HMC colorblind to red/amber LEDs?

[MichaelTiemann]

Here's a picture taken with a Canon EOS-1D showing the color-blindness of my HMC150 video camera. Note that I've zoomed in on a pair of LEDs and the vectorscope shows virtually no color difference between them, just as the LCD finder also shows virtually no color difference between them. I set the shutter speed and aperture so that the color reading on the vector scope doesn't go out of bounds, and I've highlighted the two regions to compare:

Here's the complete photo for context:

In the photo you can see there is a large range of red, amber, yellow-amber, and even yellow LEDs. They all come out virtually the same color on the HMC150. The HMC150 can see green and blue just fine, but it cannot seem to distinguish anything that's on the orange side of yellow, one from another.

Should I expect any pro video camera to be able to capture these colors as accurately as my DSLR does, or is this one of those cases where the only hope is DSLR technology?

 

[MichaelTiemann]

I guess nobody but me wants to film electronic equipment

 

[matt s.]

my question is why are you shoting with such a high shutter speed and almost closed iris for this shot?

 

[MichaelTiemann]

my question is why are you shoting with such a high shutter speed and almost closed iris for this shot?

I chose an exposure that made the vectorscope readout most clear. If I open it up, the color goes beyond full scale and starts arcing in strange directions.

 

[MichaelTiemann]

OK...150+ reads and no other responses? Anybody have similar experience with neon signs? LEDs at nightclubs, etc.?

 

[ChrisPrine]

No idea, man. I just wanna know what that crazy modular synth is.

 

[MichaelTiemann]

No idea, man. I just wanna know what that crazy modular synth is.

If the colors were better, you would instantly recognize it as a Buchla 200e ;-)

 

[studio1972]

I've had problems with purple dresses looking blue, I think it's to do with the colour space. Not noticed a problem with orange and red though.

 

[Rick Burnett]

Haha, I do not have that camera, but have to agree, nice modular synth! Although it seems to be missing about 304850 patch cords hooked up!!! (And I kid)

 

[ParrotEyes]

Some stuff that came up with a quick Google Search...

"CCDs have a very high sensitivity to light beyond 700nm where they eye and film are insensitive." (Think infrared)

And, from the Peter Gray.org website...

Why light meters can fail with CCD imagers:


This is my (tentative) theory about why light meters can fail to give correct readings when shooting High Definition (or video in general). Film emulsion are far more blue and ultraviolet sensitive compared to the other colors in the visible spectrum. So manufacturers modify film emulsions to try and get it to see all colors, more or less, equally well (or well enough to get a reasonably rounded image with an acceptable color rendition).

A CCD seems to be the opposite. Adam Wilt writes in his article, "Chip Off The Old Block", that...."CCD's do have a spectral sensitivity, and its more heavily weighted toward infrared than it is toward blue. Each CCD can be optimized for the color it will "see" by using chemical doping to alter its spectral sensitivity" (DV magazine, December, 2000). And this seems to be confirmed when setting black levels on a waveform monitor with the iris fully closed or the lens cap on. The blue signal is considerably more noisy compared to either the red or green signals. So this seems to confirm CCD's are more red and infrared sensitive compared to the other colors in the visible spectrum. And manufacturers try to optimize a CCD to record all colors more or less equally well (or again, well enough to get a reasonably rounded image with an acceptable color rendition). So similar to film emulsions, but the other way around.

Now here is the rub, light meters are configured to see light pretty much as a film emulsion responds to light. So they are biased in the wrong direction for CCD imagers.

In practice, you tend to encounter problems using a light meter to set exposures for a CCD camera with daylight exteriors (especially bright exteriors). Comparatively speaking, there is a lot more blue-wavelength light in the visible spectrum that you are trying to record in this situation. The light meter either has a similar sensitivity to the visible spectrum as film, or it is configured to compensate for this "extra" blue light in terms of the overly blue-sensitive film emulsion. But either way, this is the opposite of the requirements of a CCD imager. So while a light meter measuring a day exterior is giving a good exposure indication for film, it is giving a comparatively poor exposure indication for a CCD.

The light meter, remaining faithful to the needs of the cinematographer, prompts you to reduce relative exposure for daylight exteriors to protect against overexposing the overly blue-sensitive film emulsion. On the other hand, the CCD, with its relatively poor blue sensitivity, requires you to increase relative exposure for daylight exteriors to avoid underexposing the overall image. This is what happens in practice. In other words, your light meter is a tool that will tend to let you down if you are using it in this type of situation (daylight exteriors) to help in determining correct exposures for CCD cameras. It works well for cinematographers, but fairly poorly for videographers. A waveform monitor on the other hand, will always given a better indication regardless of the relative amounts of red, green, and blue light being captured and recorded.

 

[WpgCameraMan]

I would like to see this with the camera at a shutter of 1/60. You could dim the LED's using full neutral density on the lens, close the iris a little.

Another way is to turn off room lights, hold a white sheet of paper over the panel face. This will spread the light over a larger surface and you might be able to make out the different tones better.

And what white balance are you using? That could easily create this effect. White LED's are 6000K+ , the coloured ones might exist in some weird colourspace that CCD's can't resolve properly.

Also LED's are very bright compared to surrounding areas, once that contrast get to a certain point, you can't trust a video imager to capture things accurately.

 

[MichaelTiemann]

I would like to see this with the camera at a shutter of 1/60. You could dim the LED's using full neutral density on the lens, close the iris a little.

I had the ND at 1/64 and the iris at f/11, and the LEDs were still blowing out the color. Opening up to 1/60 just makes things worse.

Another way is to turn off room lights, hold a white sheet of paper over the panel face. This will spread the light over a larger surface and you might be able to make out the different tones better.

I tried this. I didn't save the footage, but there is no difference at all when it comes to answering the question "what color is it?".

And what white balance are you using? That could easily create this effect. White LED's are 6000K+ , the coloured ones might exist in some weird colourspace that CCD's can't resolve properly.

I did a WB using a ChromaDuMonde chart and also shot using the Daylight and Tungsten presets. I also played with Color Temp between -7 and +7 (and of course in between) and was never able to get the Red and Amber LEDs to hit the vectorscope in distinct locations. That is, I could move them perhaps 30 degrees from one extreme to another, but never were they one degree distinct one from another.

Also LED's are very bright compared to surrounding areas, once that contrast get to a certain point, you can't trust a video imager to capture things accurately.

That's precisely why I shot with the ND, Iris, and shutter as I did, to minimize the apparent contrast (stopping down black doesn't make it any blacker). I also used about 1KW of tungsten lamps to try to brighten the background to match up with the LEDs. That didn't help either.

 

[lakotawolf]

I film breaking news video and and yes red and amber flashing lights look the the same color. it really pissed me off at first but i have learned to live with it. Unless there is a fix.

 

[MichaelTiemann]

I film breaking news video and and yes red and amber flashing lights look the the same color. it really pissed me off at first but i have learned to live with it. Unless there is a fix.

Ugh. Do any video cameras get this right (like HPX370, HPX500, etc.), or does one have to use video modes of DSLRs to get accurate color information in the reds and purples?

 

[MichaelTiemann]

Aha! This article by Art Adams about the AF-100 explains what might be going on and why the HMC 150 is colorblind to my red LEDs. Looks like there's no choice but to upgrade to Varicam (or leave the family).

Bummer!