash) will show that the fast shutter speed still leaves the picture totally black. This should always be verified, but it will be true.

My bet is that this too-close distance was most of the problem. The complaint is that the near cheek is too bright. Close is good, but maybe is overdone here. Regardless of the black background, this one borders on a high key picture (it seeks a more even light, IMO).

The inverse square law says the light intensity falls off fast with distance. 24 inches is 2x 12 inches, so 24 inches will be two stops down from 12 inches, no matter what you do (i.e., the near cheek is going to be pretty bright). This is difficult to adjust for.

But if the light were at 24 inches, it takes 48 inches to drop two stops. If it were at 30 inches, it takes 60 inches to drop two stops. Point is, if the light were at say a reasonable 30 inches, the light would be much more even across the face, and could then be adjusted for lighting level. The reflector fill would be relatively stronger then too (relatively less distant in relation to the near light distance).

As to meter accuracy, there are so many other factors. I don't know this situation, but exactly where you hold the meter is important (esp at this very close distance where every fraction of an inch is so important). To meter the light on her, it should have been at her same distance from the light. This close, it is probably difficult to repeat the same reading twice.

Metering technique is all important, and frankly, it takes awhile to learn, to be consistent. We strive for consistency more than for absolute accuracy (we can always compensate for accuracy).

Was it an incident meter? I assume so, but you mentioned the gray card, which confused me. Spot metering is a reflected meter term. If actually spot metering, then which spot did you aim it at? That is the spot that will be made middle gray (a gray card? It will come out gray, if it remains at that location), and anything else (the rest) comes out however it does, following along.

If you have an incident meter, there is no reason to even think about a gray card (well, possibly for white balance, but a lighter card, or a white card, would be better for that). Metering a gray card is just done to eliminate the effect of the subject's variable reflectivity... the incident meter already does exactly that, without the middleman.

Which direction the incident meter is aimed is important...aimed at the light or at the camera lens, etc. Technique is everything, regarding consistency.

Regardless, at least until a consistent metering technique is achieved, it is a bit naive to assume we just point the meter, and it is always right. Every case will be a bit different. It is less about the meter than about our technique using it. We always have to look at the result, and maybe adjust exposure a bit. That is what photographers do. If it is not perfect, just fix it so it is. That is the easy way.

We can easily zoom the Nikon rear LCD to show an enlarged result view, to fill the entire LCD with just the face, maybe just the cheek - to see the details well. The histogram is also there, check for RGB clipping. I set my D300 LCD brightness to -1EV, to better match the way my calibrated computer monitor shows the same picture. There is a little difference, the camera LCD is showing RAW as a small embedded JPG which does have the camera settings applied, whereas the RAW image in the computer may not have those same camera settings in it. So color may differ, but we can see exposure.

The only time we would even consider calibrating the meter is that after many many times, if we observe we always seem to have the same error. Then we might calibrate to offset that systematic error. But this is anything but a one time trial.

 

[TonyW]

.....The only time we would even consider calibrating the meter is that after many many times, if we observe we always seem to have the same error. Then we might calibrate to offset that systematic error. But this is anything but a one time trial.

+1 Absolutely agree. From what I can see there is no real indication that your meter needs calibrating until as Wayne says you can prove in a variety of situations the repeatability of any errors you observe.

I think that I may even be tempted to turn the camera to manual for a while and use the Sekonic only trusting the setting it tells me is correct - and of course my technique is also correct. After evaluating the images in your application of choice after a while you will be able to make a valued judgement if the meter is telling the truth (which I suspect it is) and your technique is ok. After all that then you can compare camera and meter and if you do not get a match then you can make a decision to either match camera to meter or meter to camera or just keep in mind and make manual adjustment.

You have now got me interested and got me to do something I have intended to do for a while i.e. have dug my old Soligor Spot meter Zone System modified! out of the attic and am going to have a play. As you can probably gather generally I am/have been a fan of reflected light meters over incident for the vast majority of my work. Not suggesting this method is better or worse, they both have their pros and cons so its just my preference most of the time. That and the fact my Weston with invercone gave up the ghost a long time ago!

BTW forgot to mention nice image of an attractive girl

 

[paradiddle]

Wayne - modeling lights were on. I even had a shop light shing towards her eyes trying to get her eyes to un-dialate.

I think Weebles has the best understanding of what I am thinking. I really don't want to use the reflective meter. I am trying to get my meter to match my camera / lens combination.

 

[TonyW]

... I really don't want to use the reflective meter. I am trying to get my meter to match my camera / lens combination.

Gary, I think we can all understand this, however there is a problem imo i.e. how can you match your camera meter which by its nature is reflective (regardless of exposure mode) with an incident light meter.

As an incident light meter measures light falling on the dome (ambient light) and your camera actually meters the scene via a Matrix, Centre weighted or Spot - all of these are actually reading light reflected off subject in the scene. This may help clarify or This article

In theory if you measure a grey test card (12% reflectance) correctly it should give you the same exposure values as an incident meter does. As Wayne pointed out earlier the camera meter will adjust to whatever is put in front of it to produce the same shade of mid grey even if you use a white or black card - so the reflectance of the card is most important if you want to get accurate results. If you want to use a white or black card that is also fine but then you will have to apply manual compensation i.e. open up or close down a couple of stops depending on the card to compensate for the meter wanting to record them as mid grey

If you want to match your camera then you will need to turn it into an approximation of an incident meter and fit a diffuser over the lens take reading and set exposure manually remove diffuser and take image - seems very long winded to me! There is a gadget that purports to do this (and set white balance) its called Expodisc. Small review from Ken Rockwell http://www.kenrockwell.com/tech/expodisc.htm
Cannot make any comment as I have never used it

 

[Weebles]

how can you match your camera meter which by its nature is reflective (regardless of exposure mode) with an incident light meter.

The idea is not to match the cameras reflective meter to the handheld meter. It is to match the cameras exposure so that it's the same as the handheld meters incident exposure.

For example, if the Sekonic meter shows the correct incident exposure to be f4 and 1/160 at ISO 200, yet the camera is consistantly over or underexposing when using that exposure, then one can calibrate their Sekonic meter so that it matches what the camera is doing to within 1/10 of a stop. If the exposure variance is too great, however, then the camera may need repair.

Although Sekonic meters are assumed to be correct from the factory, there is no way to accurately calibrate the cameras exposure to within 1/10 of a stop. Hence, one can calibrate a Sekonic meter to an individual camera.

Some Sekonic meters, such as the L758DR, have an additional calibration feature (not to be confused with the process above) which, in conjunction with a Sekonic test target, allow one to calibrate the meter to a camera using photos taken of that target with the camera. The Sekonic meter then displays a graphical representation showing the minimum and maximum clipping points and dynamic range of the camera. This allows one to see whether or not the metered scene, including spot metering, falls within the cameras capability before the photographer takes the photo.

 

[TonyW]

The idea is not to match the cameras reflective meter to the handheld meter. It is to match the cameras exposure so that it's the same as the handheld meters incident exposure.

Ahh I see, so I nearly grasped the wrong end of the stick - again :smile:.

I say nearly as in one of my replies I stated "I think that I may even be tempted to turn the camera to manual for a while and use the Sekonic only trusting the setting it tells me is correct"

Gary said " I really don't want to use the reflective meter. I am trying to get my meter to match my camera / lens combination". So by the sound of things he is going to keep the camera on 'M' and just use the Sekonic suggested exposure to set the values! Nothing wrong with that but does seem a shame to loose some of the cameras metering functions.

The L758DR sounds like one hell of a meter

 

[Weebles]

Not to confuse matters, but there are times when a reflective meter is necessary, be it the cameras or that of a handheld meter that has one such as the L758DR. For example, with backlit scenes you want to use a reflective meter since the light is behind the subject and therefore you can't measure the light falling on the subject as seen from the cameras viewpoint.

And yes, typically one uses manual exposure mode when using a handheld meter. They are also useful for measuring flash ratios and make a good learning tool in that respect as well.

 

[paradiddle]

So it appears that to calibrate the hand held meter to the camera lens combination......I dial in a -.4. I was getting a 126 - 129 type of numbers when selecting the grey card.

It was really close to the center of the histogram. Even before I changed the calibration of the meter.

One other reason to not use the in camera meter in the studio......
......I am using strobes.....so prior to pushing the shutter....there is no light for the reflective meter to read!

I do use the reflective meter when I am outdoors, sports, or subject is to far away to walk up and measure.

I really wish I had spent the extra money for the L758 for the spot meter.

Thank you all for so much discussion and thought.

Gary

 

[TonyW]

So it appears that to calibrate the hand held meter to the camera lens combination......I dial in a -.4. I was getting a 126 - 129 type of numbers when selecting the grey card.....Gary

So what you appear to be doing is shooting a grey card of some undisclosed reflectance at the Sekonics suggested incident/ambient reading and expecting to get figures in Photoshop or approx R=128, G=128 and B=128 then if the figures are incorrect you are dialling in the +/- correction on the Sekonic?

It was really close to the center of the histogram. Even before I changed the calibration of the meter.

Have to say that this suggests to me do not 'fiddle' with it just yet until you have undertaken much more extensive testing - just my opinion of course!

Of course if you are happy with what you are getting and image quality is all you would want then what follows could be taken as quite irrelevant as any errors may be considered small and of course easily correctable?
Also bear in mind I have no experience of the Sekonic meter or calibration procedure :wink:.

Also I am not trying to 'muddy the waters' or make things more complex than necessary but have you considered these points:

1. As mentioned previously RGB values of 128 are not in the middle of the real world but they are the centre of a scale 0-255 used in Photoshop and other graphic programs.

2. Your grey card may when exposed correctly need to appear in PS as RGB values of 128 (it does depend on its reflectance and where you want to put it in the histogram and in your editing program). But you need to be aware that RGB values of 128 equate to luminosity of approx 21.6%. If your card is either 18% or 12% then your exposure may be somewhere off ideal if you wanted to place the card value precisely.

How important it is in the real world and how much difference it makes is up to you to decide.

Being curious (and somewhat anal this evening!) I did a little research just for the hell of it and came up with the following figures which I hope and believe are somewhere in the area of being correct. I also converted to a Photoshop image as I much prefer visual representation rather than the math.

Calculations based on using Bruce Lindblooms excellent calculator (hope I am using it correctly). Bear in mind that quite a few factors affect the calculations such as white point, RGB model i.e. sRGB, Adobe RGB or ProPhoto.
http://www.brucelindbloom.com/index.html?ColorCalculator.html


You need to agree that an absolute 18% reflectance, corresponds to middle grey, which is a Lab luminance of 50.


18% Grey Card
Ref White = D65
RGB Model = $RGB Gamma 2.2
Lab = 50 0 0.
XYZ = 18.41 Y which is luminance
Red, Green and Blue values = 118.913. call it 119


12% Grey Card
Ref White = D65
RGB Model = $RGB Gamma 2.2
Lab = 42 0 0
XYZ = around 12 Y which is luminance
Red, Green and Blue values = 99.086. Call it 99

Photoshop Model = to approx 21.6% grey card
Ref White = D65
RGB Model = $RGB Gamma 2.2
Lab = 53.6, 0 0
XYZ = around 21.6 Y which is luminance
Red, Green and Blue values = 128


Translated into Photoshop values

Subscribe to see EXIF info for this image (if available)

 

[paradiddle]

I apoligize but I don't understand all of that.

What I do know is.....it seemed when I would meter a model and later edit....my shots would be a little hot - with a small amount of clipping on the face.

I was watching a Kelby Training Class - Frank D. He said calibrate your meter to the lens / camera combo you are using. So that is what I am trying to accomplish.