Lens recommendation for Nikon D7100

[Mike Buckley]

There's absolutely no reason to feel embarrassed about not knowing how to use a histogram to your advantage, Dave. It is important though to understand that if you want to take the quality of your captures to a level that is anything higher than leaving your digital camera on auto everything and taking sharply focused pictures, the single most important thing you can do in my mind is to learn how to use the histogram. Also, when you do your post-processing, you should always keep your eye on the histogram.

I recommend checking out the CiC tutorial about using the histogram and that you start a new thread to get back to us if there is anything that seems confusing. The recommendation to start a new thread is mostly so it attracts others who will also benefit from the discussion, as one heck of a lot of people don't understand the importance of using a histogram and don't know how to use it.

[Dave A]

I recommend checking out the CiC tutorial about using the histogram and that you start a new thread to get back to us if there is anything that seems confusing. The recommendation to start a new thread is mostly so it attracts others who will also benefit from the discussion, as one heck of a lot of people don't understand the importance of using a histogram and don't know how to use it.

Thanks Mike, I am going to make it my mission to fulling understand how to interpret and understand the histogram now that I know just how important it is to achieve better results.

Dave

[Mike Buckley]

I would like to point out the evolution of the content in this thread as a learning experience about how to effectively identify problems and their solutions.

The thread began with Dave thinking that the solution to his problem regarding noise was a better lens. Then some of us pointed out the various causes of noise and that even if his equipment is causing it, the likely solution to his problem would be a better sensor, not a better lens. Now that we realize that he understandably has not yet learned how to use the histogram, we know that it's too early to know whether his equipment is causing him problems. Indeed, the only way that can be determined is if he first consistently properly uses the histogram, which will immediately produce properly exposed images.

Once we're confident that he is producing properly exposed images, his problem might go away. If it doesn't, he can then learn how to mitigate the effects of noise using software, perhaps software that he already uses. There will be compromises when using software to eliminate noise or to reduce it to an acceptable level. If mastering the software still leaves him experiencing an amount of noise that is more than he is willing to tolerate, his solution then will be to use a better sensor (again, not a better lens).

Dave is being very open-minded about identifying the cause of his problems and how to resolve them. If he had simply gone down his initially intended path of buying a new lens to solve his problem, it's likely that he would have been very disappointed.

[William W]

Originally Posted by William W
The (major) reason that the image is underexposed is likely the METERING MODE in combination with the AUTOMATIC CAMERA MODE which were selected.

I have pretty much decided that the reason an image presented here such as this one that is not an action shot and probably did not have rapidly changing light is underexposed is because the photographer either didn't check the histogram immediately after taking it or doesn't understand how to use the histogram. The former situation is really easy to fix and the latter situation is somewhat easy to fix. I'm not trying to be glib; I'm trying to be helpful.

Ah-ha! Different emphasis! I love it!

How to use another facility of the camera (the Histogram) to double check the TTL Metering System: and also how to use the Histogram to learn about what the TTL Meter is reporting, in different Lighting Scenarios.

Fantastic!

WW

[William W]

Dave,

As Mike has stated - I suggest that you do not feel embarrassed about anything; au contraire, you should feel proud that you are willing to ask questions and proud that you are open minded so to as consider all the answers.

WW

PS I didn't notice any adverse noise in that image I commented on - but I wasn't looking for noise and it is a low re-image.

The point is, whatever the level of noise there is in the full res. image, it almost certainly would have been less noise if the image were correctly exposed.

This PS is not added for your (Dave's) benefit as I am very confident you appreciate that fact - but for the benefit of the general readership of this thread.

[AlwaysOnAuto]

I'll be watching this and the other thread with interest as I don't really understand histograms either. I read the tutorial on them here and came away wondering, OK, so I know what the technical aspect is of how/what the histogram represents, but, it didn't really tell me what one should look like in the 'ideal' instance.
Is it a straight line?

[dem]

OK, so I know what the technical aspect is of how/what the histogram represents, but, it didn't really tell me what one should look like in the 'ideal' instance.
Is it a straight line?

No, the shape of the histogram cannot be controlled during the image capture. It is what it is, you can only shift it to the left or to the right by varying exposure. A uniformly grey scene will have a histogram with a single hump in the middle, while a picture of a chess board will have two peaks at each side. There is no "ideal shape" of a histogram. Just keep an eye on both sides of the scale.

If the histgram touches the right hand side of the scale, you have got completely white pixels = "blown highlights". Equally if it touches the left hand side of the scale, then you have got completely black pixels = no detail in the shadows.

For a lower signal to noise ratio it is better to "expose to the right" (ETTR) so the histogram is as far to the right as possible (without blowing the highlights) and then bring the exposure down in post processing if needed. The down side of doing ETTR is of course a slower shutter speed required (it increases the chances of motion blur and camera shake).

[Mike Buckley]

I know what the technical aspect is of how/what the histogram represents, but, it didn't really tell me what one should look like in the 'ideal' instance.

The ideal histogram varies from scene to scene. The only detail of the ideal histogram that will be the same for every scene is that the data displayed in the histogram will represent the scene itself. To use an extreme example only for the sake of simplicity, consider what the histogram of a properly exposed photo of a black wall should look like. Now consider the same thing for a white wall.

By the way, once you get the hang of this, you're gonna have to contact one of the admins so you can change your screen name. That's because then it will no longer apply. Perhaps change it then to RarelyOnAuto.

[AlwaysOnAuto]

Thanks Dem. I'll have to find some checker boards to take pictures of.

Mike, I haven't been 'On Auto' since I got my Nex-7 and started shooting my legacy glass using the old 'match needle' formula of 'spin the dials till the camera says it's OK to shoot'. I'm finding I know just enough about the relationships of shutter speed, aperture and ISO to have some fun with it. Thanks for the laugh though, I needed it today.

[Mike Buckley]

Alan,

Though Dem's description of the histogram displayed when photographing a chessboard is generally correct, the histogram will be displayed as described only if you light the board nearly perfectly evenly. Otherwise, the histogram will indicate lots of mid tones by displaying data in the middle of the histogram. Even if you light it perfectly evenly, the histogram will only be approximately as described. That's because the bright squares probably aren't so bright that the data representing their tonality will be displayed on the far right side, only near it.

I think the easiest exercise is to use a black piece of material that has as little shine as possible that fills the frame. Expose it properly (probably at about -2 exposure compensation). Then do the same with a white piece of material that has as little shine as possible. Expose it properly (probably at exposure compensation of about +1 1/2 to +2).

Finding something on the Internet that displays a photo that contains only the very darkest and the very brightest tones and its histogram would surely suffice.

[William W]

. . . I am going to make it my mission to fulling understand how to interpret and understand the histogram now that I know just how important it is to achieve better results.

-And-

. . . I don't really understand histograms either. I read the tutorial on them here and came away wondering, OK, so I know what the technical aspect is of how/what the histogram represents, but, it didn't really tell me what one should look like in the 'ideal' instance. Is it a straight line?

As Dem and Mike have already commented there is no ‘ideal’ instance and the “ideal” Histogram will vary from scene to Scene (or I would more pedantically write, vary from Lighting Scenario to Lighting Scenario).

Also the Histogram will be dependent upon the EXPOSURE the Photographer chooses for that Lighting Scenario.

It is important point to understand the different meanings of the word ‘exposure’ – in this situation the word ‘exposure’ means the INITIAL EXPOSURE of the negative or image file. We need to:

> avoid the confusion which I believe happened in Post #35

> understand that the Photographer always has a choice as to what should be the exposure at the time of the Shutter Release

This is different to the meaning of the word ‘exposure’ when we discuss the exposure of different areas of the FINAL IMAGE. Those exposures will often be a result of the (initial) exposure of the file or negative - AND - the exposure changes made during the tweaking the Image File in Digital Post Production or the exposures (dodging and burning mainly) of the PRINT during tweaking the Printing Process of the Negative in the enlarger if we are using Film.

***

A quick word or one hundred about ETTR:

On this point of choice of the (initial) Exposure of the Image File or the Negative, it is usual that in most circumstances the Photographer will want to make use of as much of the Dynamic Range of the medium, be it a Digital Sensor or Film: and similarly and also to achieve that first aim it is also most usual that the Photographer will usually want to push the highlights to near the limit of the Sensor or Film - hence for Digital Capture we often hear spoken or see written ETTR Expose To The Right – that is to push the (initial) Exposure as far to the right of the Histogram as possible.

(This is exactly what Post#34 was commenting upon and why it was concluded that the sample image was underexposed - because the histogram was not pushed as far as possible to the right)

***

Learning about Histograms:

There are a few ways to lean about how to expect your Histogram to appear, and one way which works often for my Students is simply to investigate the Histograms of different shots that you make – whilst simultaneously viewing the image - and that way one begins to assimilate “oh this shape histogram goes with that type of lighting scenario”.

As a beginning point for using the Histogram as a tool for checking you (initial) exposure, I suggest that you get the Histogram so that it is Expose To The Right (ETTR), no matter what is the shape that particular Histogram.

Expose To The Right (ETTR) is primarily to allow the widest dynamic range possible in the digital capture and thus to allow the least amount of noise in the shadow areas. BUT – there will be some times when the Photographer will purposely choose NOT to Expose To The Right, and I choose not to comment on that topic, in this commentary.

One can still 'Expose To The Right', for a low key image, simply that means (at shutter release) the brightest highlight of the scene is at the extreme right of the histogram. There might not be a large area of the scene which is bright, but what is bright, is pushed as far to the right as possible, without overexposing or 'blowing out' (i.e. unrecoverable data).

***

Another Tool – the Blinkies:

Another useful tool that many Digital Cameras have is: ‘Blinkies’. I use Blinkies more than I use the Histogram – in fact I don’t use the Histogram that much at all because when I do review a shot I find the ‘Blinkies’ a quicker much methodology of review.

***

How is the Histogram (and Blinkies) made? this is important to understand

In fact this is a crucial consideration if you want to use the In-camera Histogram and/or the In-camera Blinkies as a tool(s) for Exposure or Exposure Review:

In (nearly?) ALL digital cameras the HISTOGRAM and the BLINKIES are a function of the IN-CAMERA JPEG POST PROCESSING.

This means that the settings which ‘enhance’ the JPEG image file (in later model Canon DSLRs commonly referred to as “Picture Styles” and in Nikon I think referred to as “Picture Controls”) have a direct affect on the HISTOGRAM and the BLINKIES and, important to note, because I think in many cameras this is fact, even if the JPEG capture is not selected.

I have a few (I started with five I think and now I only use two, and occasionally a third) particular “Picture Styles” which I can set (for different lighting scenarios) to best interpret the information the Histogram and the Blinkies are providing for me.

At the very least, if you want to use the Histogram and/or the Blinkies as a tool for Exposure, I think it is very wise to set and then never move those Picture Styles / Picture Controls so there is (at the very least) a stable and a set continuity of a reference point.

*

Another Important Note on the same topic:

BECAUSE the In-camera JPEG Post Production affects the Histogram and the Blinkie - this means that – an Histogram or a Blinkie might indicate that there are “blown-out” highlights – but that might be incorrect information, in so far as those highlights might be recoverable in the Post Production from the raw file.

This is why I will often be pedantic about separating/distinguishing in my commentaries what appears to be ‘blown out’ and what is actually ‘recoverable’.

***

Some samples:

To get you started, here are some examples of Histograms of some typical Lighting Scenarios that we encounter:

Image #1

Here the Histogram crawls up the RH Side and the little spike is representative of blown windows and doors in the background – I’d be looking here that there is NOT a big bunching up at the RH side because about half of the scene is in ‘shadow’ by comparison – the point is the Photographer was looking for detail of the interior of the Mosque so this Histogram looks "OK" to me, even though it is indicating blown highlights (which there are in the doorways for example):



***

Image #2

In this shot the Photographer wanted nothing blown out and the whole image erring toward Low Key (emphasising a pensive and reflective tone) the camera angle was manoeuvred such that the lighting in the BACKGROUND was a close to black-black as possible – which was nearly achieved.

So IF this histogram were crawling up the RH side or (as in the above image#1) there were a nice “bump” in the middle of the Histogram, we would expect that the Photographer’s Vision might not be realized:



***

Image #3

On the other hand for this image the Photographer’s vision was an High Key Image (rendering ‘happiness’ and ‘joy’) – so this Histogram is somewhat like a mirror, of the Histogram in Image #2:



***

Image #4

This image, like Image #3 is also made in soft ambient room light - BUT – the Photographer’s vision was to allow the Subjects to speak for themselves and to carry the weight of the that load by their own personalities and the colours.

So – the camera viewpoint was chosen differently to that in Image #3.

Whilst there a little MODELLING on the Subject’s faces (especially OoF Woman), the lighting is reasonably flat and even across the scene – EXCEPT for the large window in the background.

So we would expect a “flat line” Histogram here – and we got it.

If we had a big bunch in the LH side we’d be thinking “underexpose” – if we got a big bunching at the RH side we’d be thinking “overexposure”:



***

Image #5

The Lighting Scenario in this image is similar to the Lighting Scenario in Dave's first sample image (the pathway through trees) - though I would bet a Mars Bars that many would not see it that way on first glance.

But if we study the image closely there is basically a large amount of highlight area and a large amount of shadow area and not that much in the mid-range tones - hence the similar "Two Humps" Histogram - but note that the exposure in this image is ETTR.



***

In concert with learning about Histograms, I advise that it would be a good idea to understand how ALL the Metering Modes on your camera work.

WW

All Images © AJ Group Pty Ltd Aust 1996~2016 WMW 1965~1996

[tclune]

A quick word or one hundred about ETTR:

On this point of choice of the (initial) Exposure of the Image File or the Negative, it is usual that in most circumstances the Photographer will want to make use of as much of the Dynamic Range of the medium, be it a Digital Sensor or Film: and similarly and also to achieve that first aim it is also most usual that the Photographer will usually want to push the highlights to near the limit of the Sensor or Film - hence for Digital Capture we often hear spoken or see written ETTR Expose To The Right – that is to push the (initial) Exposure as far to the right of the Histogram as possible.

(This is exactly what Post#34 was commenting upon and why it was concluded that the sample image was underexposed - because the histogram was not pushed as far as possible to the right)

I wonder whether this idea is at least in part equipment-centric. ETTR seems to be widely used by Canon shooters and much more rarely among Nikon shooters. Partly, I presume this reflects the more limited dynamic range of Canons. But, for many Nikon shooters, there is a trap lurking in ETTR. My camera, for example, is the D5200. While the sensor has 14 bits of dynamic range, the raw data is always lossy compressed. Nikon uses a decimation routine that aggressively prunes the brighter levels of the image. There is a look-up table that maps the original 16,384 levels per channel into 3073 values before the data is stored at all. The first level that is combined with its neighbor is 450, so the very dark values are saved faithfully. As you get brighter, more and more levels are combined into one.

If you stop and think about the ramifications of that, ETTR and then down-shifting the image would be a really bad idea on a camera that has excellent noise properties to begin with. I should mention that the Nikon D7xxx and (I believe) all the ff cameras have the option to save their raw data fully uncompressed. But the lower end Nikons do not (and the D3xxx has something like 750 values in its look-up table, so the compression is draconian.) All these compressions may be used in the more expensive Nikons, so if someone is trying to save space on their cards or looking for smaller files on their computer, they may be subjecting themselves to the same issue out of ignorance or indifference.

FWIW

ETA: Here is a useful table of Nikon model and compression values http://www.photonstophotos.net/Nikon...ompression.htm

[William W]

I don't have any data on whether Nikon users practice ETTR less frequently than Canon users: but that's an interesting thought.

Whatever the DR of the camera: and if it is a Nikon Camera with a large DR, one would expect that good design would allow for all of that Dynamic Range to be used at the capture of an image.

In the case of the sample image of the path, there was much of that DR unused, which, as I understand, means that the deep shadow area was more noisy than it otherwise would have been if, for example, the initial exposure were 2 stops more.

The subsequent improvement in the noise of those deep shadow areas would surely be irrespective of how good (or bad) the general noise stats are of the camera - it would still be an improvement in the noise in those shadow areas, would it not?

WW

[tclune]

The subsequent improvement in the noise of those deep shadow areas would surely be irrespective of how good (or bad) the general noise stats are of the camera - it would still be an improvement in the noise in those shadow areas, would it not?

But, if the noise were replaced with posterization because the shift made the decimation visible, it would clearly not be a win. As a matter of fact, Mr. Claff (in the page I linked to above) notes exactly that danger with the decimation, which he finds otherwise imperceptible. FWIW

[William W]

I read the paper and the conclusion. I found it both useful and also interesting, thank you.

In the conclusion the author offers a choice between selecting compressed or non compressed predicated mainly upon on the lighting scenario and subsequently minorly upon the intended post production - viz:

"However, if you will be post processing images with a great deal of highlight content then you may want to choose uncompressed NEF rather than compressed NEF to reduce any chance of posterization."

and therefore, because there is that choice of selecting the type of the raw capture, I am failing to understand the message, suggestion, or implication of the recent posts - the broad take-away message which occurs to me as being: it is generally a good idea to NOT use the full dynamic of the camera, be it a Nikon Camera or a Canon Camera or whatever camera.

On the other hand maybe I am missing and / or misunderstanding the intended general message - and if so could you please clarify it for me?

Thanks,

WW

[tclune]

because there is that choice of selecting the type of the raw capture, I am failing to understand the message, suggestion, or implication of the recent posts - the broad take-away message which occurs to me as being: it is generally a good idea to NOT use the full dynamic of the camera, be it a Nikon Camera or a Canon Camera or whatever camera.

On the other hand maybe I am missing and / or misunderstanding the intended general message - and if so could you please clarify it for me?

The point you've not noticed is that not all Nikons have the option (the table entries indicate whether the given camera supports onlylossy compressed or both lossy compressed and lossless raw data types). All D3xxx and D5xxx cameras ONLY have lossy compressed raw data. I used to own a D5000, which had only 12-bit data and used 769 values in its lossy compression (which is the same approach that is used with the D3xxx cameras). I both found visible noise, even at base ISO, and posterization at times when I processed images. In all honesty, I have not noticed either problem with the D5200, which has 14-bit data, better dynamic range, and more values in its decimation table. But once bitten, twice shy...

[William W]

OK get it now.

Thanks for explaining.

I guess, depending upon the type of photography that one does -and- one's tolerance for noise there would be very good reasons to trade up from D3xxx and D5xxx series cameras.

On the other hand even though you've been 'bitten' once with intolerable noise at base ISO using a D5000 Nikon, and considering your observations of your current camera, I don't necessarily see that you should 'be shy' and avoid the general usage of ETTR with your current D5200 Camera.

Accordingly I would encourage you to try it.

WW

[AlwaysOnAuto]

Bill and Tom, very interesting discussion, thanks for sharing it with this obviously rank amateur.
Being the kind of guy that I am, I set my D7k up to record 14-bit RAW and the highest quality jpeg compression available. In fact I've set all my digitals up to record the highest amount of data possible as I'm the type that believes more is better than not enough. I don't shoot enough to worry about storage space. Maybe I should change that...

Mike - I think I might be wearing the perfect item to take a picture of for a histogram study. A non-shiney, black and white checkered flannel shirt! I'll give it a shot and take a look at the histogram.

OK, you guys can continue your discussion, I'll keep quiet now. Thanks.

[AlwaysOnAuto]

So here's what I was able to come up with real quick like this morning.

[Mike Buckley]

The three humps in the histogram clearly depict the corresponding dark, medium and bright tones in the scene. That explains why that histogram appears to me to be an ideal histogram for that particular scene.