Welcome to our community

Be a part of something great, join today!

Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

Led mystery solved making way for high output fixtures.

Terry VerHaar said:
But when we shoot raw we aren't really "post wb'ing" are we? If we shoot a gray card under a tungsten light that is rated at 3200K and it turns out to be actually 2800K, by aligning the waveforms of all three sensors in RCX with the eyedropper on the card, we are not "post white balancing" as much as we are discovering the true color temp of the light. After all, with raw, WB is only metadata.

Now, if we do the same procedure but shoot the grey card under an LED rated at 5000K, the eyedropper will align the three (RGB) waveforms again - to give a true WB temp - but it might also change the tint value based on other colors present. So, I am thinking the tint adjustment will somewhat (but not perfectly) compensate for the color spike. Again - not the same as having a light with an even, full spectrum output but how far off would it be?
Click to expand...

There are 4 variables at work here:
1. the spectral emission of the light
2. the spectral reflectivity and absorption of what you're filming
3. the spectral response of the combination of the lens, the camera sensor and the Bayer filter (how pure it is, what spectrum bands it lets through, crosstalk, etc)
4. the post color processing

So given the same light, how "far off" the color is would vary depending on the subject matter, the camera and the color post-processing. Too many variables to give you an exact answer, sorry!

For example: human skin has a pretty complex spectral response. The surface has a certain spectral graph, and blood has another! (technically oxygenated and de-oxygenated blood differ too) We are super-evolved to spot these changes because you can judge something like a person's health from it. If you had a camera that missed subtle tones in human skin anyway, maybe you wouldn't notice the difference. If you used another camera and processing chain, maybe through pure bad luck the spectral responses would line up in a really upsetting way. Then there's makeup, etc.

On the other hand, a ColorChecker / Macbeth chart has pretty nice-looking big sine-waves for the color. So it will probably handle weird lighting better.
http://www.keyesphoto.com/KdktechMacbethGraph.html

Interestingly, the ColorChecker / Macbeth chart's skin patch is rather different to real skin, but a pretty good attempt:
http://www.babelcolor.com/main_level/CC_light-skin_spectral_type.gif

You can see some spectral reflectivity values for real skin types here:
http://www.image-engineering.de/index.php?option=com_content&view=article&id=544&Itemid=145
They also point out how completely different the green of a palm tree is to either of the greens on the ColorChecker!

Anyway, this is why those guys doing big feature films do tests using subject matter and equipment and workflow closest to what they need. Too many variables to figure out a generalized answer! Except that spiky is bad. And human art can overcome most issues so also don't worry too much.

Bruce Allen
www.boacinema.com
 
I am just trying to look the color quality of the light itself. On that count, reflective qualities aren't one of the variable, are they? Obviously a material reflects given wavelengths or it wouldn't emit a color at all (or even be visible). Neither are the spectral response or the post color processing issues when just looking at the quality of light itself. Yes, those vary with every specific light source, every sensor, every reflective surface and every color correction algorithm; too many variables to work through for any focused discussion.

But, isn't it true that you can determine how "far off" the light is when shooting a perfect 18% grey card? The reflective surface is a controlled variable, pretty consistent. When we are talking a high quality sensor with sophisticated color science backing it (RED), we are also talking consistency.

So, if you shoot the card under sunlight and get metadata from your Epic that shows pure grey as grey at 6400K, then isn't it true that the color temp of the light source is, indeed, 6400K - no matter how it reflects off any other particular surface? I realize your specific sensor may not be perfect but we do rely on them to be somewhat consistent and could consider the sensor as something of a controlled variable. And yes, we know that sunlight is full spectrum so we don't have any quirks there from which to adjust.

Using the example of a full spectrum phosphor light source from Cineo TruColor (PRG), the full spectrum is there but there is also a relative abundance of blue. That said, we could still shoot the grey card and get a proper WB metadata value from the sensor as processed in RCX. Perhaps with a little "help"my the tint slider.

So, the last example is just looking at a less "perfect" light source like LED - with an extra bit of green from the light source. Again, a perfect 18% grey card, a consistent performance by the sensor.

What I am asking, or proposing maybe, is that, in the absence of perfect full spectrum sunlight - which is by nature the "standard" for humans to measure/compare light - will a combination of sophisticated system of sensors and color science algorithms allow us to utilize these less perfect lighting instruments and still get good results. Certainly there are added benefits to be gained as well, in the form of power savings and avoidance of heat.
 
The Chinese have been pumping out high output LED bulbs for quite a while now. Not sure how this "mystery solved" is news but I've swapped all my bulbs out with inexpensive high output LED bulbs from Home Depot and Lowes and it's a lot nicer and easier on my eyes.
 
Interestingly, I was using some overhead fixture bulbs as practicals recently and noticed some flickering in the footage. I thought LEDs were supposed to dispense with flickering. Anyway, I recently saw some more LED bulbs that claimed *Constant light stream – no high frequency flickering (Protection for eyes). Does this sound like something new and different from older bulbs (I've had the others for about a year or longer?)
 
Terry VerHaar said:
I am just trying to look the color quality of the light itself. On that count, reflective qualities aren't one of the variable, are they? Obviously a material reflects given wavelengths or it wouldn't emit a color at all (or even be visible). Neither are the spectral response or the post color processing issues when just looking at the quality of light itself. Yes, those vary with every specific light source, every sensor, every reflective surface and every color correction algorithm; too many variables to work through for any focused discussion.

But, isn't it true that you can determine how "far off" the light is when shooting a perfect 18% grey card? The reflective surface is a controlled variable, pretty consistent. When we are talking a high quality sensor with sophisticated color science backing it (RED), we are also talking consistency.
Click to expand...

For a spiky light, no, it is not. I will try to draw a diagram.

EDIT: OK, here you go, Terry! Also... "sophisticated color science" is not necessarily a good thing! Maybe the sophistication is compensation for issues in the sensor and color filters! In microphone terms: I'd prefer a good microphone through a clean preamp with a little compression... than a bad microphone sent through a warming preamp with a whole lot of audio post stuff added.

lightcolorscience.jpg


Bruce Allen
www.boacinema.com
 
IF there was a RedUser awards program (the REDheads?) Bruce Allen's post above would have to be a nominee. For the legions of folks who don't care to read the likes of Charles Poynton (informative but dense) Bruce's graphic is gold.

FWIW, I always remind myself that what we are photographing are the reflected photons and that what the sensor "sees" is the result of rather complex equations that include optics and sensor design.

As a side note, having spent way too much time in post "fixing" green spikes in particular, never to complete satisfaction, might I suggest that the time it takes to throw a minus green gel on an offending source is worth every second. Just sayin'

Cheers - #19
 
Dear lord Bruce you've really outdone yourself on that one. You've managed to be both scientifically accurate while remaining simple and easy to understand. Not an easy feat. :cheers2:

To extend this conversation a little further here is a great news story that also applies:
http://www.npr.org/blogs/krulwich/2...d-it-to-pluto-but-not-to-pink-please-not-pink

Pink isn't a frequency. It's just "Not Green". So if your lighting was really spikey your white card could easily come out as white and your camera could be incapable of capturing pink depending on how much of the two ends of the spectrum an object reflected.
 
L. Langer said:
The Chinese have been pumping out high output LED bulbs for quite a while now. Not sure how this "mystery solved" is news but I've swapped all my bulbs out with inexpensive high output LED bulbs from Home Depot and Lowes and it's a lot nicer and easier on my eyes.
Click to expand...

Not the same thing. Most all of those bulbs are using Philips or Cree LED units in them. They can claim up and down that these are "high output", but these in no way have solved the issue of "droop". If the issue is truly solved, then that 9.5W LED that outputs 800 lumens will become something else. They still generate a fair amount of heat, in fact, some of them get pretty darn hot. A huge percentage of their power used is still converted to heat. We solve "droop" and instead of ~800 lumens, that 9W LED will be coughing out 4500 lumens and be cool to the touch. Or, you'll be able to get 800 lumens out of 1.8 watts and it will never go above room temperature.

That said, I've been converting to LED myself. They're getting darn good. The CREE (Ecosmart and Halo) recessed can inserts are awesome. Instant-on, which many of the bulbs are not and smoothly dim with most regular dimmers. They have a lower rated light output, but actually throw more usable light than most of the PAR style flood lamp LED bulbs. The 75W equivalent LED bulbs (12W?) are pretty darn good all the same. And the new CREE 60W equivalent 9.5W bulb that looks mostly like an incandescent is killer - blue package, home depot exclusive. They can't keep them in stock around here. They're $12, but output more even and brighter light than most 60W incandescent I'm aware of. And they're $12. Love 'em. Nice to get away from the power-hungry incandescents -- I've seen a noticeable drop in my utility bill over the past year as I've swapped out lots of bulbs. And CFL's are terrible. They don't tend to last any longer than incandescents, usually less time, nowhere near what they're rated. And their toxic and filled with lead and mercury so you don't want to break them. 1 out of every 10 cracks open when it burns out or blows a ballast anyway. I freaking hate CFL's and their slow warm-up times, flickering light, and I mentioned the toxic thing. People pushing for CFL's because they want to be "green" need to have their heads examined.
 
Bruce, if you have that image at high enough dpi for printing as a poster, I want it. It deserves a frame.
 
Blair S. Paulsen said:
IF there was a RedUser awards program (the REDheads?) Bruce Allen's post above would have to be a nominee. For the legions of folks who don't care to read the likes of Charles Poynton (informative but dense) Bruce's graphic is gold.

FWIW, I always remind myself that what we are photographing are the reflected photons and that what the sensor "sees" is the result of rather complex equations that include optics and sensor design.

As a side note, having spent way too much time in post "fixing" green spikes in particular, never to complete satisfaction, might I suggest that the time it takes to throw a minus green gel on an offending source is worth every second. Just sayin'

Cheers - #19
Click to expand...

http://www.reduser.net/forum/showth...or-high-output-fixtures&p=1195203#post1195203

Oh great, now, I have to beat Bruce by putting little German mountain climbers on his graphs and peaks, with speech bubbles to explain things, instead of just typing it. ;)
 
Who said that Reduser isn't a supportive community?

I am humbled that a bunch of genius guys (whose posts I love reading and I have learned a ton from) liked my drawing. Thank you so much for taking the time to write that. You made my year!

Yes, I will fix the smooth / spiky typo - and post a vector version, Jeff :)

Bruce Allen
www.boacinema.com
 
Bruce - I was buried in a project for a few days and come back to see this! Wow - pretty amazing. I simple "Thanks!" would be less than you deserve for doing this but it is all I have. It is a really generous gift of time and knowledge that you were willing to give. Again, wow.

It serves as a great basis for good discussion, too, since it very clearly shows the differences between correcting for full spectrum light and any lights with spikes in the output. It does give me a little hope, however, insofar as it would suggest that secondary color correction, for the reflections that aren't accurate, could help somewhat.

Also - I am still stuck on the idea that, when the software is adjusting for the proper color on the grey card, it is likely to add an offsetting tint (magenta, in the case of the green LED spike) to the image. I have seen that in RCX many times. If it does that, won't that somewhat help bring things back into line - with secondary being the next step? Clearly, the horrible multi-spiked light shown in your diagram would probably be a lost cause, but if you look at the real light spectrums of the PRG Phosphor lights (single blue spike) or some high CRI LEDs (single green spike) the correction might be more manageable. No?

Again, thanks for advancing the discussion by leaps and bounds!

BTW - I think the label on the lower graphic is a typo - under what looks to be a "smooth light" graphic it says "spiky light."
 
You must log in or register to reply here.
Back
Top