Acceptable CRI?
- Thread starter DC EX1
- Start date
Chris Santucci
Veteran
It really depends, but generally speaking high 80's and up is preferred.
.
.
Barry_Green
Moderator
Kino Flo, the professional standard, is 95. Lowel lamps (daylight, at least) are 90+.
Anything below 90 would be ... less desirable.
Anything below 90 would be ... less desirable.
Chris Santucci
Veteran
I'd not touch anything below 90 for film, however.
Check this out:
Candle: 1700k 100 CRI
High Pressure Sodium: 2100k 25 CRI
Incandescent: 2700k 100 CRI
Tungsten Halogen: 3200k 95 CRI
*Solux Bulb: 4100k 98 CRI
Cool White: 4200k 62 CRI
*Ott-Lite™ Pro: 5000k 82 CRI
Clear Metal Halide: 5500k 60 CRI
*Verilux® "Natural Spectrum®": 5500k 82 CRI (also called HappyEyes® and Trucolite Phosphor Technology™)
Natural Sunlight: 5000-6000k 100 CRI
.
Check this out:
Candle: 1700k 100 CRI
High Pressure Sodium: 2100k 25 CRI
Incandescent: 2700k 100 CRI
Tungsten Halogen: 3200k 95 CRI
*Solux Bulb: 4100k 98 CRI
Cool White: 4200k 62 CRI
*Ott-Lite™ Pro: 5000k 82 CRI
Clear Metal Halide: 5500k 60 CRI
*Verilux® "Natural Spectrum®": 5500k 82 CRI (also called HappyEyes® and Trucolite Phosphor Technology™)
Natural Sunlight: 5000-6000k 100 CRI
.
Chris, I wonder about the source of that table. How can a candle, or household incandescent, render blues accurately when there's so little blue in them? How can tungsten halogen not be 100 when film is made to render colors perfectly under it?
Chris Santucci
Veteran
Good question. Here's where I grabbed that from. There's an explanation on there:
http://www.fullspectrumsolutions.com/cri_explained.htm
.
http://www.fullspectrumsolutions.com/cri_explained.htm
.
Good info... the reason I ask is because I'm working on fashioning some cheap lights I can use on the road and replace bulbs etc for cheap... I've found a couple 84 CRI flo bulbs... but I think I'll just end up spending the money on TD5 Spider lights... Anyone had an experience with the Spider lights?
gco
Veteran
So much for my idea of re-purposing all those surplus sodium street lights I got a "deal" on!:violin:
Sorry, no experience with Spider lights, and I hate to go against the orthodoxy but, (shhhh, 84 CRI isn't too bad.) If they're not expensive why don't you get one and shoot a bunch of tests? Be sure you do some faces of various complexions. I suspect you'll be pleasantly surprised.
What's your source, Home Depot's store brand?
Otis Grapsas
Carbonite Member
A light source with low kelvin temperature can provide excellent blues, as long as the dynamic range of the camera is very high or the camera is optically balanced for white. If you light with a candle and use digital white balancing (not a filter) your blues will have 10 to 18dBs of extra noise compared to red, but the spectrum will still be quite smooth, which is what counts in CRI. Noise does not contribute much to the calculation until it becomes huge.
http://physics.nist.gov/Divisions/Div844/facilities/vision/color.html
Most cameras are balanced with smooth spectrums in mind, so they can have problems with others sources (CCFL, LED, etc) if their spectrum is not taken into account. Those who have a continuous Kelvin control (many stops) usually depend on simple models for the spectrum estimation, so they might be way off with these light sources. Your white will probably be accurate though, it's the color that will suffer.
Film is very happy when the CRI is high, but these days the color correction is so advanced that it can be fixed with presets that are part of a standard digital workflow.
If you design a camera for a given light source, or if your sensor happens to match a light source well, you can get very good results even though the CRI is small. On the attachment you can see a comparison of a given Kodak sensor using three light sources, the two fluos are far inferior in CRI compared to the incadescent, but they manage to outperform it on raw camera output. Which means that it will need less color correction in the camera.
http://www.dvxuser.com/V6/attachment.php?attachmentid=14836&d=1262662122
But that's not the whole story. Other colors can still suffer, even if the CRI is high and your primary colors look great, such as the skin tone region, because we are very sensitive to that. This is not taken into account by any color precision measuring tool and the skin tone region is quite small compared to the full color space of a camera. A camera with bad Imatest color difference performance can produce worderful skin tones and overall color. In-camera color correction design is an art in video and should be judged by aesthetics, not measurements.
A high end color correction scheme can theoretically use any light source with excellent results, in-camera if the light source was taken into account or in post with careful preset building based on tests and calculations.
http://physics.nist.gov/Divisions/Div844/facilities/vision/color.html
Most cameras are balanced with smooth spectrums in mind, so they can have problems with others sources (CCFL, LED, etc) if their spectrum is not taken into account. Those who have a continuous Kelvin control (many stops) usually depend on simple models for the spectrum estimation, so they might be way off with these light sources. Your white will probably be accurate though, it's the color that will suffer.
Film is very happy when the CRI is high, but these days the color correction is so advanced that it can be fixed with presets that are part of a standard digital workflow.
If you design a camera for a given light source, or if your sensor happens to match a light source well, you can get very good results even though the CRI is small. On the attachment you can see a comparison of a given Kodak sensor using three light sources, the two fluos are far inferior in CRI compared to the incadescent, but they manage to outperform it on raw camera output. Which means that it will need less color correction in the camera.
http://www.dvxuser.com/V6/attachment.php?attachmentid=14836&d=1262662122
But that's not the whole story. Other colors can still suffer, even if the CRI is high and your primary colors look great, such as the skin tone region, because we are very sensitive to that. This is not taken into account by any color precision measuring tool and the skin tone region is quite small compared to the full color space of a camera. A camera with bad Imatest color difference performance can produce worderful skin tones and overall color. In-camera color correction design is an art in video and should be judged by aesthetics, not measurements.
A high end color correction scheme can theoretically use any light source with excellent results, in-camera if the light source was taken into account or in post with careful preset building based on tests and calculations.
Chris Santucci
Veteran
Otis Grapsas
Carbonite Member
Well, thanks. I don't write articles, these days most of my technobabble related to imaging goes into this thread:
http://www.dvxuser.com/V6/showthread.php?t=132566&page=26
http://www.dvxuser.com/V6/showthread.php?t=132566&page=26
Cool Lights
DVXuser Sponsor
Thats interesting about candle light CRI 100, didn't know that. Its totally possible though since CRI and color temperature are not related. You could easily have a 10000K or even 15000K light at CRI 99, so why not a 1700K one. If you measure something in color temperature, its still considered white light. That white light either is farther to the red end of the scale or blue end. CRI is indirectly related to the green/magenta axis which has nothing to do with color temperature. The less green or magenta correction you need, the more in balance is the CRI.
We've tested high color temperature bulbs and LEDs and its not much of an issue. We're not talking about eyes and what they see directly here, its all related to what color temperature film you're using (if film based) or how well your custom white balance works if you're shooting digital mediums like video or stills. One way or another you always have a balancing feature of your medium and have to know how to use it to your advantage for both normal effects as well as special effects. Has nothing to do with your eyes other than how it relates to seeing the final product on film or video and how well you mastered that medium.
Next time I'm where we have access to an integrating sphere perhaps I'll take a candle along to test its CRI as it is interesting to know if fire is indeed that high or not. Probably needs to be a white candle as well so the colored wax doesn't contaminate the test so to speak. They may not be crazy about the soot going up on the roof and the candle wax going down on the floor of the sphere though!
You can test it yourself somewhat for that matter. Just light a color chart with a candle (or several), then custom white balance and see how well it does. If you were shooting film and wanted whites to be white, etc. though you would need a 1700K film which I doubt exists other than as a special order item and how many would need it anyway? Using 3200K film would obviously still give a very yellowish/red look but then again, that's what many would want when they were using candle practicals in a scene. Making it all look white or like lit with normal white light would not be the look they wanted for a candle lit scene.
We've tested high color temperature bulbs and LEDs and its not much of an issue. We're not talking about eyes and what they see directly here, its all related to what color temperature film you're using (if film based) or how well your custom white balance works if you're shooting digital mediums like video or stills. One way or another you always have a balancing feature of your medium and have to know how to use it to your advantage for both normal effects as well as special effects. Has nothing to do with your eyes other than how it relates to seeing the final product on film or video and how well you mastered that medium.
Next time I'm where we have access to an integrating sphere perhaps I'll take a candle along to test its CRI as it is interesting to know if fire is indeed that high or not. Probably needs to be a white candle as well so the colored wax doesn't contaminate the test so to speak. They may not be crazy about the soot going up on the roof and the candle wax going down on the floor of the sphere though!
You can test it yourself somewhat for that matter. Just light a color chart with a candle (or several), then custom white balance and see how well it does. If you were shooting film and wanted whites to be white, etc. though you would need a 1700K film which I doubt exists other than as a special order item and how many would need it anyway? Using 3200K film would obviously still give a very yellowish/red look but then again, that's what many would want when they were using candle practicals in a scene. Making it all look white or like lit with normal white light would not be the look they wanted for a candle lit scene.
Chris Santucci
Veteran
Yeah, that's it. It's interesting that it's common for cameras to have an adjustment for magenta/green and I always wondered why. But, you always lose somewhere anyway when trying to make adjustments while using low CRI sources.
I thought the candle rating was odd at first also, but isn't the standard radiator for CRI standards a "black body radiator?" Sounds a bit like a candle flame to me.
.
Cool Lights
DVXuser Sponsor
You may lose something but the results are often acceptable or can be further manipulated in post if necessary or if its really critical.
I think when they say "black body radiator" what they really mean is a piece of some kind of "theoretical" metal being heated from red hot through blue hot. Only two of the three sources I can think of that are full spectrum (tungsten, daylight and carbon arc) are close to what I would call a "flame state" and that would be daylight (the sun is a big fireball right?) and carbon arc (pretty good flame going on in there if you've ever seen one). And its no coincidence that the full spectrum sources are all the highest CRIs as well. They're all different qualities of flame though thus different correlated color temperature areas. Tungsten, I wouldn't categorize as a flame but more just like the metal being super heated (to the blackbody analogy), not really getting a chance to go to flame.
Really, when you get down to it, tungsten is the only thing that color temperature was designed to describe well. Every other white light source, from one end of the color temp scale to the other ends up with a correlated color temperature since it doesn't fit into the black body comparison but you can still find a color on the scale that will fit it.
I think when they say "black body radiator" what they really mean is a piece of some kind of "theoretical" metal being heated from red hot through blue hot. Only two of the three sources I can think of that are full spectrum (tungsten, daylight and carbon arc) are close to what I would call a "flame state" and that would be daylight (the sun is a big fireball right?) and carbon arc (pretty good flame going on in there if you've ever seen one). And its no coincidence that the full spectrum sources are all the highest CRIs as well. They're all different qualities of flame though thus different correlated color temperature areas. Tungsten, I wouldn't categorize as a flame but more just like the metal being super heated (to the blackbody analogy), not really getting a chance to go to flame.
Really, when you get down to it, tungsten is the only thing that color temperature was designed to describe well. Every other white light source, from one end of the color temp scale to the other ends up with a correlated color temperature since it doesn't fit into the black body comparison but you can still find a color on the scale that will fit it.
Way over my head, fellas! I'm going to have to research more on this topic before I can even try to contribute anything worth while...
So I went to home depot and figured I would do a little light test myself using 100+ watt output energy efficient florescent lights. I actually found Philips 5000K T12 40watt 92CRI 4 foot bulbs... bought 4 of those, put them in 2 $9 shop light fixtures and bam, I have 2 pretty decent (in my eyes) Kino Flo rip offs for about 40 bucks. Now I'm trying to bolster them together and fashion a type of grip arm receiver.
I bought a bunch of other bulbs to test them out as well... 100watt 5600K Flo bulbs as well as two giant 300W 2700K Flo bulbs. Once I have space to test them out I will post photos. Initially I'm pleased with the rip off kinos, pretty nice light, the throw is decent, obviously there is no control really, just on or off, but hey, what do you expect for $40!? :shocked:
So I went to home depot and figured I would do a little light test myself using 100+ watt output energy efficient florescent lights. I actually found Philips 5000K T12 40watt 92CRI 4 foot bulbs... bought 4 of those, put them in 2 $9 shop light fixtures and bam, I have 2 pretty decent (in my eyes) Kino Flo rip offs for about 40 bucks. Now I'm trying to bolster them together and fashion a type of grip arm receiver.
I bought a bunch of other bulbs to test them out as well... 100watt 5600K Flo bulbs as well as two giant 300W 2700K Flo bulbs. Once I have space to test them out I will post photos. Initially I'm pleased with the rip off kinos, pretty nice light, the throw is decent, obviously there is no control really, just on or off, but hey, what do you expect for $40!? :shocked: