A new player in DCI compliant projectors?

[Marcel Birgelen]

If you mean moire and speckle, as it was explained to me, it's the effect of the very precise wavelength of each laser diode combined with a perforated screen. The output of some of the diodes simply goes through the perforations, while that of others hits the white surface. Defocusing the lens slightly will make the artifacts go away - but at the cost of image resolution, of course. It can also be helped by adjusting the angle of the perforations of the screen. The last time we had an SP4K in the shop in preparation for onward shipping to a customer, someone from a screen manufacturer came by to do some experimentation on determining the optimum angle for the rows of perforations to minimize this problem.

As Mark points out, post houses don't like laser projectors, as a general rule, this being the main reason. Of course they also don't like how the color temperature of a xenon arc bulb's output changes over time (and over the last third of its warranty hours, a lot), meaning that they have to reshoot colors very frequently for applications where color space is critical. But as a general rule, they consider this to be the lesser of two evils.

What you see here definitely isn't speckle. Speckle, especially in its RGB incarnation, looks a bit like someone puked pearlescent beads all over your screen... What we see here also doesn't look like traditional moire to me, because to create this particular pattern, the perforations in the screen need to be in a perfect grid (you should preferably use a screen with random perforations for laser) and perfectly horizontally and vertically aligned, which almost never happens.It almost looks like a double-slit-pattern-like affair. You remember those double-slit pattern experiments back on high-school maybe, where you shot a laser through a double-slit and saw how it started to behave like a wave in water, rather than a stream of particles? If you never did that or if you were busy doing other things or simply don't remember, just Google it.

Laser interference patterns come in many shapes and forms. Just google it, some look like waves in a pond, others are far more complex. But they happen when a laser hits something that's smaller than its wavelength. The trouble is, we're sending very strong coherent light sources through a bunch of micro-mirrors, so this kind of interference patterns may simply be created by the spaces between the DLP mirrors for example.

People are using actual DLP and LCoS chips to create laser-interference patterns for stuff like holographics, so it is a thing and this thing may just be the cause of the funky shading Steve is seeing here.
Again, this is a theory, but if you have a better one, I'm all ears.

Phosphor machines do have less speckle right out of the box, because two of their primaries are created by exciting a phosphor, which is in turn split by a prism. This light isn't monochromatic anymore like the laser light itself. The remaining primary can still cause some speckle, but at least it won't look like rainbow glitter.

The peak response frequency and the filter shapes vary person to person. But, our brains compensate based on our experience with broad band illumination. It seems that we would all judge the same colors if we varied the amplitude of monochromatic sources that are at the peak response frequency. Person to person variations would be where the curve slope is the smallest (flat top). However, making lasers for these frequencies appears difficult. So, laser-based projectors often use frequencies that are on the skirts of the filters. A small variation in filter frequency or in filter shape can result in a large variation in the output of the XYZ filters resulting in variation in perceived color.

I think that your "simplistic view" is a pretty good approximation of how humans perceive colors and that's one reason why laser projectors are a problem. It's not just people who register colors differently, there even is a difference among camera sensors.

We need to see a spectrum of Xenon or HPM lamps AFTER the color filtering. No digital projector shows the full Xenon or HPM spectrum.

The interesting thing about a Xenon DLP projector, unless you're doing Dolby 3D, is that there aren't any dichroic mirrors or filters in the light path, other than the UV, IR filter which cut-off light that would be invisible anyway and some spectral coatings on the prism, to get a hard cut between the R, G and B component. I believe the core design idea of the DLP lightpath is to throw away as less of the initial spectrum as possible, to avoid compensating for it later on with filters, to be as light efficient as possible.

Barco does very good with their flagship DP4K-60L - it uses special speckle mitigation techniques, one of it is it's use of very many laser diodes and wavelengths. I think they also modulate the spectrum and so make it wider than it is normal for a single LD. They also use this on this unit to create the 6 wavelengths for spectral separation 3D. Comes at a price, though.

Barco's flagship 6P machines do a good job at speckle mitigation, even without weird stuff like screen shakers. But despite their 6P implementation, they still suffer from this "bad color syndrome" to some extend. For me this is especially visible in 3D mode using Dolby/Infitec color separation glasses, which I've nicknamed NTSC glasses... NTSC as in... Never The Same Color.

[Mark Gulbrandsen]

Not even seeing the laser projector on their web site right now....

The interesting thing about a Xenon DLP projector, unless you're doing Dolby 3D, is that there aren't any dichroic mirrors or filters in the light path, other than the UV, IR filter which cut-off light that would be invisible anyway and some spectral coatings on the prism, to get a hard cut between the R, G and B component. I believe the core design idea of the DLP lightpath is to throw away as less of the initial spectrum as possible, to avoid compensating for it later on with filters, to be as light efficient as possible.

Keep in mind that the optical coatings on various components also affects the spectrum going through...

[Marcel Birgelen]

Keep in mind that the optical coatings on various components also affects the spectrum going through...

Sure, but afaik, the only thing that's usually coated is the prism and those coatings are only there to filter out the stray light in the spectrum for each component. It will have some impact on the final spectrum of a "pure white" image, but it's supposed to be fairly small. The other coatings applied are there to block stray light entirely, that shouldn't have an impact on the spectrum itself, but only on the efficiency of the lightpath. Every optical component will have some impact on the spectrum, but, unless I'm remembering stuff wrong, the principal design idea is to filter as less as possible, in order to maintain the spectrum of the xenon light source. I also remember this to be part of the marketing strategy of DLP v.s. stuff like transmissive LCD: The general lack of filtering of the visible light component makes DLP one of the most efficient light engines.

[Carsten Kurz]

Marcel, tell me how the red light get's only to the red DMD on a xenon or HPM projector?

And what would the spectrum of a full white look like measured off the screen? Yes, a lot broader color components than for LED oder LD light sources, but not the full xenon spectrum reconstructed as you depicted it above. And yes, of course a broader color component spectrum helps with reducing color metamerism and speckle as well.

As a result, basically you need to get rid of most of the quality properties of laser light that makes laser light so special in order to make it suitable for high quality image projection applications. The good thing is, you can still call it 'Laser' for marketing applications.

[Mark Gulbrandsen]

Sure, but afaik, the only thing that's usually coated is the prism and those coatings are only there to filter out the stray light in the spectrum for each component. It will have some impact on the final spectrum of a "pure white" image, but it's supposed to be fairly small. The other coatings applied are there to block stray light entirely, that shouldn't have an impact on the spectrum itself, but only on the efficiency of the lightpath. Every optical component will have some impact on the spectrum, but, unless I'm remembering stuff wrong, the principal design idea is to filter as less as possible, in order to maintain the spectrum of the xenon light source. I also remember this to be part of the marketing strategy of DLP v.s. stuff like transmissive LCD: The general lack of filtering of the visible light component makes DLP one of the most efficient light engines.

Actually, everything in the NEC's that I have worked on has some sort of basic optical coating. It's mainly to increase the efficiency of any given optical part and eliminate stray reflections. Weather they are multi coatings or just a single coating I can't say, but I suspect multi coating.

[Marcel Birgelen]

Yes, a lot of stuff is coated to remove stray reflections, but this coating should block out light from the entire spectrum equally, it's not a color filter. In practice, some coating may block out e.g. red a bit more than blue, but the net effect of this should still be mostly marginal, or else the coating isn't doing its job.

Marcel, tell me how the red light get's only to the red DMD on a xenon or HPM projector?

Let's focus on xenon here, because that's where the discussion is at. Your average xenon 3DLP projector does the majority of the legwork via the prism, which isn't filtering but splitting the spectrum in three components. The sides of the prism, as I mentioned earlier, are indeed coated to get a more clean cut and to filter out any stray light that doesn't belong in that spectrum.

And what would the spectrum of a full white look like measured off the screen? Yes, a lot broader color components than for LED oder LD light sources, but not the full xenon spectrum reconstructed as you depicted it above. And yes, of course a broader color component spectrum helps with reducing color metamerism and speckle as well.

I don't have a graph here at hand, but there should be a relatively sharp cutoff < 420nm and > 650 nm, because that's beyond the human vision and filtered out by UV/IR filters. For the rest, the curve should be pretty flat, but you'll see two dips, one around 500nm and the other one around 560nm, that's the "cut-over" between B/G and G/R.

No, it doesn't exactly follow the output of the xenon lamp, but since the filtering is at a minimum, it should still get pretty close. It's generally known that colorists do prefer xenon projectors, because they're the best at accurately reproducing color.

As a result, basically you need to get rid of most of the quality properties of laser light that makes laser light so special in order to make it suitable for high quality image projection applications. The good thing is, you can still call it 'Laser' for marketing applications.

The biggest benefit of modern light sources is their efficiency, compared to letting some gas glow in a vacuum bulb. The fact that most of the light output is in the visible spectrum and not inside the spectrum we don't need and which essentially translates to heat is the biggest benefit.

I'm by no means an expert in this field, but I see two options for laser sources to fix this "spectrum gap". One is by simply combining much more sources that are all tuned to a slightly different frequency. The other may be using something like tuneable lasers, which are seemingly a thing, but obviously still very expensive. A "tunable laser" could "rotate" through a "spectral range" and as such simulate a broader spectrum. The disadvantage of such a solution is that we're essentially back to a color-rotation scheme, but maybe this can be executed at such frequencies that there are no visible artifacts.

[Mark Gulbrandsen]

Yes, a lot of stuff is coated to remove stray reflections, but this coating should block out light from the entire spectrum equally, it's not a color filter. In practice, some coating may block out e.g. red a bit more than blue, but the net effect of this should still be mostly marginal, or else the coating isn't doing its job.

I can speak from experience as a lens technician that getting optical coatings right, even in this time and age is extremely difficult. Even the manufacturers will tell you if you want matched coatings then it has to all be done in the same vacuum chamber. Strong used to sell lamp houses with matched reflectors that were coated at the same time. Don't know about Kinoton and others, but I assu,me they would offer the same.

[Marcel Birgelen]

I can speak from experience as a lens technician that getting optical coatings right, even in this time and age is extremely difficult. Even the manufacturers will tell you if you want matched coatings then it has to all be done in the same vacuum chamber. Strong used to sell lamp houses with matched reflectors that were coated at the same time. Don't know about Kinoton and others, but I assu,me they would offer the same.

Well, designing a lens system that compensates for optical aberration has always been something like an art only a few dark wizards really understand, even with modern software, it's a quest better left to those who know what they're doing. But, if I ever I embark on such a mission again, I know how to find you.

[Carsten Kurz]

Marcel, as you certainly know, a prism will refract white light into a continuous spectrum and as such is of little use to separate RGB from a white light source. What xenon and HPM projectors (and phosphor conversion lasers) use is a dichroic prim. It's surfaces carry dichroic coatings and yes, these DO filter the light. The prism part is used to reflect and spatially separate the color components so that they can go to the three separate imagers. The resulting white light on screen after recombination of these three color components thus is not a continuous spectrum as it was before. So the comparison of a full xenon spectrum with RGB LEDs or LDs doesn't help much here.
I hope the deteriorated business perspective for exhibitors and manufacturers in view of the Covid-19 crisis does not keep future light source improvements from being brought into the cinema projection world. The spectrum and line width of RGB LEDs seem to be much better suited for projection, without the need for speckle mitigation

Just that few companies nowadays seem to invest into high power LEDs for this application anymore.

(Small etendue is another reason why laser light sources are preferred currently - but essentially, it contributes to speckle as well, or goes with it.)

[Marcel Birgelen]

Marcel, as you certainly know, a prism will refract white light into a continuous spectrum and as such is of little use to separate RGB from a white light source. What xenon and HPM projectors (and phosphor conversion lasers) use is a dichroic prim. It's surfaces carry dichroic coatings and yes, these DO filter the light. The prism part is used to deflect and spatially separate the color components so that they can go to the three separate imagers. The resulting white light on screen after recombination of these three color components thus is not a continuous spectrum as it was before. So the comparison of a full xenon spectrum with RGB LEDs or LDs doesn't help much here.

Carsten, I put it in to illustrate that xenon is currently the most "spectrally consistent" (if that's a thing) light source we have, as the output of a xenon light bulb is pretty linear over the visible spectrum. DLPs design incentive was to filter the least light possible, compared to transmissive technologies like LCD. Yes, there is some filtering going on in the bandwidth where the colors cross over, around 500nm and 560nm. In practice, this isn't a complete blackout though, and just a dip in the plot. The resulting plot is the closest to a full spectrum we have, compared to any other light sources, including HPM, which have a high peak in both blue and green, phosphor laser, LED and certainly n3P RGB laser. The result is: colors that look good for everyone.

Besides inconsistent colors and speckle, the point is that laser projection seemingly has other issues too: One of them seems to be color convergence when patrons are wearing "optical systems" as in glasses themselves. And how do you explain those ugly "banding" artifacts you see on the pictures Steve has posted for example?

I hope the deteriorated business perspective for exhibitors and manufacturers in view of the Covid-19 crisis does not keep future light source improvements from being brought into the cinema projection world. The spectrum and line width of RGB LEDs seem to be much better suited for projection, without the need for speckle mitigation

Just that few companies nowadays seem to invest into high power LEDs for this application anymore.

(Small etendue is another reason why laser light sources are preferred currently - but essentially, it contributes to speckle as well, or goes with it.)

At least the LED market itself is moving forward and those developments can also be adopted for projection. I'd expected we'd see high-powered LED based projectors first, before we'd see laser. LED-powered projectors are nothing new, but they never really made it beyond the meeting room.

LED systems will never have the same small etendue like laser systems, but with the right optics, they can still be pretty efficient and since most lasers used for projection are essentially driven by LEDs, their durability, if properly cooled, should be comparable. The problem with LEDs though is that they're also almost monochromatic and require all kinds of coatings to enhance their spectrum.

Heck, maybe we should focus more on firing lasers at phosphors... we used to create fancy images by shooting electron beams at phosphors for decades, so maybe the future is shooting high-energy photon beams at it.

[Mark Gulbrandsen]

Well, designing a lens system that compensates for optical aberration has always been something like an art only a few dark wizards really understand, even with modern software, it's a quest better left to those who know what they're doing. But, if I ever I embark on such a mission again, I know how to find you.

DLP lenses are all computer designed today. No dark wizardry involved.

[Steve Kraus]

The spectra of the light should not matter so long as you can stimulate our retinas appropriately in the R, G, and B areas of sensitivity. I agree with Harold's explanation.

Just think how much more complex our color imaging systems would need to be if we could not cheat and use just three colors to reproduce to what our eyes see as full color. Think of a visiting alien with a broader range of color perception asking why we image bananas as red + green instead of in true yellow. (Okay, a banana probably does give off R & G as well as true yellow light.)

[Sean McKinnon]

The interesting thing about a Xenon DLP projector, unless you're doing Dolby 3D, is that there aren't any dichroic mirrors or filters in the light path, other than the UV, IR filter which cut-off light that would be invisible anyway and some spectral coatings on the prism, to get a hard cut between the R, G and B component. I believe the core design idea of the DLP lightpath is to throw away as less of the initial spectrum as possible, to avoid compensating for it later on with filters, to be as light efficient as possible.

Not to quibble but some Xenon machines do use notch filters in the light path.

[Marcel Birgelen]

Not to quibble but some Xenon machines do use notch filters in the light path.

Like indicated in the whole discussion. Yes: All xenon DLP projectors use some filtering on their prism, to filter the two cross-over bands.

Just think how much more complex our color imaging systems would need to be if we could not cheat and use just three colors to reproduce to what our eyes see as full color.

I could imagine a laser-scanning solution with a tunable laser. If you could tune the laser to any frequency within the visible band and if you could modulate this sufficiently fast, then you could possibly create a "true color" laser-scanning projector.

Tunable lasers are still experimental, but may solve both the speckle and color problems of modern laser light sources.

[Mark Gulbrandsen]

It's the new GDC Projector. It mentions their name right in the Youtube blurb.