Ludvig, we haven't posted many of the project details as of yet but hope to relatively soon. We are using that LCD shield presently. It's buttons have been mapped to menu items at this stage. It's more the readout of the RGB values I'm hoping to somehow squirt into the circuit.
In the interest of progress for now, for the sake of repeatability we can simply create a decal and eyeball the knob positions.
I dont know how that three knob controller works but I guess that its PWM and has some transistors inside. If you wanted to you could easily drive the arduino with those pots and have arduino pulse the transistors. Then just write the values out on the lcd shield. Or perhaps just hook up the pots in parallel to three analog ins on the arduino and read the values and display them on lcd with the arduino. Last but not least you could expose some wires soldered onto the pots to the outside and read the resistance with a multimeter.
Ludvig
You need more than three LED colors "RGB" to get a CRI of over 85.
Alternatively, you can mix warm white (~3000K) with cool white (6000K) and add green and red in for balancing the spectrum out. All colors individually controlled. Split "Green" into "Cyan" and "Lime Green" for even better control. Don't forget that your "White LEDs" are blue and yellow heavy ... ;-) just the nature of the beast. CRI over 95 with good R9 is possible this way.
Since this is a fairly large area "light box", color mixing shouldn't be too much of an issue. Matter of diffuse reflection and diffusors and distance of course.
CCT and CRI will shift with temperature but camera RAW post processing in case of stitching can take care of this. Cooling the LED's is important as many times noted in this thread.
Some insightful ideas Ludvig, thanks! Once things are up and running and demonstrable and repeatable, oh, and viable : ) I look forward to devising the optimal method for dialing in those values.
Amedeus, I see what your saying. However, we are a bit invested (with no regard for anything but cost/convenience/and a bit of RGB separation) and I wonder if you see viability in the Ribbonflex RGB LED system for this application or not? The white temps you refer to, are they that different from being able to dial in a combination of R, G and B values? Each color is independently controlled, off to full power. I haven't taken an image of the spectrum and honestly was going to use actual RAW files to make those judgements. Peter has already done this with RGB LED and I was impressed with the tonal separation, and with only the usual amount of work, the color as well.
But how important is CRI? I know that drumscanners use Tungsten but the Nikon Coolscan series wich is more or less the goldstandard for 35 an mediumformat scanning ises rgb led.
Are digital cameras better at seeing peak R, G and B values or everywhere in between values? Are there specific wavelengths we should be targeting?
In my tests with negative color film I have had much better results with rgb led then with sunlight. I think that when you stimulate the r g and b photosites without any splill between them you get a much bigger gamut. the negs look supersaturated with rgb and with sunlight they are more of a mush so its easier to get balanced positives from the negs with rgb leds, they have to be desaturated after balanced though. Perhaps for positives its a different result, i have not tested that.
The pro flat beds, such as a Cezanne, use 1990s florescent bulbs. I doubt their CRI is very high.
“You often feel tired, not because you've done too much, but because you've done too little of what sparks a light in you.”
― Alexander Den Heijer, Nothing You Don't Already Know
This response curve is for ektar. Not exactly a good "cri" or blackbody response. In my experiments I have chosen leds with a similar wavelength to the peaks in colornegatives and my sensor.
Coincidently sensors are quite similar in their response curve.
I fully understand the desire to keep it simple RGB. Makes life easier in light of the use and the need to be able to generate pure R, G and B separately, this is, within the limitations of the choosen LEDs. Saturation is great though and as Ludvig mentioned, desaturation is often needed to restore a sense of natural.Originally Posted by Daniel Moore
I'm not familiar with the Ribbonflex RGB LED system but will look into it.
The white temps I'm talking about are being dialed in changing power to up to 7 LED channels. Not that different than RGB but with a more accurate "broadband" spectrum. The spectrum of LEDs is "spiky/peaky" with a limited bandwidth. The Bayer filter combination in an imager (or RGB filter solution with 3 imagers) have a broader spectrum with more accurate coverage than what the LED is able to produce. RGB LED's combined also have a different response curve than what the eye perceives.
You can't record what's not there ... you can fool the eye briefly in sequential viewing, not in side-by-side. The results may be different for negatives than for positives. I'm mostly dealing with "positives" in my work.
As for Color temperature CT is only valid for black body curves ... we talk about CCT for everything else. Two different animals. 3000K CT from a true black-body radiator is unlike the 3000K CCT from a LED RGB source or CFL.
Hence the need for CRI and when that was no longer sufficient ... planckian locus, McAdams ellipses for more accurate definitions.
Posting Permissions
Reply With Quote
Bookmarks