Help Me Understand Dynamic Range and the K-5

<p>I was watching a Creativelive.com class and the instructor was talking about off camera lighting and making sure that all of the light was within 3 stops of the main or key light. This was to ensure that all of the light was within the capture range of the camera. Now 3 stops was the number that he used because he assumed that his camera had a 5 stop total tonal range. He then postulated that he'd lose the top and bottom of the range, as it could not be pushed any further. </p>

<p>Being a bit of a nerd, I then looked up the K-5's dynamic range (I know its great, but I wanted specifics). So at the basic settings using the extended dynamic range of 80 iso on the highest quality, this website indicated that the K-5 raws had a 7+ dynamic range with the least amount of noise. At the lowest quality that number jumped up to 14 stops. </p>

<p>Working with the 7 stops, and discarding the top and bottom of the range, that leaves 5 stops. </p>

<p>So here are my questions:</p>

<p>1. The dynamic range of the JPG preview is different than what is in the raw file. Is that correct?</p>

<p>2. If so, then what is the best setting for the K-5 jpg file to capture the most dynamic range (assuming that the highlight and shadow protection is turned off). This is important because I would be trying to make sure that the data that is captured is within the dynamic range (+/- 2 from the key exposure) and not blowing out. But if this changes between the capture jpg preview, then how will I know that I have blown out the image or got enough shadow detail? Do I have to adjust for this based on each jpg style?</p>

<p>2b. Is it true that the histogram is also based on the jpg file and not the raw file? If so, then the histogram also does not give a true representation since the jpg will blow out much sooner than the raw file. </p>

<p>3. Assuming 18% gray is the middle stop, does that mean that I have 2 stops below and 2 stops above in the dynamic range? Again, measuring with an external light meter. </p>

<p>4. Or does the dynamic range refer to just the amount of light captured at all times in the raw file? So if the scene uses matrix metering, then k-5 will evaluate to try to fit the entire scene with the dynamic range as best it can? If so, does the K-5 look for middle gray or does it do its calculations based on the scenes total tonal range?</p>

<p>Thanks!</p>

 

<p>1. Yes and this is not specific to K-5. In general, the DR of a JPG is resulting from applying a processing curve to the RAW data and this can result in clipping information. That is why it is recommended to shoot RAW and to process the RAW data later.<br>

2. Depends on the scene and on the metering of that scene and on the camera settings. There is no easy answer here - you'll have to figure it out based on your specific use.<br>

2b. Yes.<br>

3. Question does not make sense to me.<br>

4. DR basically refers to the range that the camera can possibly capture. This is just a simple definition. You may have to underexpose or overexpose according to the metering, to fit a scene into the DR of a camera. Metering is not perfect and the best thing is to learn how it fails and dial in compensation. Or if you can afford it, shoot, examine result, and adjust. Like for artillery fire. In theory, everyone can figure out the trajectory of a shell, but in practice it's just trial and error correction.</p>

<p>I don't think my response will help with your inquiry, but reading your situation and your questions leads me to respond by offering to simply expose so the histogram curve is mainly right of center without clipping, which should offer the greatest editing latitude once it's in the computer editing program. And capturing and working with RAW files (question 2) will offer greater latitude over JPEG files. Shadows will show detail (without going totally black) and highlights won't be clipped. Yes, there are unique situation where this won't be the case (the dynamic range is wider than can be caught with a single exposure), but your situation describes what sounds like controlled studio lighting. (If it's for outside lighting conditions, you're likely adding fill lighting to lighten dark shadows of an up-close subject, thus narrowing the dynamic range.) And unless there is movement in what your are shooting (i.e. a model that is constantly moving, wind blowing hair, etc), you could also set up to bracket the shots with 1/3 or 1/2 stop exposure compensation and chose the best starting image once in the computer.</p>

<p>I'm really out of the league with this thought, but I'm wondering if those class instructions are holdovers from film days (but some films had quite wide DR, didn't they?), or at least previous generations of digital when sensors and their processors didn't offer the dynamic range digital cameras offer today?</p>

Hi Jamal: Both Laurantiu an Steve gave good answers. I can add this: Dynamic range (DR) as you

know is the range of tones that is recorded so that there is detail in the darkest and the brightest area

of the image. In Raw file each type of sensor has its specific DR. You do not see this range as you

cannot display Raw file as it has to be processed for display. Commonly it is processed to be

displayed as JPG. You do not capture all the DR of RAW on conversion. How much DR you do not

capture depends on what you use for converting from RAW to JPG. Although most converters

probably are not that different.

 

As for as 18% grey exposure is concerned, most DSLR cameras give a choice of how to meter

exposure. There are settings such as evaluative, average, spot, etc. You will get different tones of the

image with these different metering settings.

 

I believe that the instructor was referring to light settings for pleasing portraits and not for all kinds of

images. Hope this helps. Sandy

<p>Jemal, to amplify Laurentiu's comment: You are throwing away lots of DR if you are shooting jpgs. It's not hard at all to process RAW files, and it is really fun to pull color and detail out of murky darkness. I'm processing a shoot of an early-morning event shot in low bright sun, and the contrast was so extreme that with jpgs I would have very few keepers. With the K-5's RAW files, though, the exposures are no problem.</p>

<p>Rick</p>

<p>Thanks everyone for your responses. Just a few clarifications. <br>

1. I shoot RAW, not jpg. But the problem is that the camera (all cameras) convert the RAW file into a jpg preview (As I understand it). So I would a. like to get this preview to be as accurate as possible in terms of dynamic range, and b. to set up a jpg that is close to how I would process the final RAW file for those occasions when I don't want to be bothered with a lot of post processing. <br>

2. As to question 3, all camera's really only see 18% gray. I guess my question is where does this 18% gray land in the dynamic range. Is it in the middle of the range? (Ignore the fact that whatever I meter off of will be treated as being 18% gray). So for instance, if the K-5 has 11 stop DR, would 18% gray be 5 stops from the top of the range and 5 stops from the bottom of the range (I.e., in the middle of the DR)? Or is 18% gray two stops from the bottom? Hopefully that clarifies things some what.<br>

3. I also process all of my RAW files in either LR 3 or PS CS4.</p>

<p>I don't know, Jemal. I'm sort of thinking that the instructor made things unnecessarily complex with his statements.</p>

<p>Thinking out loud via my keyboard...</p>

<p>--How about this- take some sample photos, a small variety of subjects and lighting conditions. Put the RAW images on your computer monitor in LR or CS, and then compare the unedited RAW images on the computer monitor to the images as they appear on the camera's display monitor. The camera operating manual, pages 297-298, describe how to adjust screen brightness and the color of the camera's rear monitor. This might be as close as you can get to matching how things look at the time of the shot compared to how they would look (in general) on your computer monitor. This won't affect the camera histogram at all (which is representing the camera generated JPEG version of your photos, which I would hope isn't far off from what a histogram of the RAW file would look like) but it might help your camera and computer monitors be a bit more in agreement with each other when using your eyes to subjectively judge your captured exposure on the camera screen. (Balancing out the "looks bright/dark on the camera, but looks brighter/darker on the computer".) The computer monitor will show more dynamic range than the camera monitor is capable of (I'm assuming), so maybe trying to make the camera monitor emulate the computer monitor more closely is a work-around to your inquiry. Maybe? Especially in concert with referencing the histogram. This may be as close as you can get to making the camera display JPEG and the RAW image on your computer screen look similar to each other, much less an edited RAW image file. I don't think you can get any closer than that- your camera isn't as powerful/flexible as computer-based editing software and a calibrated computer monitor. Any idea how old that online tutorial was? If it was made even a couple camera sensor generations ago, his comments may no longer have the relevance they once did.</p>

<p>--His comments may mean more with spot or center weighted exposure metering, but may be of less value with evaluative metering when the camera takes the entire scene into account.</p>

<p>--Me thinks the instructor was going a bit overboard. The current generation camera sensors (K-5 and newer) and on-board processing, and the computer software we are now able to utilize, are so powerful that we do have the ability to make up for things we could not just a few years ago (shy of utilizing HDR to conquer DR issues). A year ago I used my K-5 to shoot a downtown skyline that was nearly in silhouette due to catching the setting sun and illuminated clouds in the shot. About as wide a DR possible. I captured the image as a RAW, and rescued it with CS5/Adobe Camera Raw just playing with the sliders.</p>

<p>--Another thing that may be helpful is test shots using an 18% gray card, which will be beneficial in post processing.<br /> Other thoughts... Some help here? http://en.wikipedia.org/wiki/Middle_gray<br /> http://en.wikipedia.org/wiki/Gray_card<br /> http://www.wikihow.com/Use-an-18%25-Gray-Card<br /> My thoughts here are that if you have some reference in the test photo where the middle of the light scale is at (as found on the gray card), it may help you judge the lighting, the dynamic range, across the rest of the image once the image in on your computer. The first of these links does state that 18% gray is just under the half way point between black and white, 46.6% on the brightness scale for the sRBG color scale, and 119/119/119 for a 24-bit RGB scale (assuming the 0/0/0=black, 256/256/256=white brightness range, 119 being 46.6% of 256).</p>

<p>Hope this helps.</p>

<p>Jamal, on my istD and K10D there was a considerable difference between the histograms I saw on the back of the camera and those I saw in Lightroom. Following the advice in a PDML thread, I adjusted the brightness and contrast of the camera display until they matched (more or less).</p>

<p>Steve - Thanks for the information re the Gray card. That answers my question that I have roughly two stops above and two stops below middle gray to work with. I hear middle gray is zone V on the zone system.</p>

<p>Everyone - I have already adjusted the brightness of the LCD on the back of the camera down because I kept getting dark exposures which looked great on the back of the camera. That is an excellent tip.</p>

<p>However, it does not resolve my current problem which both Steve T. and Rick W. recognize. The fact that the histogram on the back of the camera is tied, not to the RAW image, but the jpg. Since the jpg throws away a ton of data that it does not use, it seems hard to rely on it for determining when an image will clip, particularly in a situation where you are trying to push the light as far right as possible. I have noticed for different picture styles (i.e., bright vs. natural, etc.) that the clipping for highlights and shadows occur differently even if the camera is on a tripod and not moved. I would like to know where that clipping actually occurs in the RAW file so that I do not have to guestimate or hope that the shot is within the dynamic range when I finally get back to a computer.</p>

 

<p>Jamal,<br>

I think you are seeking a formula or a recipe. An exact one doesn't exist because there are so many variables. That's where your human judgement and experience is most excellent.<br>

I feel that your instructor also infused a cookbook approach to this, rather than an approach that focuses on the art of the image including composition. In a studio where you can control the light so well, I believe that proper exposure should be a non-issue. DR takes care of itself, especially with a camera like the K-5.<br>

To me composition is the major issue--though white balance has been my personal challenge when shooting a group composed of people of diverse ethnic backgrounds (the word "race" seems obsolete these days).<br>

To try to sync your histogram with a RAW output, you should start by testing the full range of in-camera contrast settings. Aim the camera at some kind of tone step pattern used to measure DR. I'd also advise setting the camera to Adobe RGB as the color space. It would help to get the white balance accurate in camera as well.<br>

All this seems too hard to me. I really only use the LCD to verify the framing of a shot and for LiveView closeup focusing. My energy has gone into learning software tools that will sync the RAW data with my personal vision of what the shot should be.<br>

I'd also recommend you upgrade to LR4.x. Its handling of highlights in particular is far superior to older versions.</p>

<p>ME</p>

 

<p>Hi Michael:</p>

<p>Thanks for your input. Actually, the instructor did an excellent job of explaining why you need to control the light, particularly the main or key light. The issue that he explained is that you want to capture as much detail as possible, even in specular highlights. What I found revolutionary (is anything ever) and a point that I had heard before, was that instead of measuring the overall scene, one needs to measure the light for the key object of the story and build your lighting around that point of interest/item. In the case of his example that was the highlight along the body lines of a car. He then set the rest of the lights to be within 2 or 3 stops of that light down. His shot was excellent and well thought out.</p>

<p>I had heard similar for an outdoor photo shoot of a model where the photographer was going to key in on the bright rim light along the left side of her face. He wanted that rim to be no more than 1 or 2 stops over her face, so that the area lit would not blow out, but would in fact contain data. </p>

<p>Shooting different skin tones and white balance. As you know, I'm black and shoot mostly black folk. i've struggled with trying to get skin tones right, and have not found a lot of the action sets to work particularly well with black skin. The two things that have helped me most in learning is this: 1. Get a light meter. The meter does not care about the skin tone of the subject. It only measures the light at 18% gray. The meter in your camera changes because it is measuring the amount of reflected light which is different based on the color of the object it measures and forces to be 18% gray. 2. White/light skin shape is defined by the shadows. Dark skin shape is defined by the highlights. Both great advice. </p>

<p>White Balance: If you shoot raw, white balance doesn't really matter, you adjust it to taste in LR or Photoshop when you develop the raw file. I have been playing with Pentax's CTE mode which is describe as "anti-white balance". It takes the colors of the scene and lets them dominate the color balance. For instance, if you are shooting in a field of green grass and brown and green tree leaves, then your photo should have a greenish color cast. So far, I am loving the CTE processing. Try that for your next white balance. The other thing I learned was that you can get a white balance lens cap, put it over your lens and then shoot into your main light. You use that photo to set white balance. The other thing is to get a passport color checker. (I can't afford that).</p>

<p>I use the LCD only to check framing, and Liveview, and to check the histogram. The most important is using the histogram to check highlights. At least for me. i have learned not to rely on the LCD image for anything other than framing, focusing and the histogram.</p>

<p>LR4.x and upgraded photoshop are on my to buy list when means permit.</p>

<p>The suggestion for +/- 3 isn't so bad for a rule of thumb. You'll probably clip to white close to +3 (if not sooner) and will be close to black at -3. Assume that 0EV is that 18% gray metering. With RAW processing you can probably recover a bit more shadow detail beyond -3, especially with the K-5's files.</p>

<p>If you want to see better what you've captured on the in-camera JPEGs, you might consider reducing the JPEG contrast settings.</p>

<p>Jemal,</p>

<p>From your reply to M.E. on this page, I understand specifically what you are getting at.</p>

<p>Your instructor was teaching you a zone concept of metering where you basically judge the tonality of the scene,determine what is key (whether it be highlight, shadow or mid tone) and then base your 3 stops around that.</p>

<p>I used this exact same system when I was learning photography, and I still do use it today. Only, I really had to think about it back then. This method, however, will always ensure you have your key within a proper exposure and since there is only so much you can do with a single exposure the stuff beyond the range of the key end up off the exposure as specular highlights or black shadows (clipped).</p>

<p>However, I have an issue with your rationalization of how the in camera light meter works.</p>

<p>Nikon might have a 2000pxel (or whatever) color meter, but Pentax and the other brands (canon, sony) all use a simple 18% grey light meter, just like the K1000 and Spotmatic did. Sure it's fancier, with all sorts of zones, but I assure you, unless something changed with the K-5II, Pentax just measures reflected light. And Pentax, if anything, is slightly troubled with "light fur" over "dark fur", so I'd assume, dark skin doesn't give Pentax an issue either. BTW, I think Nikons meter still just measures reflectivity of light, but it uses algorithms of stored scenes to determine the scene and adjust for the scene when in Program or Av/Tv priority modes.<br>

<br />So my question is, are you using a handheld reflective spotmeter or an incident light meter?</p>

<p>If you are using a reflective meter, I have some bad news for you, it works just like the spot meter in the K-5. Pentax has always had a nice tight spot meter in its flagship SLR models, and while they don't publish the size of the spot for the K-5 (at that I know of) I bet it's pretty darn tight. Probably not as tight as your handheld, but you can "zoom" in with the camera lens if you are using a zoom, and if not you can just walk up to your subject.</p>

<p>Anyway, as a general rule, if you were to spotmeter off a dark skinned persons face, you could simply guess how much tonal difference it was over a 18% grey face. My skin is pretty close to 18% grey, but not exactly. So I'd often just take a meter off my hand and open up +1/3 stop. If your subject is dark skinned you might take a meter off her face and close down/compensate -2/3 stop. This would make your subject be slightly darker and not turned into a muddy mid tone. </p>

<p> </p>

<p>Andrew: Thanks, I'm going to try your tip.</p>

<p>Justin: My question was really related to a handheld incident meter. I have been using Pentax's multisegment metering mode lately on the P setting. Sad, but true. I started out using the reflective spot meter in my earlier photos. That worked great when I slowed down to set exposure lock and then recompose. Unfortunately, I am not that disciplined. But as you can see from this discussion, I am working my way to trying to be much more creative with lighting (both strobe and natural). That means, understanding the DR, and making sure my metering is ... thoughtful ... and within the DR of the K-5. Nice to know that the Pentax Spot Metering is pretty tight, because I am going to start back using it with the idea of actively choosing what I want as Zone V. </p>

<p>So I just popped over to DPreview to check out the K-30 review and what do you know, they have an excellent discussion on dynamic range. The new tools allow you to see the DR of the different jpg picture modes as well as show the general DR of the camera. http://www.dpreview.com/reviews/pentax-k-30/13</p>

<p>So assuming the K-5 and K-30 use the same algorithm for the various scene modes, it appears that Portrait and Natural gives you a DR of -5/+3 with 18% at 0. Radiant gives you -5/+3, but shifts 18% gray -1 stop. Muted appears to give you the widest DR at -6/+3, but also shifts 18% gray -1 stop. Landscape gives you the most highlight information with a DR of -4.75/+3.5 with 18% at 0. While the smallest DR is with the jpg set to Reversal Film with a DR of -4/+3 with 18% shifted 1/3 stop or so. </p>

<p>It appears that Muted will be the "out of the box" choice for jpgs to determine the most latitude before clipping, while Reversal Film should probably be selected if you want to ensure that the entire scene is will not be clipped at all. </p>

<p>I hope that DPreview updates the K-5 review when they test the new K-5 II/s. </p>

Jemal, here's another tip that I

sometimes use:

 

Instead of futzing with grey cards,

meter off the palm of your hand.

Decide that your palm is +1 (or +2/3,

whatever you wa t). If you meter from

a known tone, it won't matter what

tones are in your subject.

 

Whatever meter you're using you

probably need to apply some amount

of calibrate-to-taste. You may find your

incident meter doesn't quite match

what you get with the in-camera meter.

 

I know a good portion of traditional

zone system is best applied to b/w

negative film/processing but reading

about it sort of opened my eyes a bit

and I could start to see ways that some

of it could be applied to shooting

digital. One thing to be careful about

though is avoiding the white clipping

(blown highlights) as these are rather

ugly on digital while on negative film

the transition was much smoother and

more acceptable. +3 may get you

more of this than you want. The

highlight correction feature on more

recent pentax models helps with this

somewhat by deliberately

underexposing highlights by a stop and

boosting the remainder of the exposure

to where you would expect 'proper'

intended exposure to be.

<p>Hi Andrew:<br>

Thanks for your input. I'd heard of the meter off the palm of your hands suggestion before. I'll have to remember it. As for clipping whites, I think it's pretty safe to say that whatever 18% gray is, Pentax will only allow about 2 stops over that before clipping. Of course, it can go 3 stops over in the RAW file, but that would give you latitude to push the picture when you get it home. <br>

Everyone: Assuming that the jpg safe DR is 4 stops under and 2 stops over 18% gray, this now changes what I decide is 18% gray. Since I have far more latitude in the shadow range, maybe I'll move 18% gray up a stop or two (Scene dependent) in metering, since I would have 4 to 5 stops of shadow to use for recovery when I get home to process the picture. </p>

<p>That's pretty much what I was getting at -- I think you get more than +2 but +3 is probably too much. A mild RAW recovery works OK but a bit more and you're likely losing some color accuracy because the three channels probably don't clip together. This is one reason I like the RGB histograms on most of the newer Pentax models. And since K-5 RAWs (esp at lowish ISO) have such good shadow quality you can bring them up as needed even if you underexpose a little to keep your highlight detail.</p>

<p>Well I have settled on turning Highlight Recovery (or whatever it is called) on in camera. There really is no benefit to having shadow protection on as it reduces dynamic range. I have also opted to using Reversal Film as my primary jpg, and will switch to Muted depending on the situation (for Raw DR checking mainly). I have also decided to push compensation +1. Hopefully, I will get to shoot this tomorrow or the weekend and can post some sample photos in the POTW.</p>