Comparing LED video lights

I went into the shop today to evaluate some of Video Gear's new LED lights and see how these compare to the industry standard HMI.  In the list were three lights from Litepanels, two lights from Film Gear, and one from newcomer Ushio.

---

---

There were really two parts to this evaluation:  First is total light power, second is color rendering.  In order to evaluate light power I set the light 10 feet from the test chart and measured exposure using my Sekonic light meter at 1/50th, ISO 100 - just because it seemed reasonable.  If the light beam spread was adjustable I measured in both spot and flood mode to see the range of exposure I could expect from the light.

For color rendering I set a Macbeth Color Checker 10 feet directly in front of the light with my camera set to a white balance of 5500k.  Each of the lights was theoretically a 5500k source according to the manufacturers, so this seemed like a way to at least take relative color measurements.  I obviously don't have a colorimeter that can measure the exact correlated color temperature (CCT).  The reason I say CCT is that LED's have a discontinuous spectrum that severely lacks cyan so you don't expect an LED light to render color as well as a tungsten nor HMI light.

1.  Litepanels Sola 6.  This is an LED based Fresnel with a built in lens that goes from spot to flood.  The power supply is housed just under the light (where you see the "Lite Panels" logo in the picture).  The light is dimmable and does not produce flicker nor audible noise except for a whisper quiet cooling fan.  No gloves are required since the light fixture only produces a small amount of heat.  In my measurements I found at the conditions listed above the light was able to get an exposure of f/4.0 to f/1.4 (spot to flood).  Not enormous amounts of light, but more than sufficient for the new line of native ISO 800 cameras.

2.  Litepanels Sola 4.  This little brother to the Sola 6 is equivalent in feature set, but is significantly smaller and obviously has less light output.  I was able to get exposures of f/2.8 to f/0.7(.8).  Again, this is about a two stop difference, which tells me that the two models probably share a similar design.

3.  Litepanels 1x1.  This is the industry workhorse when it comes to LED lights.  However, with this light you get a softer spacelight with no focus adjustment (at least on the model I used).  Since the dimmer goes down to zero output I was only able to evaluate the light at full output.  My exposure on this light was f/1.0(.6), a little more than one stop more than the Sola 4 in flood mode, but 1.5 stops less than the Sola 4 in spot mode.  In my opinion that's why the Sola's are worth the money - versatility.  You have to use an crate over this light to turn it into a spot and that only loses output.

4.  Ushio Pro Panel 1x2.  Ushio is the new kid on the block.  This panel features the ability to not only dim, but to set the color temperature from 2700k-6500k via an attachable control module.  For my tests I set the color temperature to 5500k.  Since this light only has a flood mode I found the maximum exposure I could get was f/1.0(.8) - essentially the same as the Litepanels 1x1.  Although the light fixture is double the size and obviously double the softness given the large diffuser. 

5.  Film Gear LED Flo (2ft x 4 bank).  This is a unique light that emulates the Kinoflo lights, but uses tubes of LEDs.  You can easily swap out the daylight tubes for tungsten balanced tubes.  The setup I was using came with a dimmer that allowed me to switch on the lights in two banks.  You can assign the banks however you wish since the connections to the LED tubes are arbitrary and user assignable.  My full output exposure reading on this light was f/1.0(0.3).  That's 1/3rd stop below the 1x1 and a half stop below the Ushio panel.

6.  Film Gear Daylight Fresnel HMI 575W.  I measured this light mainly as a reference since HMI's are industry standard.  What the photo doesn't show is that the light comes with a set of flags and a set of 5 or 6 (I don't remember) glass lenses that are put over the front of the light.  This light gets HOT!.  Not tungsten-immediately burn your fingers off hot, but still HOT!  Gloves are still recommended.  The fact that it doesn't have a built in Fresnel lens is less handy, but it does allow for focus adjustment of the PAR via a knob in the back.  Because this light has a number of external glass lenses I felt is was necessary to measure the output with more than one lens and also at spot and flood focus.  With the 10-degree lens the output measured f/11(.5) to f/2.8(.5).  With the 50-degree flood lens the output ranged from f/5.6(.5) to f4.0(1).  With the frosted diffuser the output was f/4.0(.7) to f/2.8(.7).  That's A LOT more output than any of the LED lights.  In fact, I took the glass lens out and measured the raw output...f/22(.2)!  Now that's a real light source.

Next, let's look at color rendering.

---

---

First, we'll start off with white balance.  Keep in mind that I was using a standard setting of 5500k on my camera.  The HMI light came out pretty much neutral at that setting, however we are looking at the relative color rendering since cameras can be custom white balanced.  With that said, it's also convenient to have a light that can mix with daylight and not cause a color shift.

Let's start with the neutral patches.  HMI has a slightly blue shift compared with absolute neutral.  The Sola 6 is lacking green, creating a magenta shift.  The Sola 4 has a much lower lack of green, so its magenta shift is fairly minor.  The Litepanels 1x1 and Ushio Pro Panel both have a large shift toward red.    The Film Gear LED Flo has a very, very minor shift toward red.  So the Sola 4 and Film Gear LED Flo seem to have the best neutrality of the bunch.  However, this neutrality is all relative.  I was surprised to see a difference between the Sola 6 and Sola 4.  It seems like they would be essentially the same technology.

---

---

Next, we'll look at the top line of the chart *after* a custom white balance on the neutral patches.  The top line of the chart is arguably the most important because if a light can't render skin tones properly then it's nearly unusable.  After the custom white balance the color rendering matches reasonably well with minor shifts.

---

---

Then finally we'll look at lines 2 and 3 of the Macbeth chart.  I see some slight differences on line 2 in the way orange is rendered.  Green seems to vary a bit as well.  Nothing hugely alarming here, but the differences do exist.

On line 3 I see some slight differences in the primary and secondary colors.  I expected to see some darkening in cyan since white LEDs lack cyan in their output spectrum, but so do the color filters on CMOS sensors.

---

 

---

If you're looking for a winner here, well there isn't one exactly.  The lights all work well and can produce nice results given the correct DP.  There are differences between lights, but the only colors that can easily be identified as wrong are skin tones.  Those seemed to look OK with a custom white balance.  Now if you're going to mix and match lights, this is an entirely different story.  The neutral patches above somewhat reveal which lights you should and shouldn't mix and not drive your color grading person nuts.

As always, if you have questions please call the shop.  We're here to help.