Monday, December 30, 2013

Flashlight Ceiling Beamshots (Take 1)


I've decided to take my experience in compressing a photograph's dynamic range and apply that to rendering flashlight beam-shots in such a way as to show all the subtle colour differences and beam artifacts. There are many subtleties to the beams produced by various flashlights and while it is relatively easy to see these differences visually when shining against a blank white wall, it is not so easy to capture on camera, at least not with a single photo. Combining the large, 'domed' LEDs with high precision, polished reflectors and anti-reflection coated lenses in modern high-end flashlights can also create colour artifacts in the beam, sometimes quite obvious and other times more subtle.

I have endeavoured to somewhat accurately portray these differences in the gallery which you can access by clicking on the above image. Not only might these beam-shots be of interest to other flashlight aficionados, those that sometimes call themselves "flashaholics", but some of the subtle patterns and coloration are also quite beautiful in their own way. I need to be careful, since with my growing collection of lights, I may be well on the way to "flashaholism" myself! A great resource, if you want to learn more about modern high-end flashlights, is CandlePowerForums.

The middle image above (and the first image in the gallery) show the collection of lights, diffusers and filters that I photographed. There are many notes following in this blog entry, describing the test conditions, which lights were photographed etc. One thing that I will stress right from the get-go is that the photos in this gallery can in no way be used to judge the relative brightness of these lights. More details follow…

[update 12/31/2013]: added one missing photo and revised the order of the gallery and descriptions slightly to make more sense.
[update 1/1/2014]: Added labels to each shot in the gallery and fixed a few minor things in the article. Happy New Year!
[update 1/6/2014]: Added in the Thrunite TN32 beam-shot and took out the laser+refracting head to keep the gallery at a clean 35 shots.

General Notes

I photographed these with my Fujifilm X-E2 and XF 14mm f/2.8 lens. I have included EXIF data so you can very roughly judge the necessary exposures, but only gross differences should be used as an indication of relative brightness. For example, the coloured, secondary LEDs in the SRT7 (photos 14-16) are much dimmer than the main LED, and you can see that in the exposure time. In most cases for this gallery, I have equalized the brightness of the beams in post processing, in order to show tint as well as density and coloration artifacts, not to show relative brightness differences between the lights. Luckily the raw dynamic range of the Fujifilm X-E2 is actually extremely good, better than many full-frame DSLRs even, so that helped me to capture all the subtleties of the beams with just a single carefully chosen exposure.

Photos were all single shots, captured raw and processed in Lightroom v5.3. I used dramatic application of the highlight recovery, shadow fill and exposure sliders in order to compress the high dynamic range in such a way to show what one can see visually in the beams. In addition, I did a perspective correction and crop for all the beam shots. You can see in photo 2 what the ceiling looked like, uncorrected for perspective. Photo 3 shows the crop and perspective corrections used for all the beam shots. There was a convenient black dot on the ceiling which I used to line up the lights (you can see it on photos 2 & 3) but for aesthetic reasons, I retouched it out of the centre of all the beam shot photos. As you can see, the ceiling does have texture and is very slightly blotchy (needs a new coat of paint!), but I don't think this affected the beam-shots to any significant extent. I was hoping to use a 4x8 foot sheet of white foam-core material, so as to have a perfect smooth white surface, but there was too much glare with distracting hotspot visible as a reflection. I would have had to angle the camera further than I was comfortable with when it came to the required post-processing perspective corrections.

Note that I did photograph a bracketed set of shots for each light's various output modes but for this gallery at least, I only chose a single shot that allowed me, after post processing, to best portray the beam artifacts and colour for any given light's brightest setting.

Here is an example to show the degree of post-processing applied, photo 4, the beam-shot from my SupBeam X40. This first photo is the untouched version, just as the raw file was imported into Lightroom…



After my perspective correction, cropping, white balance equalization and dynamic range compression, we get this…



Speaking of white-balance, all the beam-shot photos were locked to a custom white balance when shot and then set to an identical white-balance in post, specifically 4400K with a +47 magenta bias. To my eye, visually, my EagleTac MX25L3 with its gorgeous Cree MT-G2 LED provides the most neutral white beam of all my lights. It is not a particularly high CRI LED but it is rated by Cree as having a 5000K colour temperature. Of course, Kelvin ratings won't tell the whole story as far as tint either (hence the +47 magenta bias), but in the end, visually I felt the MX25L3 was very nearly "pure white" so I used the area just outside the bright core of photo 5 as a click-white-balance point in Lightroom, giving the 4400K+47 setting. I then synchronized all the beam-shot photos to that same white balance.

Note that on the neutral-patch of an Xrite ColorChecker Passport, a click white-blalance of my MX25L3's beam will give a white balance of 4950K +45 magenta, so very close to Cree's 5000K average Kelvin rating, with a slight green bias, requiring the +45 magenta correction. The 4400K here was likely due to the ceiling being slightly off-white. Also note that full noon sunlight comes up at right around 5000K as well, with a minimal green/magenta bias (+7) when you do a click balance on-the ColorChecker with my Fuji camera. Since we're also relying on the rendering of the camera, this won't necessarily be identical to what a calibrated colour-meter would read from the same light source either.

As far as colour accuracy, on my NEC SpectraView PA271 display, hardware-calibrated to 5000K, the beam-shot colours in the the photos are generally quite accurate to what I see visually in real-life, apart from a slight boost in saturation detailed below. The main exception is the very core of the brightest light, the SupBeam X40. The highlight recovery done in Lightroom will often not render accurate colour in the very brightest, partially clipped areas. The next darker exposure I did for the X40 was a little too dark to show the spill pattern as well as I would have liked. Note that the vast majority of non-calibrated displays are too cool (a bluer colour temperature) and also have too much contrast, so how accurately you see these beam-shots depends on the quality of your monitor and how well it is calibrated.

In post, I did push up the saturation of the beam photos in Lightroom slightly (+10), in order to show the colour differences within each beam more obviously. To my eyes, the tint differences are definitely a little more subtle visually than in these photos, but not by all that much. In addition, the beam artifacts are also slightly more noticeable in the photos than they are in real-life. Also keep in mind, that any one light when seen on its own, generally appears more white than what these photos suggest, mainly because our brain is very good at auto-white-balance! However when you shine two lights side-by-side, then suddenly any tint differential between the two becomes very obvious.

Anyway, to summarize all this, basically I have tried to portray the beam tints, tint variations and artifacts reasonably accurately but slightly enhanced in such a way as to be a little more obvious too. Mostly, this gallery is meant more to be interesting and pleasing to look at, and not to be a scientifically accurate comparison! Also, photographic colour management is not an easy task either and as mentioned, what looks accurate on my display very likely may not on yours. However the differences between the various lights, presuming you have a half decent display, will hopefully come across fairly well.

As far as the setup, the lights were all shot perpendicular to and 50 inches away from the ceiling, give or take a 1/2 inch. There was a very small amount of ambient incandescent light, filtering up from a stairwell. This had no effect on the majority of the images but I did switch the light off for photos 13 and 14, the secondary colour LED of the SRT5, since that LED is extremely dim and the incandescent ambient light was quite visible in those much longer exposures. So, that's about it for the setup overview, now on to the details of all the photos…

Specific Photo Descriptions

For simplicity I will leave off the "XE2_BCxx" for each shot. That naming convention I use refers to the camera used (XE2), where it was shot (British Columbia) and the year (xx where 13=2013) for example. Manufacturer's links, where available, are included for all the lights as well.

1) 11326: A group shot of all the lights, filters and diffusers I used in the photos. See the end of this section for a description of which light is which.

2) 11072: An uncorrected ceiling shot.

3) 11072: The perspective-correction and cropping used for all the ceiling beam-shots applied.

4) 11075: SupBeam X40 (3x Cree XM-L2 Cool White), max output rated at 3480 lumens.

5) 11340: *new* Thrunite TN32 (1x Cree XM-L2 Cool White), max output rated at 1702 lumens.

6) 11087: EagleTac MX25L3 (1x Cree MT-G2 5000K Neutral), max output rated at 2750 lumens.

7) 11251: EagleTac MX25L3 with diffuser supplied in filter kit.

8) 11256: EagleTac MX25L3 with 58mm Hoya #85 warming filter. The MX25L3 has standard 58mm threads and can accept any photo filter of that size. Nice!

9) 11261: EagleTac MX25L3 with red filter supplied in kit.

10) 11267: EagleTac MX25L3 with green dichroic filter supplied in kit.

11) 11273: EagleTac MX25L3 with blue dichroic filter supplied in kit.

12) 11275: EagleTac MX25L3 with blue dichroic filter supplied in kit. This shot with boosted exposure so you can see the surrounding purple fringe spill caused by the filter. Note that the beam core's colour is totally overexposed and wrong in this shot. The green filter also has a similar effect.

13) 11103: Nitecore SRT7 (1x Cree XM-L2 Cool White), max output rated at 960 lumens. This shows the main, white LED.

14) 11119: Nitecore SRT7, max output with an ArmyTek diffuser in-place. This diffuser can be seen at the very front of photo 1, just right of centre.

15) 11115: Nitecore SRT7, three secondary tricolour LEDs outputting red.

16) 11116: Nitecore SRT7, secondary LEDs outputting green.

17) 11118: Nitecore SRT7, secondary LEDs outputting blue.

18) 11129: Nitecore SRT5 (1x Cree XM-L2 Cool White), max output rated at 750 lumens, main LED. This shot with a home-made flip-up diffuser attached but opened. Some of the beam's peripheral area spill is cut off by the diffuser body as you'll see in photo 19.

19) 11124: Nitecore SRT5, main LED diffused, with flip-up diffuser closed.

20) 11134: Nitecore SRT5, main LED with flip-up diffuser removed, so no spill cut-off.

21) 11147: Nitecore SRT5, secondary LED outputting red.

22) 11148: Nitecore SRT5, secondary LED outputting blue. No green option as with the SRT7. The secondary LED is extremely dim as well, far dimmer than the three secondary LEDs of the SRT7.

23) 11151: ArmyTek Predator Pro v2.5 (1x Cree XP-G2 Cool White), max output rated at 670 lumens. As you can see, the Predator has an extremely well focused beam with a great deal of throw for its compact size.

24) 11164: ArmyTek Predator v2.5 (1x Cree XP-G2 Warm White), max output rated at 640 lumens. Besides the warm LED, the regular Predator does not have the same degree of mode or regulation customization that my cool-white Pro version has, although its colour is a little more suiting to light-painting in some situations.

25) 11183: Zebralight SC52 (1x Cree XM-L Cool White), max output with 14500 Li-Ion at 500 lumens.

26) 11228: Filzer I-BEAM X4 (1x Luxeon Rebel 90), max output rated at 100 lumens. Beam can be focused with a TIR optic and this photo is at its tightest setting.

27) 11231:  Filzer I-BEAM X4, this photo being at its widest (defocused) setting.

28) 11204: ArmyTek Wizard Pro headlamp (1x Cree XM-L2 Cool 5500K), max output rated at 1010 lumens. Wide-angle TIR optic with a 70º beam spread.

29) 11210: ArmyTek Wizard Pro Wide headlamp (1x Cree XM-L2 Cool 5500K), max output rated at 1010 lumens. Wide-angle TIR optic with a 90º beam spread. Beats me why this wide version is listed as discontinued? As much as the difference seems subtle, in actual use, the wider more diffuse beam of the wide version is much nicer than the regular narrower beam. There is talk of a refined UI coming, so maybe they will give it a new name?

30) 11235: Big blue lead-acid Canadian Tire brand "3.5 million candlepower" rechargeable spotlight, this photo with its 21x unknown, cool-white (almost bluish) LEDs in a "shower head" configuration switched on.

31) 11240: Big blue spotlight: this photo with its single halogen incandescent spotlight active. It seems to have a surprisingly tight focus for such a wide, shallow reflector, but in practice the beam diverges more than it seems here at longer distances and is riddled with artifacts. I brought up the brightness of this shot quite a bit in order to show these artifacts. In reality, the beam spreads a little more than the ArmyTek Predator v2.5 for example and has maybe 60% of the total lumens output of that much smaller, single LED light. I included this light to demonstrate how far modern LEDs have come. I also have a 3x D cell halogen MagLight and it is far dimmer than this spotlight too with even worse focus. I could't be bothered to put fresh batteries in it for this test.

32) 11279: 4Sevens Quark Mini 123 (1x Cree XP-G), a tiny little keychain light with max a output rated at 189 lumens.

33) 11294: Generic green laser, probably 5 mW output.

34) 11304: Laser beamshot, yeah! However… it's pretty dim... but reasonably safe to use at least.

35) 11329: In case anyone was wondering, shot one was taken with this setup. White seamless paper as a background, a medium soft-box and a Profoto strobe. The quartz modelling light is on, which is giving the warm light. The strobe, when fired, is daylight balanced.

That's it, whew!  A few additional comments…

Most of these LED flashlights have seen some light-painting duty and none of them have a high CRI (colour rendering index). The best of the bunch is definitely the 5000K EagleTac MX25L3, but even though its beam is quite neutral, it still requires a customized calibration with an Xrite ColorChecker Passport for proper colour rendition when being used to light a photograph. However as mentioned before, all of these lights when used as flashlights, put out quite a nice quality beam (even the big spotlight is okay) and at night with dark-adapted eyes, the brightest ones are quite staggering in their brightness and throw! The variances in beam tint seen in these photos are also not at all very noticeable in actual use.

Here is the list of lights from the first photo, the back row starting from the smallest on the left: 4Sevens Quark Mini 123, Zebralight SC52, NiteCore SRT5, ArmyTek Predator Pro v2.5, ArmyTek Predator v2.5 (warm), NiteCore SRT7, EagleTac MX25L3, SupBeam X40 and the big blue spotlight. The second row, starting on the left: ArmyTek Wizard Pro Wide (with headband), ArmyTek Wizard Pro (with pocket clip), EagleTac green, blue and red colour filters, the 58mm Hoya 85 and the EagleTac diffuser. In the front, starting on the left: my home-made flip-open diffuser for the SRT5 (a Butler-Creek clear scope cover with diffusion), the ArmyTek diffuser used on the SRT7 and then the little white-bodied green laser.

On January 6, 2014 I updated the beam-shot gallery to also include the impressive new Thrunite TN32. While it isn't the brightest flashlight I own in terms of total lumen output, it is by far the most focused, with the brightest beam core, the least spill and is the farthest throwing by a long shot. For an outdoor beam-shot comparison at a whopping 400 meters, as well as a photo showing the TN32 compared to a few of my other lights, see this blog posting here: Thrunite TN32 comparison...

Lastly, an astute flashaholic might notice the Predator Pro v2.5 does not have a crenelated bezel and they would be correct. It is actually sporting the non-crenelated bezel from an ArmyTek Viking Pro, which does indeed fit!

Anyway, that's it for now. A whole lot of flashlight geeking out! If there is any interest, I may decide to post a more "scientific" and less "aesthetic" gallery with beam-shots that have totally consistent processing so that accurate brightness comparisons can be done. However for now, that is definitely it…

(and I'll be amazed if anyone actually reads this to the bitter end!)

1 comments:

Post a Comment