Can I rely on My K1000's Centre Weighted Metering?

[wiltw]

The thinking behind center weighted is to limited the amount of exposure given to sky. Some makers such Miranda Sensormate used bottom weighted in which the metering pattern was etched into the mirror with heavy strip on the bottom of the horizontal orientation with a spike in the middle to catch the sky. I found bottom weighted to work quite well for landscapes but used a hand held meter for portraits Nikon center weighted metering was good enough that most pro Nikon users relied on it in most lighting without issues. But just as with spot metering, matrix metering there is tricky lighting when a hand held incid;ent meter is best.

Even modern camera Evaluative metering is 'weighted', but we do not know what the weighting is (what percentage of the metering result is biased, and to what degree). Here is a Canon dSLR using Evaluative to 'weight' the zone where the AF point is located.

...and although an 18% gray card is the point of focus, it is severely underexposed because of the sky preponderant in adjacent metering zones. Clearly the 'weighting' is insufficient to render the gray card anywhere close to its inherent brightness.

It should have looked like this brightness level (metered via spotmeter)

[

 

[Paul Howell]

Matrix metering compares the segments to stored information to figure out what the scene is, backlit, landscape, ect the provides the exposure that best fits the scene. I use matrix metering on my F4, most of my Minolta A mount bodes and Sigma SA and SD bodies with good results. I have noticed as wiltw did that a very bright sky will cause my Minolta and Sony bodies to underexposed, the foreground, while my old Sigma SD10 and 14 8 segment cell meter seems to do a better job. With my Nikon and Minolta and Sony bodies I switch to spot metering.

 

[wiltw]

Matrix metering compares the segments to stored information to figure out what the scene is, backlit, landscape, ect the provides the exposure that best fits the scene. I use matrix metering on my F4, most of my Minolta A mount bodes and Sigma SA and SD bodies with good results. I have noticed as wiltw did that a very bright sky will cause my Minolta and Sony bodies to underexposed, the foreground, while my old Sigma SD10 and 14 8 segment cell meter seems to do a better job. With my Nikon and Minolta and Sony bodies I switch to spot metering.

At one time, it was Nikon that stated that there was a database of photos whose metering zone brightness were factored into a choice used in Evaluative metering by their film SLRs. I do not know that similar claims were ever made by any other manufacturers. And whether or not Nikon dSLRs still use the same photo database as their film evaluative metering has not been stated by Nikon, to my knowledge.
AI claims to know:

• "Here is how the system works and has evolved:
  • The Core Database: The camera divides the scene into multiple zones and reads brightness, color, and contrast. It then checks these variables against an on-board database of over 30,000 reference photographs to determine the best exposure.
  • Modern Upgrades: While the underlying database concept remains, modern Nikon mirrorless and advanced DSLR cameras (like the Z8, Z9, and Zf) supplement this with advanced AI, deep-learning subject recognition (for human/animal faces, eyes, and vehicles), and newer modes like Highlight-Weighted Metering.
  • The "Evaluative" Name: "Evaluative" is the terminology used by Canon. If you are looking for this equivalent setting on a Nikon body, simply look for Matrix Metering"

[edit] I just found this reference in Nikon published product information

https://cdn-4.nikon-cdn.com/en_US/IMG/Assets/Common-Assets/DSLR_DigiGuide.pdf

• "Nikon engineers set their sights on expanding the potential of the renowned 1,005-pixel 3D Color Matrix metering sensor. Through this process, the Scene Recognition System was born. This amazing system provides unmatched detail analysis of any scene, recognizing both color and brightness to a much higher level of precision. Milliseconds prior to shooting, the Scene Recognition System recognizes a scene’s essential elements and then applies that information to the camera’s other analytic processes. As a result, AF, AE, i-TTL flash control and AWB have achieved an entirely new level of accuracy and performance that dutifully reflects the expectations and desires of today’s photographers. Moreover, Nikon’s SRS operates at every level from the D3000, D3100, D5000 and D5100 all the way up to the flagship D3 series cameras.

 

[koraks]

AI claims to know:

And is obviously wrong. E.g.:

The camera divides the scene into multiple zones and reads brightness, color, and contrast. It then checks these variables against an on-board database of over 30,000 reference photographs to determine the best exposure.

This cannot apply to the matrix metering in esp. older cameras, since those had no physical way to store a database of 30k images, nor to pattern-match that data within milliseconds, not even if the data are abstracted into a handful of bytes of data per image (which would be inadequate). Inefficient and essentially impossible with the technology as it was when Nikon's matrix metering/scene recognition came into being.

The way it likely works is kind of similar to how AI itself works (ironically, since "it" apparently doesn't "realize"): you take a large set of data, in this case images with an accompanying dataset of abstracted color and exposure data (as would be generated by the matrix sensor), as well as a set of desired outputs of the scene recognition system (i.e. focus point choice, desired depth of field, exposure settings). This doesn't happen in the camera; it's part of the R&D effort, so it happens "in a lab" so to speak.

Then there are basically two ways to go forward: one is to take a neural network and train it to these data, after which the resulting neural network can be hard-wired into an ASIC that forms the 'brain' of the SRS (black box approach). The alternative is to take a white-box approach and purposefully design an algorithm that performs the desired permutation on the inputs, which would then also be hard-wired into an ASIC or alternatively run as firmware on a microcontroller (the former would have been used up to and into the 1990s and early 2000s, the latter is likely how it's done today as it's more flexible).

In this approach, the camera doesn't have to carry around a large database of image data. In fact, it doesn't have to contain any data at all: it only embeds a (hardwired/programmed) way to interpret data acquired from its onboard sensors.

Note that while the AI output is technically incorrect, for practical purposes it can be a sufficient way to understand (approximately) what's going on. AI is also not solely to blame - manufacturers also have made a habit of trying to simplify technical underpinnings to explain a product to the user base, and in doing so make up all manner of fancy stories that really aren't true, but sort of explain what is going on under the hood. To draw a somewhat silly parallel: as long as you don't open the bonnet of your car, a sufficient understanding might be that there's gnomes in there that pour cups of gasoline into engine bits, or opening and closing valves, and other gnomes running around with thermometers and a bunch 'office gnomes' who manage the whole lot. It's not accurate, but it kind of explains some of the principles behind how a modern car engine runs.

 

[tcolgate]

.Then there are basically two ways to go forward: one is to take a neural network and train it to these data, after which the resulting neural network can be hard-wired into an ASIC that forms the 'brain' of the SRS (black box approach). The alternative is to take a white-box approach and purposefully design an algorithm that performs the desired permutation on the inputs, which would then also be hard-wired into an ASIC or alternatively run as firmware on a microcontroller (the former would have been used up to and into the 1990s and early 2000s, the latter is likely how it's done today as it's more flexible).

In the 80s Olympus did the training and then baked the resulting data onto a patch that sits over the shutter curtain (IIRC). The meter reads off of that curtain. I vaguely recall that the Leica M9 (and maybe M10) use the same approach.

 

[wiltw]

And is obviously wrong. E.g.:

This cannot apply to the matrix metering in esp. older cameras, since those had no physical way to store a database of 30k images, nor to pattern-match that data within milliseconds, not even if the data are abstracted into a handful of bytes of data per image (which would be inadequate). Inefficient and essentially impossible with the technology as it was when Nikon's matrix metering/scene recognition came into being.

I think you are too literally interpreting the use of a database of pictures, 'stored' within the camera. Digital storage was not employed, but the analysis of many images to determine an algorithm to embed. Nikon says the data was analyzed 'in an analog way'...
Nikon debuted Automatic Multi-Pattern metering in the Nikon FA in 1983, but it relied on algorithms and a 5-segment meter rather than a reference photo database.

I found Nikon's own description:

Nikon Family Cousins, Part 14 : "Nikon FA" and "Nikon FE2" | Camera Chronicle | Nikon Consumer

"It was a device to measure the information on the brightness of the subject in each area of 24-segmented image plane, by interpreting the graphic output produced on paper by this pen recorder. At present this is processed with a personal computer in the way that the output signal of the metering device is input as a database by means of an A/D (analog-to-digital) converter, but in those days, it was processed in entirely 'analog' manner....​

"Further, they took pictures of variety of scenes as far as they can imagine, and repeated the brush up of the algorithm by analyzing the results of them. As a matter of course, staff members went to many spots of the world to collect the data of other than Japan that is located in the mild temperate zone.​

​

Before the world first "Matrix metering" was completed and incorporated into Nikon FA which was released in September 1983, six years and the efforts of many people involved in the project were required."​

In 2005, Ken Rockwell's write-up:

"The superior Matrix meter is the main reason to choose Nikon over other brands.​

​

"Nikon's Matrix metering, introduced as "Automatic Multi-Pattern" (AMP) metering in the FA camera in 1983, was the world's first meter that actually measured exposure, instead of just light. It is one of the most important advances in photographic technology. This meter knows how to make white snow or sand look white, instead of a conventional light meter's making everything look medium 18% gray. It applies the zone system automatically to attempt to render a correct exposure under difficult and contrasty situations. When shooting in a hurry under rapidly changing conditions, which is the whole point of using a small format camera like a Nikon, there is no better way to meter your exposures....​

​

"All the other major SLR makers since about 1990 have imitated this meter under many different names. Canon calls it "evaluative," and most camera makers brag about it by specifying how many sensors they use. Even Leica attempts to copy it. Today's Leica R8 has about the same technical sophistication as the 1983 Nikon FA."​

 

[IpseLux]

Put a white coffee foam cup over the lens and take a reading (incident). Or point to the clear blue sky, 90 degrees from the sun. Both should get you a good printable exposure.
Traditional auto metering works if the lights and darks in the image average to middle gray. In real circumstances they seldom do.
Ansel Adams teaches how to overcome that, by placing white and black in certain zones. Worth googling it.
Good luck.

 

[koraks]

a white coffee foam cup

What's the appropriate wall thickness and opacity of the foam for this?

 

[IpseLux]

What's the appropriate wall thickness and opacity of the foam for this?

Give it a try. Consider it a working hypothesis….
You can always adjust exposure if it’s a bit off, with whatever cup you use.

 

[koraks]

Give it a try.

Kind a hard, foam cups are virtually nonexistent here.
The gist of my response of course is that a makeshift diffuser in principle will work fine, but it generally needs to be empirically determined how much light it blocks in order to determine a compensation factor for use.

Carrying a hand-held meter with a dome also works quite well; incident meters are quite affordable, esp. second hand ones. They're also often smaller and less fragile than a foam coffee cup.

 

[IpseLux]

Kind a hard, foam cups are virtually nonexistent here.
The gist of my response of course is that a makeshift diffuser in principle will work fine, but it generally needs to be empirically determined how much light it blocks in order to determine a compensation factor for use.

Carrying a hand-held meter with a dome also works quite well; incident meters are quite affordable, esp. second hand ones. They're also often smaller and less fragile than a foam coffee cup.

I use a little Sekonic digital. Some of my older cameras don’t light meter.
The cup works relatively well, white, translucent, illuminated by the available light, and the meter will render it middle great. It works in a pinch.
I always wondered why cameras did not include a white, lens cap to serve as impromptu incident meters, or had an incident meter function? I suppose as you state, light meters are readily available and relatively affordable.
Here in the American South, foam cups, red plastic cups, and all matters of environmental waste are not regulated much. States like California are much more cautious.
Kind regards K

 

[RalphLambrecht]

My first real 35mm camera was a Pentax SP 500. Eventually added a Pentax ES, screw mount cameras. I made great pictures with these cameras. Just matched up the needle, or with the ES set to auto. I had to be cool so I traded these in for Nikon F2S, this camera is more center weighted. I bet my color slide success rate dropped by 75%. I rarely use in camera meters unless it's something like a Nikon F5 with matrix metering on auto pilot. I'm better off guessing. I'm actually pretty good at guessing. I use an incident meter. Pretty boring approach

prior to matrix metering, center-weighted metering was the best there was. By adding personal experience, both can be unbeatable.

 

[Andrew O'Neill]

prior to matrix metering, center-weighted metering was the best there was. By adding personal experience, both can be unbeatable.

I agree up to a point. A hand-held spot metre is unbeatable in some situations.

 

[Paul Howell]

Regardless of the metering method, all meters can provide an inaccurate reading for a given lighting situation. Nikon matrix metering was optimized for transparencies, with the move to digital I guess the metering was tweaked. For B&W and color negative Sigma version of martix was as good if not better than Nikon or Minolta. A good TTL average meter with a 18% gray card works as well as an incident meter, the advatnage of a inciedent meter is not having to carry a card.

 

[mshchem]

I agree up to a point. A hand-held spot metre is unbeatable in some situations.

Especially when you are shooting over differently illuminated subject zones, like gnarly old trees in front of haunted abandoned provincial hospitals.

And making amazing alt process prints

 

[Kilgallb]

The problem you describe is why there is a rubber cover for the eye piece attached to the carry strap. I still have mine.

Perhaps something is wrong with mine but the meter seems to change if light enters from behind the camera through the eyepiece. The problem is corrected when my eye is close to the eyepiece shielding it from light. Anyone else notice this?

 

[wiltw]

Perhaps something is wrong with mine but the meter seems to change if light enters from behind the camera through the eyepiece. The problem is corrected when my eye is close to the eyepiece shielding it from light. Anyone else notice this?

You describe a not-uncommon weakness of TTL metering systems using photosensors up in the pentaprism area, and it is not such an issue for metering systems whose photosensor is on the back of a semsilvered reflex mirror (Topcon), or in the base of the camera (Canon dSLR), or reading the film surface or front surface of the shutter curtain (Olympus). In fact, some designs incorporate an eeypiece blind/curtain to overcome the issue during tripod-mounted usage.

 

[Kodachromeguy]

Some companies formerly sold a translucent dome to attach to your lens to simulate an incident dome. I did not see any for sale now, but possibly I used the wrong search terms.

I have used the broad area Spotmatic light meter totally successfully. I always point it slightly down to measure a mixture of ground material, grass, pavement, whatever. But obviously, if the subject is dark, like a facade under a porch, I open a stop. Formerly, I measured the exposure of a scene with the Rolleiflex selenium meter the same way: point downwards.

 

[snusmumriken]

You describe a not-uncommon weakness of TTL metering systems using photosensors up in the pentaprism area, and it is not such an issue for metering systems whose photosensor is on the back of a semsilvered reflex mirror (Topcon), or in the base of the camera (Canon dSLR), or reading the film surface or front surface of the shutter curtain (Olympus). In fact, some designs incorporate an eeypiece blind/curtain to overcome the issue during tripod-mounted usage.

With the Nikon FE I used to have, I found I could hang the black cap from a film canister over the eyepiece when the camera was tripod-mounted (or tape it there if windy!). That was handy for photographing wildlife from a hide, for instance. Otherwise, unless the light was changeable, the better alternative was to manually set whatever exposure was indicated when my eye was present at the viewfinder.

 

[warden]

With the Nikon FE I used to have, I found I could hang the black cap from a film canister over the eyepiece when the camera was tripod-mounted (or tape it there if windy!). That was handy for photographing wildlife from a hide, for instance. Otherwise, unless the light was changeable, the better alternative was to manually set whatever exposure was indicated when my eye was present at the viewfinder.

Light entering the camera from the eyepiece is a minor issue for me because I only care what the meter says when my eye is blocking the light anyway, but still, I’m now wondering if the battery drains faster due to light entering the rear of the camera and keeping the meter awake while the lens cap is on. No big deal, just a curiosity.