My camera can't do square roots

<p>While working on his homework my son ask if I'd like to calculate the determinant for a 4X4 matrix. I just said use the graphic calculator. However, his instructor told him to complete the assignment longhand.<br>

That' what galls me about digital photography: you purchase an expensive camera, lens, software, etc. and <strong><em>you have to get the color correct ... longhand</em></strong>!<br>

Imaging working out a simple percentage problem: take out you calculator and compute 7% of 1,295 <strong><em>but don't rely on the answer until you view it on a calibrated screen. </em></strong>Let's add up your camea purchase on the printing calcualtor <strong><em>taking into account the numbers aren't accurate because the printer isn't profiled.</em></strong><br>

OK. I've had sufficent experience to know these factors exsist but still haven't reached the point where I've perfected anything. I remained amazed the extend of subjective input necessary to get a good, final copy.</p>

<p>Joe</p>

<p>I have the opposite experience. There's absolutely nothing subjective about the laws of physics. All Schrodinger's Cat jokes aside, there's nothing you can do (subjectively) about photons, wavelengths, diffusion, diffraction, reflectivity, dynamic range, light sources with dodgy spectral output, colored walls, chromatic wackiness in lenses, etc. They are what they are, whether or not you see it, understand it, or care. Getting your camera's recording to wind up on a piece of paper in your hand, and to have it "look" like what you saw ... is a process with a lot of variables. They are all objectively understood, and we have accurate tools by which to measure, adjust, and compensate for them. Sometimes we can't (reasonably) overcome things like limited dynamic range ... and so we have to make judgement calls about what aspect(s) of the image are more important to us. <em>Those</em> choices are "subjective" because they depend on what you want to communicate.<br /><br />Otherwise ... shoot RAW, include a WB reference, use calibrated display and printing devices, and you'll be close enough for most any reasonable use. Beyond that, you're dealing with issues like the light in which the print will be seen, and other stuff that requires special (and subjective) consideration beyond "accuracy" in the sense that I believe you mean it.</p>

<p>Hey, speak for yourself. My camera sure can take square roots! It can even do x^2.2 and x^(-2.2). In fact, it does this every time it takes a photo and does a gamma conversion. Unfortunately, reading the result as a number is a bit tricky.</p>

<p>Sorry to interrupt the worthwhile discussion of subjectivity.</p>

<p>Tom M</p>

<p>Matt,<br>

I know. My frustration originates in the ability to recognize the variables and compensate in the outcome. Actually, there's no way the engineers designing the optics, sensors, etc missed anything therefore <em>I'm describing factors I've yet to fully control. </em><br>

That said I really loose all confidence when someone post a photo only to have others comment "color shift" or "too much yellow". And when I get noticeable changes in the color when using a gray card reference / sampling in post processing then I feel really lost. <br>

Ahem ... I took a photo with a gray card, adjusted color to the card, printed it, held it up against my red car and had a very, very good match (good enough for me). When I repeated this exercise with another photo and detected a color <em>cast on the gray </em>card I begin to think something isn't working. OK, it's me but why isn't this more foolproof?<br>

I constantly check here to learn more. Matt, your postings are quite helpful (well, except the photo of the toilet innards :) ) so thanks for your reply.<br>

Joe</p>

<p> </p>

<p>We humans are the weak link in the chain since our vision is perceptual (e.g. we don't need to WB our eyes for different light sources). If anything, our electronic imaging devices are <em>too </em>accurate, picking up nuances of light that exist but are largely invisible to us. Calibrating is an accommodation for the limitations of our vision, and to compensate for drift/wear/aging of both us and the electro-optical devices we use to view the images. You could think of a display calibrator as a translator to get some very accurate numbers into a form that's digestible for us.</p>

<blockquote>

<p>OK, it's me but why isn't this more foolproof?</p>

</blockquote>

<p>I ended my frustration with an x-rite Color Checker. Nothing to adjust in-camera...just shoot and it makes a correct camera profile in post. Not totally foolproof, but it's close.</p>

<p>Don't forget that it's possible to have something in your throw some reflected (colored!) light onto your WB target. Say you're hodling it up next to the hood of a red car, and you're catching a whiff of that color on the card, rather than strictly the sunlight that's lighting the scene as a whole. If you didn't have to make more than one WB shot, then you certainly do have to wing it in post, when you see something like that. Nothing wrong with seat-of-the-pants, sometimes.</p>

<p>Um.. that should say "... something in your <em>scene</em> throw some ..." Woops!</p>

<p>Matt,<br>

I though that as well. The red finish didn't cause a reflection on that particular instance, it was another case with the gray card setting in the yard against the lawn. Kinda shakes your confidence when you can't trust <em>the standard.</em><br>

DB, I profiled my monitor but still can't figure out why a maple tree looks slightly brown instead of a light gray (Nikon, in addition to face recognition, must use object recognition and just plugs in a color. Probably why all my fish look gold) When I change the WB in post to Daylight from "As Recorded" the tree then changes closer to the correct color. But why Daylight, was Shade correct, or Overcast? The fact I had to override the camera and <em>subjectivly</em> pick the WB doesn't always work in my case because, farnkly, I'm not to good at it.<br>

As I mentioned, there's no question many of those posting at this site know exactly what they want and that their judgement yields wonderful results. I'm not expecting I will do the same: I'm just tired of orangeish family (and not incandescent lighting).<br>

PS: Back in the day, I could never figure out why machine shop/manfacture would insist machinists measure their work with some 30yr old micrometer despite having the latest $150K Cincinnatti Milling Maching CNC <em>that was designed to produce an item to unparalled tolerances. </em>(No produce plug here). The mahine was the culmination of thousands of man-years of knowledge and experience. The CNC got it right much like I thought the camera would get it right.</p>

<p>It is not just digital photography; we have always had to compensate and correct images.</p>

<p>In black and white photography, most photos require some dodging, burning or contrast compensation, even if you use the Zone System.</p>

<p>Color photos have the color adjusted - either by the operator or the mini-lab itself. Of course you do not see the process (assuming the lab does a good job), but it still happens behind the scenes.</p>

<p>Photography is not the only activity that requires adjustment. Your car requires adjustments - it called a tune up. Your Hi-Fi set (that dates me) or iPod or iWhatever has sound adjustments - minimum bass and trebble.</p>

<p>As for your son doing a determinates long hand, that exercise should have been to give him a better understanding of how determinates work and when to use them. On the other hand, he may just look on it a drudgery and learn nothing. It is all a matter of attitude.</p>

<p>In a like manner, making the adjustments and understanding why we make them may make us better photographers rather than just camera operators.</p>

 

<p>Brooks,<br>

I realize those points but expected closer results pretty much most of the time. My tree image, for example, tempered in the context the surrounding portions of the image may influence the outcome.<br>

I'm not attemting to belabor this discussion for the sake of its continuation (all the good posters will rightfully drop out). Just posed the question: mostly serious, a little whimsy. <br>

But I can argue my point by citing all the products on the market intended to derive accurate color from digital images. Even the recommendation to shoot RAW in some way validates the inabiliy of the camera to get it right (all the time, every time)<br>

Hey, I'm not really arguing other than to say how difficult a time I have taking an image and feeling confident I can get reasonable results.</p>

<p>Chipping in my two bits, any good painter will tell you that it isn't a matter of matching a pigment mix to a reference in the scene, but of getting the right color relationships, pigment mix to pigment mix, in the painting. A painting can't get the range of luminances and color values in the original scene, but a good painter can make you think it does--with pigment mixes having surprisingly little to do with the hues and values of the original scene. More generally, any snapshot can tell you what a person looks like, but a great portraitist like Karsh or Ingres can tell you who that person is.</p>

<p>If you really want to emulate a milling machine, include a color chart in one frame and make adjustments until a print of that color chart can't be distinguished from the original. Then make a print of the frame of your choice with the same adjustments. It's possible that you have a color blindness that prevents you from matching the print to the chart as a normally sighted person would. In that case I suppose you'd use a colorimeter. You can get a faithful reproduction, which isn't to say a good one.</p>

<p>Someone who says "color shift" or "too much yellow" wasn't present at the original scene and can't judge the influence of color reflectances outside it. He's basing the criticism on inherent properties of the photo, which may indeed be a perfectly faithful reproduction of what was there. You still have to arbitrarily decide what to do to make a photo <em>look</em> natural, in spite of the fact that it already <em>is</em> natural. It won't help to burn joss to the God in the Machine.</p>

<p> </p>

<p>Charles,<br>

Right. I have read about using a color chart within a sample image as a method to make sure the color is accurate. And that makes sense, establishing a standard, kind of like placing a ruler within frame for authenticating dimensions. But if it is necessary to establish color, or avoid color changes, by use of a known standard it somewhat vaidates my original comment; the color isn't very automatic despite the sophistication of the modern DSLR. And that really doesn't make sense given the abilities of the manufactures, which I know, but I keep encountering examples to the contarary.<br>

I'm not really disagreeing with anyone; I respect the people that have commented. I just hoped the digital was, in fact, a process rather than an art.<br>

And, I'm not color blind, although my father is, but if I ever post an image with a blue grass and green sky I may plead otherwise.</p>

<p>Expecting digital cameras to correctly gauge color in each scene is just like expecting them to nail exposure on each shot while in [P] mode; while it may do a good job many times, it would be unreasonable to expect it to know what your intention was in the shot.</p>

<p>It is nothing new that just like exposure, color temperature has to be measured in each scene. You can either let your camera make its best assumption with it's built in metering, take an external meter reading and find out what is present more accurately, or use one of the devices that allows the camera to function more like an incident meter. Color, like exposure is in the end a decision left for the photographer to make based on their vision and interpretation of the scene, whether accurate to the original scene or otherwise.</p>

<p>Nothing new, it is just that you are the one doing the adjustments instead of a lab. Color negative film was a lot worse IMO, thats why I used color positive film (slides). So i could see the 'correct' colors, which of course were bias based on the film brand and type. But, good luck on getting slides printed correctly. Prints could never match what is seen on the light table. To be honest, while it is a hassle I get the results I want now with digital.</p>

<p>If you look around you may find a service shop that will print your photos for you. </p>

<p>Someday display technology will evolve and made color correction easier. Hopefully sooner than latter. </p>

<blockquote>

<p>DB, I profiled my monitor but still can't figure out why a maple tree looks slightly brown... ...The fact I had to override the camera and <em>subjectivly</em> pick the WB doesn't always work in my case because, farnkly, I'm not to good at it.</p>

</blockquote>

<p>Neither am I, hence the x-rite Color Checker. It calibrates the <em>camera profile </em>(the camera's response to the light), not the monitor, to get the color spot-on in almost any light. The camera profile is used to adjust the image files...no change is made to the camera itself. It's analogous to using CC filters with film, only with a <em>much</em> higher degree of accuracy. You just shoot the color target under the light you're actually shooting in. Then in post, a profile is generated off that target image by the computer, and you apply that profile to your other images shot under that particular lighting condition, and voila - perfect color. You can even tweak the profile if you want. These days I shoot RAW and auto-WB and use the x-rite...it's less to think about so I can concentrate on shooting.</p>

 

<blockquote>

<p>Back in the day, I could never figure out why machine shop/manfacture would insist machinists measure their work with some 30yr old micrometer despite having the latest $150K Cincinnatti Milling Maching CNC <em>that was designed to produce an item to unparalled tolerances. </em>(No produce plug here).</p>

</blockquote>

<p>Funny that - BITD I worked for two years for Milacron in Plant 1 at the Oakley complex in Cincinnati, in the gear cut department. I can vouch that the company had all of us use either their or our own mics and indicators to check the work, and some of the machines we used were capable of incredible accuracy - if there were set up and operated correctly. The machine I ran most of the time - a German Blohm-Hauni NC serration grinder - adjusted in increments of .001mm (roughly .0000040"). It was used to put 360 or 720 teeth on the control couplings that indexed the worktables of those CNC mills to within .0001" over 360 degrees. No matter how well any machine is built, over time, it'll drift, wear, etc. As accurate as that machine was, it'd drift as much as .0008" to .0012" on a single part <em>according to the weather</em> (Milacron finally had to put it in an environmentally controlled room). You'd be surprised how much metal can move, and how often high-tech machine tools have to be adjusted or 'dialed in' by the operator. Did you know that even the liquid cutting coolant can wear and affect accuracy?</p>

<p>We were required to submit our mics and indicators (company or personal) periodically for documented calibration against master gauges in a lab in Milacron's tool inspection system (they engraved my personal mics with their system control numbers). We also had to have our work verified daily by an inspection department - we called it 'the gauge room'. After a setup, we ran one piece and had to have that verified by our department's inspectors in the gauge room before proceeding, and depending on the part, we also had to submit others in the run for complete inspection (this was besides checking them ourselves). The gauge room's instruments were extremely accurate, but they also had to be periodically calibrated against lab master gauges and documented. You can imagine I understand the value of proper and periodic calibration of anything to maintain top accuracy.</p>

<p>DB: In some way your description of the calibration procedures to verify accuracy of machinery and measuring equipment describes what I assumed one could accomplish with the entire digital flow: camera set-up, capture, post, output. The procedure you describe using a color profile, X-rite, etc. will produce accurate color but then I question why the camera missed in the first place. If a camera could get within the variblity you cite for a gear cutter then my comments would be moot. I've come to understand that color & exposure accuracy fall more in control of the operator, but when I read instructions that say "... move the slider until the color cast is removed .." that's the point I feel lost. Oh, I may have worked across the street from you one summer although possibly I'm confusing my Cincinnati locations (long defunct Williamson Company).<br>

Justin: I don't disagree with the comments of the others responding to my post. My comments relate more to my suprise color & such can wind-up so far off that intervention is in order. Makes the learning curve - mine at least - quite steep. <br>

Matt B: I think your comments may describe my experience, a hassle, but if done correctly, better results. Just that darn learning curve.<br>

Charles: Color relationships. Testors Model Masters Technical Guide describes <em>scale color theory</em> (Distance Effect). Although the amount of white paint to add to the base color seems broad, I can count drops of paint! I wish I cound match digital color the same way. (OK. Not the same as color accuracy but it seems to fit somewhat within the context of this post and his earlier comments) </p>