Chroma Confusion - 445 vs 532 - WTF Mate?

I've kicked around in Chroma off and on, but I've never scrutinized its output, or had to rely on it's figures. In fact, I've generally booted it up only when someone has cited its figures relative to those produced by my own brightness comparison tool.

But this evening, I actually had occasion to need it's output figures for the balancing of a 445, 532, and 640. I know that Chroma is relied on by many, but speaking from my very limited colour mixing experience alone, Chroma is WAY OFF. See screenshot:



In what universe does 150 mW of 445 balance out 180 mW of 532? It doesn't matter if you're talking about the beam (considering Raleigh) or the dot (not), 180 mW of 532 is substantially brighter than 150 mW of 445.

Second, look where Chroma is sticking 445 on the spectrum. I'm willing to accept that we all see colour differently, that there is no standard monitor calibration, and that not every diode we call a 445 is actually lasing at 445, etc etc etc. But this is too far from the reality of a 445's colour to attribute to perception, display, or diode variance issues. That's plainly too far into the violet.

There's a difference between brightness and color balance. 50mW of green will be substantially brighter than 200mW of 660, but together they make a perfect yellow.

That is one drawback to using the color 445 in a projector. The blue isn't very bright.

About the color that it attributes to 445, that's just a stock photo of the CIE colorspace. The math is sound.

so wait in my projector 200mw of 532 should only have around 180mw of 445nm?
That seems off to me.
post link where to dl chroma please?

rhd said:

Second, look where Chroma is sticking 445 on the spectrum. I'm willing to accept that we all see colour differently, that there is no standard monitor calibration, and that not every diode we call a 445 is actually lasing at 445, etc etc etc. But this is too far from the reality of a 445's colour to attribute to perception, display, or diode variance issues. That's plainly too far into the violet.

Click to expand...

Looking at a calibrated screen here, i'd agree that the color in the diagram is 'too purple' for the output of a 445 nm laser diode. I think the fault is in the rendering of the diagram though - even a few nanometers make the difference between blue and purple at these wavelengths, and the diagram probably is not accurate enough to show a fair difference between 450 and 440 nm.

Rule #1 of lasers: always use chroma.
Rule #2 of lasers: never trust chroma.

However with that said, a 1:1:1 ratio of 642nm to 532nm to 445nm does produce a nice white (talking from experience, not from chroma). It's surprising how much you can change the colors and still get a nice white.

I appreciate the everyone who has pitched in has more experience with colour mixing than me, so I accept your position as accurate.

What's driving me nuts now, is the "why?"

200mW of 445 vs 200mW of 523, when viewed on their own, have clearly different brightness levels. Cyparagon touched on the fact that brightness does not equal colour mixing. I'm having trouble conceptualizing how this is the case though. If the 532 perceived as substantially bright than the 445 (which is is when we're talking about the same mW of both), then how could they mix to produce a colour falling at the midpoint between them - why wouldn't it skew towards the green side of things?

Yes, you can have a relatively large deviation from chroma and still have a good white. Notice that the sun looks white, but so does an incandescent lamp. If you have more colors on the blue side, your white appears "cooler" If you have more red and green, your color appears "warmer". Cool white and warm white are both still white.

rhd said:

how could they mix to produce a colour falling at the midpoint between them - why wouldn't it skew towards the green side of things?

Click to expand...

Yeah, you'd think so, but that isn't the case. One reason is the "blue" cones are activated better by 445 light than say 473. Run the figures again for 473, and you'll find you need more, even though it is brighter.

Cyparagon said:

Yeah, you'd think so, but that isn't the case. One reason is the "blue" cones are activated better by 445 light than say 473. Run the figures again for 473, and you'll find you need more, even though it is brighter.

Click to expand...

That's remarkable. You're right, more 473 is required than 445, to balance identical values for RED and GREEN.

Looking at the specific RG and B cone response curves, it makes sense that 445 would appear brighter than 473. At 473 you've got the following cones picking up some of the the wavelength:
- roughly 0.5 of the BLUE curve
- roughly 0.25 of the GREEN curve
- roughly 0.15 of the RED curve
- in aggregate, 0.9 units of sensitivity

At 445, you've got essentially
- all 1.0 of the BLUE curve,
- roughly 0.15 of RED/GREEN combined.
- in aggregate, 1.15 units of sensitivity

Now here's my question -
We're all familiar with this chart:


Why doesn't this chart reflect the reality that 445 is going to be perceived as brighter than 473? If it did, we would see an additional upwards pointing peak around 445.

In fact, I'm fairly impressed with this graphic, as it would seem to better reflect my expectations based on the discussion in this thread:

Just throwing this out there, maybe the green cones send a stronger signal, and thus produce a higher PERCEIVED brightness per unit?

rhd said:

Why doesn't this chart reflect the reality that 445 is going to be perceived as brighter than 473?

Click to expand...

445 excites the "blue" cones more than 473, but 473 excites the green and red cones more. White is based on the combination of the three cones' signals. Here's a better approximation, more to scale:

sorry that i cant find the thread, but a while ago i read that you can even get a pretty decent white with just 2 colors also.


michael.

Cyparagon said:

445 excites the "blue" cones more than 473, but 473 excites the green and red cones more. White is based on the combination of the three cones' signals. Here's a better approximation, more to scale:

http://www.unmc.edu/physiology/Mann/pix_7/cone_sen.gif

Click to expand...

So what wavelength of light should be perceived as brighter, ignoring the context of colour mixing, when comparing an identical power rating - 445 or 473?

I'm still struggling with why 445 could have more influence mw for mw than 473 in the context of colour mixing, yet not exceed 473 on the overall foveal sensitivity line.

For white, each of types of cones need to be stimulated. For brightness, the stimulation of each cone type only needs to added. 445 stimulates blue, and very little green and red. 473 stimulates less blue, and more green and red. This means it appears brighter, yet more is needed to adequately stimulate the blue cones enough to be perceived as white.

chipdouglas said:

you can even get a pretty decent white with just 2 colors also.

Click to expand...

I did it with 488 and 589

LoL. ^^^ that.