Trouble with Nikon D7000 files from concert

<p>Hi all,<br>

I student of mine is getting some strange looking files from her Nikon D-7000 when she shoots under gelled lighting at concerts.</p>

<p>This appears as a blown out, etched looking highlight, a highlight that isn't really there when viewed in some programs other than ACR and PS.</p>

<p>Any suggestions on what to tell her to do under these conditions, I'm stumped.<br>

<a href="http://www.eyelement.com/photos/concert.jpg"><img src="http://www.eyelement.com/photos/concert.jpg" alt="" /></a></p>

<hr>

<i>Moderator note: I've pruned this thread as much as possible to refocus on solutions. My apologies if there's a problem with continuity or context. Hopefully the essentials have been preserved. --LJ</i>

<p>Tel her first to leave the exif in te picture.</p>

<p>Hans, here's a bit of detail: Auto WB, 800 ISO, 1/20 @ f1.4, 50mm f1.4, Spot metered.<br>

No changes in ACR, this is as seen straight out of camera.<br>

Histogram looks nearly perfect on exposure, well-balanced, except for that "white etch" causing a blue spike on highlight end. This only appears under blue lights, red gel shots OK.<br>

<br />Could this be from UV, aka, black light?</p>

 

<p>A few thoughts:<br>

First off, the file is a jpeg, and only 169K at that. I don't know if it was shot as jpeg, or converted in ACR, but that much compression is destructive. These conditions require shooting in raw.<br>

When I look at the histogram, I see the graph hugging both the low and high end - your student is losing detail at each extreme. If this is the case even in the RAW file, there is simply too much contrast for the camera, and you need to choose what to expose for.<br>

Generally, I choose not to overexpose the highlights, as that looks awful to me (faces, hands, generally), and I just accept the deep blacks at the other end of the scale. If your student is using spot metering, the spot being metered is not on the most brightly lit part of the subject. You have to try to keep the histogram from hitting the right end - unless you have items in the background that you don't mind overexposing (say, a floodlight) - you can turn on the blinkies on the monitor to check what is overexposed.<br>

You mention that this happens only under certain colored gels. I have found that if I spot meter on the part of the image I want to expose for, the color of the gel doesn't make a difference. But, when I have a choice of metering off one half of a face (red gel), or the other (blue gelled light), my eye isn't much help in deciding which one is the brightest. So, I try one, and check the histogram.<br>

Last, I find it helpful in ACR to adjust white balance first, then the exposure parameters. Too hard to judge the image otherwise.<br>

Hope that helps,<br>

Chris</p>

<p>All I can suggest is that gelled lights should really be shot at a Tungsten WB, because that's basically what CT theatrical lights are. Strongly coloured lighting will throw an AWB off and add a cast of the opposite hue to the lighting gel(s). Maybe you should try and educate your students out of using AWB - not that it should matter for RAW shooting.</p>

<p>Another issue is that some colour management settings of PS can give rise to posterisation if the OS tries to colour manage as well. Try turning all CM settings to "off" in PS and see if that improves the look of things. Definitely don't use the awful Adobe Gamma "calibrator" to set up and manage a monitor.</p>

<p>Original file was shot in RAW, and does not appear to be even close to being overexposed. Highlight end rolls off nicely well before the highlight end, then bang - etched white. You're seeing a small area here in a cropped version.<br>

Attempts at temperature / exposure / highlight recovery do nothing to the etched area, it just doesn't seem to act like a normal blown highlight.<br>

The strange part is that in viewing some of the preview files in "Previews" on a Mac, don't show the blown out area. Weird.</p>

 

<p>She needs to try another raw converter. Lightroom and ACR/PS is the same so for instance Capture One, DxO or some of the others.</p>

<p>I think a variety of issues can affect, for instance auto wb as Rodeo Joe pointed out should really be set to tungsten or possibly daylight as it is HMI and LED lighting involved in the fixtures. Also when one color channel is blown out the raw converter may try to salvage the colors instead of going pure white. Great for skintones under normal lighting but not so great otherwise.</p>

<p>Actually the WB setting is always involved when clipping one or several channels because otherwise clipped color information would not go to white but rather the complementary color of the clipped channel. For instance if you clipped the red the color, but green and blue is unaffected the resulting color would go towards cyan. But the raw converter will try to salvage the color when one or two channels are blown out and go to white when they are all blown out. As it tries to salvage the colors that are not there it's using nearby colors.</p>

<p>With pure colors I think it's likely that the highlight recovery algorithms used in ACR simply cannot cope.</p>

<p>PS. Actually I think every decent raw converter should be able to show if one or two channels are clipped (in raw). Clipped in raw is not the same as clipped in the image. But I'm not aware of any software that can do that except rawnalyzer but that doesn't support newer cameras. and the author has unfortunately passed away.</p>

<p>Could this be a combination of motion blur (if this was shot at 1/20) and lights that have time to change colour and intensity during the time the shutter is open? If there's a chance to test again in similar conditions, perhaps going for e.g. ISO 3200 and 1/100 might end up being a better trade-off?</p>

Thanks for all of the suggestions. I will try Capture One at work tomorrow, and try to obtain the EXIF data.

 

The more I look at the files, the more I think it is indeed a RAW converter issue. Shots that were made at the same

exposure, but under different color gels look fine, so I think that would rule out a camera problem.

 

I have seen some D300 files shot under deep blue Rosco gels that exhibited some cyan rollover, but nothing this drastic.

 

Thanks again!

<p>Larry, I think there are two issues. One, the colour of the gelled lights. The lights are magenta ... and if you think about what a Bayer pattern looks like, magenta is easily the worst colour for a digital camera to deal with. Magenta-gelled lights will always always always result in terrible tonal range and bad contrast, because the sensor is less sensitive to magenta than any other colour. Red gels are a fairly close second, for the same reason. Those wavelengths also take up less space in the visible spectrum, which means that between the spectrum and the sensitivity, only getting 2-4 tones out of a magenta-lit subject is pretty common.</p>

<p>Secondly, the image doesn't look like it was processed great. You can maybe improve the image with better processing (or better software), but it's nigh-impossible to ever have a really good digitally-recorded image shot under magenta light, regardless of exposure or processing.</p>

<p>EDIT: I realize after that post that fifteen people are going to post good images shot with magenta light. I'm not saying it's impossible. I'm saying that 1 out of 50 might be good under magenta light, while 1 out of 5 might be good under blue or yellow light. Nigh impossible. Not actually impossible.</p>

<p>The problem is the color of the lights. Also, Pete mentions LED lighting, if this is LED lighting, it is very difficult to impossible to correct with WB adjustment. Also, as it was shot in RAW, the comments about WB are irrelevant and pointless.</p>

<p>There isn't an easy solution when this situation exists. If flash is allowed, it's far better to use it properly (typically with slow sync to allow the colored light ambience to remain) than to try and shoot without it. If it's not allowed and the lights are changing, wait for the lights that are not magenta. Greens tend to work best.</p>

<p>When things just can't get fixed in color, the best thing to do is convert the images to black and white.</p>

<p>Digital cameras, particularly those that use a Bayer pattern, have as many green photosites as all other colours combined. Magenta is the opposite of green. Thus, digital cameras will have worse contrast, tonal range, and sharpness under magenta light than any other colour. They will do the best under green lights. All other colours will render somewhere in the middle, with near-greens like blue and yellow faring better than red or cyan.</p>

<p>Film usually fares better, since colour receptors are randomized, meaning there is a better balance between all other colours. Worse rendering of green, better rendering of magenta.</p>

<p>i shoot in these environments often. i never use auto WB indoors. looking past the lighting issue, 1.4 and 1/20 is not what you want for capturing moving subjects. for live action, 1/100 is the minimum shutter i would use, with 1/200 being what i usually aim for. i wouldn't shoot wide open if it can be helped; 1.6 or 1.8 will at least give you a bit more sharpness; f/2 is even better. the solution here is to raise the ISO. with a d7000, i would shoot at ISO 3200 with the fastest shutter i could get away with, or auto-ISO set to 3200 max, as stage lighting can vary. also, spot metering is sometimes advisable in these situations.</p>

<blockquote>

<p>This appears as a blown out, etched looking highlight, a highlight that isn't really there when viewed in some programs other than ACR and PS.</p>

 

</blockquote>

<p>The issue the OP has is trying to help a student 'fix' things....either to get it better next time <em><strong>AND/OR</strong></em> get the best image from the RAW files already taken.</p>

<p>The fact that the effect is visually variable with different converters, implies it can be optimised.....we want a RAW file to play with!!</p>

<p>Even though the light source is un-balanced, if the actual exposures per channel are not blown, it should make an odd, but believable, Black and White conversion? No?</p>

<p>If, as it appears, the lighting is magenta/blue heavy, how about those old fashioned things, CC filters? Y 50 would cut out a lot of the excess wavelengths. A Wratten 11 would be about right, but maybe not <strong><em>strong</em></strong> enough?</p>

<p>..... and while we're here, how about a Black Light/UV remover, something a little stronger than your average Skylight filter, maybe a Wratten 2E? </p>

<p>Not sure if the camera will successfully complete a grey-card preset, sometimes if it's too far off, it wont work and will show the No Good message. Maybe it <strong>will</strong> if the camera is suitably Corrected?</p>

 

<p><strong>EDIT/Correction</strong>.... Maybe a Wratten No.8 (or K2 Yellow), rather than the overly red-cut No.11</p>

<p>Black and white conversions will not help, because the camera didn't see enough information in the magenta channel to produce more than a couple tones. Colour balancing also will not help, as it will produce an image that is colour-corrected, but still only has a few tones. A colour filter would help, but only if you can get it on the camera and adjust your exposure to compensate for the two or so stop light loss before the LED colours change.</p>

<p>Yes you can 'help' the student. But the way I help my own photo students is by not encouraging them to polish turds.</p>

<p>The easiest solution is next time, wait a few seconds for the LED lighting to change, and THEN take the picture. Patience is the answer - not software.</p>

<p>Each pixel/photosite/whatever you want to call it is receptive to a single wavelength of colour, and only that wavelength. In a Bayer pattern, they are divided into blocks of four - one red, one blue, two green. A photosite sees its opposite colour only when all the photosites BUT that one trigger. So if every photosite in a four-block group triggers, it sees white. If every photosite but red triggers, it sees cyan. If none of them trigger, it sees black.</p>

<p>Any time you have an image that is lit with a single colour, that image is going to be worse than when shot under white light, because you have at least 1/4 of your photosites seeing absolutely nothing. Magenta is the worst single colour, because it means that half your sensor (the green photosites) see nothing. If you were to compare all the possible solid colours, magenta is the closest to pure black, at far as how many photosites trigger.</p>

<p>This is why noise is largely magenta. This is also why all-magenta images are destined to be terrible - because half your camera sensor doesn't even turn on. And this is why all-green images come out better than any other colour - you have the most active photosites. I don't know the math well enough to say, but I wouldn't be at all surprised if the maximum dynamic range of your average magenta-lit image was exactly half that of your average green light-lit image from the same sensor, with white light-lit images being far and away the best..</p>

<p>This has to do with the colour AS IT ENTERS THE LENS AND STRIKES THE SENSOR. That means that white balance doesn't affect anything, since WB is something that the processor does to a captured image. Your exposure doesn't matter, since you're merely increasing or decreasing the brightness of what is bound to be an awful image.</p>

<p>Literally the ONLY possible fix is to use a coloured filter to change the image away from magenta BEFORE the light hits the sensor. But this is a really dumb idea for concert photography. Not only will you lose 2-3 stops of light to any filter that is dark enough to make a difference, but by the time you get the filter out of the bag and on the lens the light probably would have changed anyway.</p>

<p>There's no way around it. It's no different than trying to take a photo of a black cat at night. Some things just make for awful pictures, and the only solution is to not take those pictures - there are no magic beans.<br /><br /></p>