AVS Forum banner

1 - 15 of 15 Posts

·
Registered
Joined
·
4,077 Posts
Discussion Starter #1
The reason I ask is because a 6x color wheel would seem to have that much time spent on each color segment of the wheel. If two DLP projectors were stacked and the second running 4ms behind the first, the color wheel would be on a different color than the first. Wouldn't that eliminate rainbow effect for those that see them ? In addition to adding brightness, of course, by stacking.

This would even allow stacking of non-RGBRGB projectors which are brighter to begin with. A 6000 lumen rainbow-free DLP image for under $1500 would be possible, no ?
 

·
Registered
Joined
·
4,077 Posts
Discussion Starter #3 (Edited)
Does it matter ? When each frame starts, the driver in the projector is going to check the position sensor on the wheel to ensure the frame is rendered correctly. Whether the delay is 4ms for close to one color space on the wheel or slightly more or slightly less, the two projectors should still be rendered two different colors at the same time. 4ms is only 1/10th of a frame on 24p material, so it isn't like two different frames would be displayed from the two projectors. They might create some motion blur to instruments but the eye would never see it.

Actually, only a portion of the segment on the wheel is used. "Brilliant Color" works by trying to use more of the segment and ends up sometimes using part of the wrong segment just because of the sync issues you mention. But otherwise, the mirrors all turn to off before the edge of the segment gets close. If both projectors have Brilliant Color turned on full, you might have as many as four colors being rendered at the same time. If both had BC turned off, any sync issues would probably fall in the "off" period for one or the other projector.
 

·
Banned
Joined
·
11 Posts
the two projectors should still be rendered two different colors at the same time.
I guess, that two different colors will rendered (most of the time) without any HDMI delay, simply by using two projectors, because the internal timing is not synchronized to any (external) high precision time source.
 

·
Registered
Joined
·
946 Posts
So it would be an automatic 12x color wheel when two PJs are stacked?
 

·
Banned
Joined
·
3,901 Posts
I'm not sure if there would be even a 4ms delay between two identical projectors (or electronics therein) paired over HDMI, otherwise passive 3D wouldn't work and certainly stacked 2D projectors wouldn't work either. Splitters introduce barely no lag, AFAIK (I could be wrong though, has anyone measured it with Leo Bodnar? I'm sure Google would have some answers to that).

It's an interesting question but I don't think it's pertinent to dual modulation.

DMDs being small (say, 0.47 inch for the latest 1080p models) makes it extremely hard to align them in series. You have to use a lens to square up the projection, and relay lenses inherently will blur the image a bit, so the back DMD will need some processing. Ok, you can do that in a shader by driving each projector independently from a PC. The latency between the first and second DMD is the least of one's worries.

But, you can delay the colour wheel in the service menu by changing its phase, that should do it. Put a different phase value in each projector until the colours look right. Done.

Plus, if Brilliant Color is off, all the mirrors are off during transitions between the colour segments, reducing rainbows and lumens by a bit (5%?). I learned how it works just recently. I like Brilliant Color due to the more lumens it provides but it's less accurate, certainly.
 

·
Banned
Joined
·
3,901 Posts
So it would be an automatic 12x color wheel when two PJs are stacked?
Only for stacked 2D projectors in parallel (and out of phase), not DMDs in series. They'd have to have their colour wheels synced, so still 6X no matter what. Or 6X with one RGB and one in grayscale.

Like I told dreamer in a PM, I would run the back DMD in grayscale mode (this is trivial to do on any DLP projector, either through NVidia control panel to make it black and white (digital vibrance control set to 0), or by just not connecting the CbCr channels). Then unplugging the colour wheel will triple your lumens (from the back DLP only). Most dual-modulation designs do one in grayscale and one in RGB, in either order (logically it shouldn't matter, but there are practical concerns like aligning sub-pixels which is 3x harder than aligning pixels, e.g. for stacked LCDs).

If you want to modify the phase (differential latency) between two identical or similar DLPs, I would run them analog (component or RGB) and do it via an in-line phase shift. Analog VGA and Component have the added benefit of being true 10-bit in 4:4:4 even at 1080p60. I've run a VGA cable to my w1070 to test it and it works. The ADCs in the analog components of all these DLPs is 10-bit, meaning if you manage to feed UHD Bluray content to them which is native 10-bit, you should get some benefit in terms of banding reduction, all without having to drop to YCbCr 422 (which is the max for HDMI 1.4 at 1080p60 for 10-bit).

Of course dual modulators in series will multiply not only the contrast ratio, but also the effective bit depth of the whole system. So you'd get 16-bit per channel colour (48-bit Deep Color, they call it), through encoding, say one projector's incoming video signal to a different gamma value. Of could you have to do this anyway, because dual modulation would attenuate the signal twice, meaning the luminance, if from 0-1 normally, is now squared (so smaller). So you have to feed the square root of the video signal. I've done it and it works with dual LCDs.

I had a plan to take out the colour wheel from my w1070 and operate it in grayscale 10-bit, focus the image onto my 1440p 5.5 inch topfoison LCD panel, then re-project that out. Basically, a relay projector which would work in theory with any projector out there as the backend one. A 3LCD projector would have half the lumens of a DLP operating in grayscale due to the use of a polarizer which cuts randomly polarized light from the lamp in half (integral of a cosine from 0-1 limits is 0.5).

So, a w1070 would have 6000 lumens where it normally has 2000 in RGB. Then you of course need to take a hit for the polarizer for the LCD so you're back to 3000 lumens, so comparable to 3LCD projectors, but you get 1M:1 native on/off, 16-bit colour in the bargain. I might try finishing up my DIY project but I wanted to buy a new projector first. I've nearly broken my w1070 beyond repair several times now and I have it dialed in perfectly right now with zero dust specs inside the light cavity, so the image looks like it was on day 1. (after several rebuilds).
 

·
Registered
Joined
·
4,077 Posts
Discussion Starter #8 (Edited)
Only for stacked 2D projectors in parallel (and out of phase), not DMDs in series. They'd have to have their colour wheels synced, so still 6X no matter what. Or 6X with one RGB and one in grayscale.

Like I told dreamer in a PM, I would run the back DMD in grayscale mode (this is trivial to do on any DLP projector, either through NVidia control panel to make it black and white (digital vibrance control set to 0), or by just not connecting the CbCr channels). Then unplugging the colour wheel will triple your lumens (from the back DLP only).
Yes, this topic was for a much more straightforward stacking of two projectors for the usual reasons of increasing brightness and not the light-modulator experiments we PM'd about.

It just occurred to me that two DLP projectors could also eliminate rainbows if they could run 4ms apart. I was a little surprised to find switches, splitters and cable lengths have virtually zero lag. That would have been a simple solution while keeping all connections vanilla HDMI from any source -- not HTPC VGA. I did find some indications that processing inside older DLPs added lag. Even simple sharpness controls added lag. But no hard numbers. If setting the sharpness high on one projector and zero on the second would produce 4ms lag, that would also be convenient. Generally, the 'gaming' modes cut lag on some projectors by much more than 4ms, right ?

As far as running simple stacked projectors with one in grayscale mode goes, wouldn't that just wash out colors ?

I love the idea of running your W1070 projector as the grayscale light source for a color LCD panel from a monitor or tablet. You would still need an entire second lens assembly after the LCD panel, wouldn't you ? Removing the color wheel seems infeasible, though. I thought the position sensor on the wheel was necessary to the DMD driver so it knows which mirrors to flip ? I had always thought a white field would have all mirrors "on" continuously across all color wheel segments -- like the wheel had been replaced by an ND filter. Your 6000 lumen comment for the W1070 sans wheel means I am wrong. Where are the 4000 lumens disappearing to in a color wheel projecting a white field ? Because the white light is being filtered down to just three frequencies ? Would you replace the wheel's sensor with some timing chip if you removed the wheel, or maybe replace the wheel with a clear wheel ? Maybe replace the wheel with one shattered but still spinning the sensor ?

I can see doing it with a cheap but bright used LCD projector as the light source -- just replace the dichroic mirror with a FS mirror and remove two of the three LCD chips. You could do that for cheap without waiting until you decide on a projector to replace the W1070. Unless the output of the LCD projector would have the same polarized light problems you schooled me on in our PM.
 

·
Registered
Joined
·
4,077 Posts
Discussion Starter #9
Plus, if Brilliant Color is off, all the mirrors are off during transitions between the colour segments, reducing rainbows and lumens by a bit (5%?). I learned how it works just recently. I like Brilliant Color due to the more lumens it provides but it's less accurate, certainly.
I think the difference in lumens between Brilliant Color on or off is usually much more than 5% -- more like 20% -- but it depends on how projector manufacturers implement it. Benq's on/off is pretty conservative in not risking running into the next segment, while Optoma's 1-10 setting allows a lot of slop or none.

I thought fiddling with color wheel phase caused banding and skewed colors. Delaying the signal before then would seem more likely to work.
 

·
Registered
Joined
·
4,077 Posts
Discussion Starter #10
I guess, that two different colors will rendered (most of the time) without any HDMI delay, simply by using two projectors, because the internal timing is not synchronized to any (external) high precision time source.
Actually, I think the HDMI signal qualifies as a highly precise timing source. They will be exactly in sync unless some delay is intentionally added somewhere.
 

·
Banned
Joined
·
3,901 Posts
I'm sure there are (expensive) gen-lock devices out there that can sync or de-sync HDMI digital signals.

Best to use the same electronics controlling both modulators to guarantee them being within a narrow latency range from one another.

Like I said in PM, dual-LCD is the best for now and the future, because there are no relay lenses necessary, you can use the exact same LCDs for each, grayscale panels inside Epson projectors are gamut independent so DIY'ing an RGB laser P3 or rec 2020 projector is a piece of cake (you can even stick the lasers somewhere else, and put some tubing to make sure nobody puts their eyes in the way or gets hit by a stray reflection). Of course with lasers you also avoid the 50% lumens drop due to the polarizing grid on the LCD panels, if you align them to match. Actually, you pretty much have to otherwise the lasers would likely melt the LCDs due to overheating. A polarizing grid on the back of the LCD is only really necessary for lamp-based light sources, so it could even be removed. But best to leave them in there.

ps does anyone have a pair of 3LCD projectors they're not using? :)
 

·
Registered
Joined
·
4,077 Posts
Discussion Starter #12
I'm sure there are (expensive) gen-lock devices out there that can sync or de-sync HDMI digital signals.

Best to use the same electronics controlling both modulators to guarantee them being within a narrow latency range from one another.
An inexpensive HDMI repeater-type extender (as opposed to signal booster) will add about 30ms according to a gamer forum I was reading. Any converting back and forth from HDMI to Cat5 must add delay. And any internal processing in the projector, including sharpening, keystone correction, FI, etc. will add delay. I just can't find hard numbers. I'd have to spend $100 on a Bodnar sensor to get measurements.

As far as lasers go, it sounds like fun, but getting lasers powerful enough seems tough. Looking around a bit, depending on the color of the laser you only get 200 lumens or so per watt. If you can believe the specs, this place has 10W lasers in many colors. http://www.dhgate.com/wholesale/10w+laser.html

Dialing the beam out to 0.74" diameter would require what sort of lens ? And expanding lens and then a collimating lens ? And what RGB wavelengths make for 2020 color rather than rec709 ?

Was it you that shattered his color wheel while repairing a projector or someone else ?
 

·
Banned
Joined
·
3,901 Posts
DCI P3

Red: 615nm
Green: 540nm
Blue : same as rec 709

Rec 2020:

Red : 630 nm
Green : 532 nm
Blue : 467.1 nm

http://www.set.org.br/artigos/ed161/161_revistadaset_pag62.pdf

The idea is not to buy single powerful lasers, but make diode banks of multiple cheap ones. Adds redundancy too.

Diode banks' lasers can be focused into a fiber or straight into a collimating rod. Or course it's hard to make multiple lasers coherent (understatement) but that's well beyond DIY territory.

Using fiber coupling costs about 20% of lumens compared to just directly beam combining onto the imager (I saw that tidbit in a commercial laser projector post on dci forum).
 

·
Registered
Joined
·
4,077 Posts
Discussion Starter #15
I run my Sharp 30K 3D DLP's in a stack for extra brightness. I didn't notice any changes with perceived RBE, these projectors are generally good with this to begin with.


source -> Denon 7200W dual HDMI output to each projector.

I don't see RBE on my Benq W1070, either, but I have not had anyone else over to watch it. I just think it would be an interesting side-benefit of stacking if it eliminated rainbows for those people who do see them. I happen to have a second W1070 that the power failed on. If stacking gave me both extra brightness and eliminated any chance of RBE, it would give me incentive to repair it. This video of the brightness increase of stacked projectors intrigues me, especially the way city lights in the distance appear in the night shot. Do/did you notice a dramatic increase with your Sharp stack ?

 
1 - 15 of 15 Posts
Top