NikonGear'23
Gear Talk => Other => Topic started by: James Farrell on January 17, 2016, 18:58:51
-
MY QUESTION: This post inquires whether LED lighting (as the sole light source) would adversely affect the auto focus system of a typical, current generation Nikon body, e.g., a D750 or a D7200. I recall reading somewhere about incandescent lighting possibly affecting focus. But I don't know about LED lighting. Please review the following background info about the specific issue that has me baffled.
BACKGROUND: Most of you are familiar with some of the different kinds of 'third-party', auto focus adjustment fine tune accessories on the market like the Lens Align-Focus Tune system. With this system, one aligns a high contrast target parallel to the sensor plane of the camera, take a series of shots and examine where upon the slanted ruler (attached to the target) is the most sharply-focused point - in front of the square target (the zero line) on the ruler would be front focusing, behind the zero line would be back focusing. You're supposed to shoot Jpg fine, standard picture control (and default in-camera sharpening), aperture wide open, tripod mounted, shutter delay, remote release, with the target properly illuminated (white balance set for the temperature of the lighting) in AF-S mode, and center focus point only with a high enough shutter speed to avoid blurring of the target but not too high an ISO setting as to create a lot of noise. (Note: Focus Tune software can be used to evaluate the test shots at different in-camera focus adjustment settings, but this post is about just the process of focusing on the target itself under artificial lighting.)
MY ISSUE: The issue I am seeing is that the auto focus performance for the same overall settings, is quite different when shooting these test shots indoors with my focus test rig illuminated by a 3,000 lumen LED light with a 4000K temperature rating versus taking the same shots outdoors in daylight (highlights NOT blown out) with the camera set to auto white balance. The significant difference for both normal PDAF focus AND live view focus between the test results indoors versus outdoors has me wondering if these cameras react (focus) vastly different depending upon the kinds of light sources. Yes, I understand that a JPG image will look very different depending upon the light source and the white balance settings (which normally I don't worry about since I shoot exclusively raw), but he difference I am seeing in a controlled (as much as I can) test environment has me completely baffled. The words "different wave lengths of light" come to mind. But I am not schooled in the science of lighting. Thanks for reading this convoluted mess. >> Jim
-
Your question can be broken down to:
Does the spectral composition of the lightsource influence AF or is the AF color blind?
-
No optical system is entirely "colour blind". Fluorescent light is troublesome for most digital cameras so should not be used for this kind of tests. Continuous-spectrum sources are better. Using a green filter over the lens might improve accuracy more.
-
If you were using Matrix Metering, then repeat the experiment using Spot Metering. I was thinking that how the camera applies the results of the Matrix Metering algorithms could perhaps affect auto-focus.
-
No optical system is entirely "colour blind". Fluorescent light is troublesome for most digital cameras so should not be used for this kind of tests. Continuous-spectrum sources are better. Using a green filter over the lens might improve accuracy more.
I don't understand the term "continuous-spectrum" ... I am simple 'meat and potatoes' Irishman (even though I hate green beer and don't like Guinness).
If artificial light does affect focus performance, how do studio photographers adjust? What do they do?
-
The LED light may be flickering at higher rate than the frequency of the AC current (50Hz) to lengthen the life of the LED and to save the power. Like the image sensor, the PDAF sensor is also working at a certain "frames per second" to read out the image info. There can be some interference between them.
-
The sun, an incandescent light bulb, or an electronic flash delivers a continuous spectrum out. A fluorescent tube has narrow spikes in various parts of the visible spectrum and almost nothing in between.
I don't think studio photographers are too much worried by fine tuning. If there really is a consistent focusing discrepancy, they let a repair shop recalibrate their gear.
The modern DSLRs usually have a user-controllable setting for flicker reduction (50/60 Hz).
-
The LED light may be flickering at higher rate than the frequency of the AC current (50Hz) to lengthen the life of the LED and to save the power. Like the image sensor, the PDAF sensor is also working at a certain "frames per second" to read out the image info. There can be some interference between them.
Thanks you Akira for your thoughts about the LED lighting. And thanks to the others who responded as well.
-
The sun, an incandescent light bulb, or an electronic flash delivers a continuous spectrum out. A fluorescent tube has narrow spikes in various parts of the visible spectrum and almost nothing in between.
I don't think studio photographers are too much worried by fine tuning. If there really is a consistent focusing discrepancy, they let a repair shop recalibrate their gear.
The modern DSLRs usually have a user-controllable setting for flicker reduction (50/60 Hz).
Bjorn: Just to clarify ... I am not using fluorescent lighting - using LED. Having just read about the LED flicker issue posted by a cinema photographer in Switzerland, and now understanding more about it (and what Akira had to say), my question is would one try to have their flicker reduction setting in their DSLR be the same or as different as possible than the flicker rate/frequency of the light source? Sounds like a naive question, I know, but I'm closer to getting a grasp of the issues having read the responses here.
-
James.
"Contiuuous Spectrum" means that all colors of the rainbow contribute to the mixture we call "white"
That is usually the case if light is emitted as a side effect of heated matter (as in the sun or a light bulb or a flash unit)
But this is not the only way to "make light"
a fluorescent tube makes Ultraviolet light first and then converts
it using a fluorescent foil that emits slightly smeared out single
colors. 3 colors in older. 4 colors in typical. 5 to 6 colors in so called
"Full Spectrum" tubes, like the ones I use in my studio.
LED light is similar to that. Single colors smeared out by another
physical effect. If you combine LED of different colors you might
also have a white light of several colors.
but
the AF system might work better in green light than in red light or in blue.
Questions?
-
James: set the flicker reduction in-camera to that of the mains: 60 Hz in USA, 50 Hz in Europe etc.
-
There are two types of "white" LEDs: the one with combined red, blue and green LEDs and another one with a UV emitting chip to induce the UV-fluorescening substance. The former is the better one and the latter is generally used for cheaper products.
If it is an RGB one, you shoule be able to see the RGB spots when you put a piece of white paper right in front of the LED and observe the reflection (some ND filter should help you see the reflection without damaging your eyes). If it is a UV one, you would observe some yellow tint in the reflection if you do the same.
-
Is the distance to target the same for outdoor scenes and indoor scenes? If too close to the target, the distance between the central focus point and the laterally positioned scale could come into play if the lens has field curvature. For instance my Nikon 12-24 mm will focus very differently depending if a peripheral focus point is used vs. the central focus point, while other lenses are insensitive in this respect, so it is not a sensor adjustment issue.
I have yet to detect any difference in focusing accuracy of the central point whether my D7100 and D5100 is used with a warm LED bulb/warm fluorescent bulb mix or used in outdoor in cold blue winter light.
-
There are two types of "white" LEDs: the one with combined red, blue and green LEDs and another one with a UV emitting chip to induce the UV-fluorescening substance. The former is the better one and the latter is generally used for cheaper products.
If it is an RGB one, you shoule be able to see the RGB spots when you put a piece of white paper right in front of the LED and observe the reflection (some ND filter should help you see the reflection without damaging your eyes). If it is a UV one, you would observe some yellow tint in the reflection if you do the same.
I made an image (attached) of the LED light that I was using and to which I referred in my original post in this thread. Note the small 'yellow' squares (about 3 millimeters square - shown enlarged in the inset on the lower right) in the middle of the light fixture that's about 18cm in overall diameter. All light from this fixture emanates from these yellow squares. It's extraordinarily bright - 3,000 lumens using only 40 watts (US 120 volts AC). So, I assume this to be the cheaper LED light source and one not recommended as an artificial light source for indoor photography. Or, is that an overstatement?
My Nikon D750 was set for 'auto' flicker reduction. I could set it for 60Hz (US grid). But I would assume that the auto flicker reduction setting (which is the default) would have made the appropriate in-camera adjustment. Perhaps that's a flawed assumption.
In any case, I appreciate everyone's input on this. Once again, I thought I was being clever and tried to do something 'on the cheap' (buying inexpensive lighting). Another lesson learned. I'll just stick to Mother Nature's light source from now on. Thanks again, Jim
-
These are HiPower LED as you will finbd them in car lights or flash lights
If you can identify who made the light you might also find technical data. In the technical data you will also find a "spectral emission curve" that tells you which colors are present in the light and in which proportions.
I did never hear that LED "flickr" and I do not know how they should "flickr".
I guess LED are part of an electronic array with condesors and that they fire at random in such a way that the light emission is more or less the same at every point in time.
-
James, thanks for the details of the light source. Apparently the LED is of an UV induced fluorescent type.
Frank, flickering high power LEDs reduces the heat. By flickering, you can reduce the duration of the electric power pouring into the LEDs, which reduces the wattage per hour, which reduces the heat and elongate the life of LED chips.
I'm not sure whether the lighting system James is using is flickering, but that could be a potential reason for the problem.
-
James, thanks for the details of the light source. Apparently the LED is of an UV induced fluorescent type.
Frank, flickering high power LEDs reduces the heat. By flickering, you can reduce the duration of the electric power pouring into the LEDs, which reduces the wattage per hour, which reduces the heat and elongate the life of LED chips.
I'm not sure whether the lighting system James is using is flickering, but that could be a potential reason for the problem.
Notwithstanding the fact that doing all of this silly focus testing would be easier with natural sunlight, I wonder if it's possible that at the "Auto" flicker-reduction setting (in Nikon's camera setup menu) is not the best choice. I think I'll test this light with setting the Nikon flicker reduction setting at 60Hz (one of the settings available in the setup menu) instead of Auto-reduction. Maybe ... just maybe ... the camera is not able to correct for the right auto flicker reduction solution. In other words, maybe I should treat this light as a fluorescent fixture instead of an LED fixture. Thom Hogan suggests that if using fluorescent lighting to use shutter speeds that are multiples of 120 (120 volts AC and 60Hz), e.g., 1/30, 1/60, 1/120 etc.) and, of course, appropriate white balance. This interesting (well, to me at least) issue has become more of a educational science experiment now versus a photography question.
-
The frequency of the flicker of LED lightings should be much higher than 50 or 60Hz because the LED reacts much faster to the on/off of the AC power supply than the fluorescent light bulbs (tubes). The so-called switching power source is commonly used for the power source for the computers, audio amplifiers, etc. The switching power source oscillates square waves of very high frequency out of the audible range. I'm not sure whether such a power source is used for this particular product. But if it uses a switching power source, the anti-flicker function of any digital cameras that are optimized for 50/60Hz wouldn't work or make any difference.
-
Late to the party but I did some research a while back because I needed a cooler (ºC) light source for my copy stand. I have been using two 80W halogen floodlights but they got very hot, and consumed much electric when copying images for many hours. Also the halogens were too warm (ºK) for accurate colour reproduction. Daylight isn't an option in my situation.
Eventually I found these GU10 lamps which are pretty good compromise, the spectrum while not full, is the best I have found and they seem to do a much better job than the halogen units without the heat. They are not in what I consider to be an expensive price range, ~£4.50 ish, from memory? Compared with my Nichia 365Nm UV chip they are for nothing!
The "Switching Cycles" frequency quoted in the spec's is 50,000Hz. I don't see such a high frequency affecting the AF? More likely the gappy spectrum I would have thought?
These lamps were by far the best spectrum I could find, some had complete gaps and spikes.
http://www.farnell.com/datasheets/1782809.pdf
(https://farm2.staticflickr.com/1572/24440188546_4a3d8552f9_o.jpg)
I chose the 4,000ºK lamps and use six of them to illuminate my copy stand, they are on adjustable mounts, three each side, so I can move them as required to make the illumination even depending on the subject. The gap on blue-green doesn't seem to make a noticeable difference in my eyes anyway.
-
Robert, thanks for the follow-up. Judging from the spectrum chart, these seem to be of the blue/purple (sorry, not UV!) induced yellow fluorescence type.
-
Akira, you posted while I was composing. We seem to agree on the frequency thing. There is little or no UV output but a little IR in these lamps, others vary a lot, it would be good to be able to see the output spectrum for other lamps but most manufacturers don't seem to publish that data.
-
James,
back in January 2011, the D7000 was just released, I used the new camera with the then current generation of fast lenses (AFS 24mm/1.4G, AFS 35mm/1.4G, AFS 85mm/1.4G). Much to my disappointment, the AF was way off for longer distance when shooting with an open aperture. What started as a seemingly simple issue to just correct a misaligned AF module in the D7000, became a much broader effort. Over the course of a few months the error "extended" to many more combinations in my gear set.
Ultimately,
the following camera's made the trip to Nikon's main service center for Central Europe multiple times: D7000, D3, D3s, D3x, plus the following lenses: AFS 24mm/1.4G, AFS 35mm/1.4G, AFS 85mm/1.4G, AFS 24-70mm/2.8, AFS 200mm/2 VR, AFS 70-200mm VR1, AF 85mm/1.4D. (many more lenses were impacted, but I didn't sent all. The lenses above represented a good sample). Some of the gear made the trip 3 or 4 times. As the backfocus/frontfocus pattern had so many variations (camera/lens/type of light/focus distance combinations), I sent in the whole gear set to allow the Nikon technicans to check their desired combinations as well. I spent many hours to use a similar lens cal device like you (lenscal) to check AF precision and had many phone conversations with the Nikon staff in the service facility in Düsseldorf (their main facility in Central Europe). Sent Nikon extensive test protocols and pictures to support their work finding the "pattern" of the errors.
I documented many of my findings in a thread in the german NFF forum. Sorry, it is in german, but anyway, here is the link for those who can read german, or use machine translation:
http://www.nikon-fotografie.de/vbulletin/nikon-d7000/151532-fokusprobleme-mit-neuen-lichtstarken-objektiven.html
In hindsight, a few things I remember (almost 5 years already passed by):
1) First check should always be if your lens is decentered. Particular important if you use an AF cross sensor on the testpattern. Makes your AF measurement random.
2) AF adjustment in the camera is distance related. Adjusting the focus at 3m right is no guarantee that it is equally precise at 10m.
3) Faster lenses are more affected (due to shallower focus). Only f2.8 or faster lenses are impacted. Focal length doesn't make a difference if a lens would be impacted or not.
4) Most affected where the generation of cameras between 2007 and 2010. With the D800/D4, quick checks didn't produce this erratic behavior of the previous generation.
5) Quality of light definitely impacted AF test results (incandescent, outdoor, daylight bulbs, ...).
6) Nikon's factory error margin for the D7000 AF was outside my understanding what infocus or back/front focus means (The D7000 was the first 16MP DX Camera and my camera came early - don't know if the QA process was later adjusted)). FYI, Nikon's guidelines prescribed a 50mm with f2.8 to calibrate the AF
7) Lower cost cameras as well as higher priced cameras where affected
8.) One lens (AFS 24mm/1.4G) was replaced by Nikon
9) After 2 months focussing on AF focus issues, the situation improved to a state, that it was sufficient for me.
10) The D2X and D2Xs where not affected, the D300 and D300s to a smaller degree
11) I learned a lot about potential AF errors.
12) It was my first needed engagement with Nikon's service organisation - until then, things worked fine
13) I am now much more relaxed about AF systems :) (but I am still curious about the performance of the D5 and D500 AF systems :) )
Please let me know if you have any further question, happy to try to answer it. Good luck with your gear.
rgds,
Andy
PS:
I can't find the info anymore but I believe I've read years later, that the randomization in the AF systems were due to IR sensitivity in the AF module. But as said, just a memory snippet.
-
The frequency of the flicker of LED lightings should be much higher than 50 or 60Hz because the LED reacts much faster to the on/off of the AC power supply than the fluorescent light bulbs (tubes). The so-called switching power source is commonly used for the power source for the computers, audio amplifiers, etc. The switching power source oscillates square waves of very high frequency out of the audible range. I'm not sure whether such a power source is used for this particular product. But if it uses a switching power source, the anti-flicker function of any digital cameras that are optimized for 50/60Hz wouldn't work or make any difference.
I think I understand this well enough to draw a few conclusions that the LED lights I've tried to use will be useless for indoor lighting with my Nikon bodies, and best used to illuminate the driveway for night-time snowshoveling (yes, where I live in Arizona I get snow). One interesting and perhaps telling characteristic is that despite the very high lumen power (3,000) of this fixture which is only 40 watts, it hardly produces any heat.
Thanks to all the contributors to this very informative discussion: Akira, Bjorn, Frank, Andrea, Seapy, Olvind, and Andy (if I omitted anyone, I apologize). Thanks again, Jim
-
You can demonstrate the pulsed nature of LED illumination by swinging your camera rapidly past some LED lights with the shutter open and pointed at the lights. Same as with sodium and mercury vapor, fluorescent tubes and neon signs.
Even the expensive, so-called "high color rendering index' (high CRI) LED lamps intended for photographic use may require color correction filters in order to closely color-match traditional, continuous-spectrum photographic light sources such as daylight or tungsten.
EDIT: Yesterday I rode past some still-deployed Christmas lights of the LED variety. They must be going at a high frequency to make such small traces: