Talk:DLP

From Wikipedia, the free encyclopedia

Contents

[edit] DLP vs DMD resolution

I tried to fix some errors in the DMD section. For example the article said that each mirror represented a single pixel. That is generally not true in modern DLPs due to Wobulation. I think what I wrote better reflects the current state of DLP TVs, but it still implies that DMDs come in 1920x1080. My understanding is that the vast majority of DLPs sold as 1080p sets, do not have a 1920x1080 DMD in them, but one half that size. I could not find the website where I read this and was working from memory, so I did not want to get into too much detail about it. Maybe someone else could make it a bit clearer which sizes modern DMDs come in, and how they correspond to the HD resolutions.128.117.194.163 21:30, 24 July 2007 (UTC)

[edit] Color primaries and mixing

Actually, red, blue and green are not primary colors. They are base colors for subtractive color mixing, but by definition, the primary colors are red, blue and yellow. -Anon

The anon has it the wrong wko;hkl;jay round. Theresa Knott (Not the skater) 16:23, 24 Oct 2004 (UTC) Double Anon April 27th, 2005

Actually, neither is right. Red, green, and blue, (RGB) are the base colors for additive color mixing. Additive color mixing is commonly used when the light source is integral to the system (think television, computer monitor, LCD screen etc.).

Cyan, magenta, and yellow are the base colors for subtractive color mixing commonly used when an external light source is utilized (think paintings, printed documents etc.). It helps to think of Cyan, magenta, and yellow as being anti-red, anti-green, and anti-blue.

Yellow is often included in the list of primary colors due to a fluke in synthetic pigments. As described here: http://science.howstuffworks.com/light7.htm

I personally dislike the term 'primary colors' when it is used in reference to red, green, and blue in discussions of color science. It's use implies that color mixing is property inherit to the nature of light when in reality its root lies in the nature of human physiology and the nature of the human eye.

-NPW

NPW is right. Red Green and Blue are that way because we have cones in our eyes sensitive to red, green and blue light. Some people (all of them are women due to the nature of the genetic anomaly) have a fourth set of cones sensitive to another primary color. Fortunately (or unfortunately) that fourth color is aparently very close to red. Otherwise TV would probably look very artificial to them. Nsayer 19:02, 25 July 2006 (UTC)

[edit] The "Rainbow Effect" and three-DMD systems

The article makes the claim:

Three-chip projectors do not suffer from the "rainbow effect", since all three components are present at the same time.

While it's true that three-DMD devcies are far less prone to the "rainbow effect", it's not true that they are completely immune. Because the DMDs are (fundamentally) binary devices, they are operated as Pulse Width Modulators and the red, green, and blue beams are each individually turned on and off. This means that for certain colors, there are definite times when a (say pink) displayed object is illuminated solely by the red beam and at other times by all three beams. If your eyes are in motion, such an object could still be perceived as being striped in red and white, even in a three-DMD system. The reason the effect isn't seen as much in a 3-DMD system is that the pulse-width modulators operate a lot faster than the frame rate of the color wheel in a single-DMD system so there's less distance separating the colors when your eyes are in motion. Also, the produced color artifacts are (usually) a lot less distinguishable than the obvious red, green, and blue artifacts.

This isn't so important that I'm going to edit the article, but we should be aware of this.

Atlant 17:34, 28 Mar 2005 (UTC)

I've actually been to a DLP presentation by the head of Texas Instruments UK (who's personally quite heavily involved in DLP products) and they actually have a clever way of doing the modulation that means (in your example) the blue and green would modulate very very quickly while the red stays on, meaning the interval is so small that no one could possibly see it - unlike the color wheel problem, which is (or used to be) only just outside the bounds of normal human vision, hence some people can. --Dtcdthingy 20:29, 28 Mar 2005 (UTC)

You're correct that the DMD mirror isn't simply switched on and off once per frame for a single variable-width pulse, but the switching rate of a DMD mirror isn't all that high; it's only about 5 kHz tops ([1]) so a given mirror won't switch more than (say) 82 times per field for about 41 "pulses" of light. (The switching time isn't zero and there is a wear-out mechanism in the torsion beams that suspend the mirrors.) Your eye can sweep across an entire screen width in one field, so do the math and then tell me that you wouldn't be able to see the banding for certain well-chosen colors.

As I said, the effect is far less noticeable than with a single DMD device, but it's still there.

Atlant 01:05, 29 Mar 2005 (UTC)

Let's go with your numbers. The eye moves across the frame in 1/60th of a second, during which time 41 pulses of light are sent. The other number we need is the "shutter speed" of the eye - let's assume it's also 1/60th of a second. What you'd get is 41 images printed on top of each other on the retina during the exposure, equivalent to taking a photo, moving the film 1/41st of a frame, and repeating 40 times. If we imagine that film, the motion blur from the movement of the film would appear in 41 separate steps rather than being completely smooth. I really don't think the brain would be able to tell the difference, to be honest. --Dtcdthingy 02:01, 29 Mar 2005 (UTC)

(For ease of discussion, let's assume there's a single line of 50%-saturated pink displayed from top to bottom on an otherwise-black field.)

You don't need the "shutter speed" of the eye at all. If your eye sweeps left or right across the field in 1/60 of a second, then what you will see (thanks to persistence of vision and the like) is a screen that has 82 stripes. Half of the stripes will be white and half of the stripes will be red.

If you don't think you'll be able to see this (a pattern of 82 white-and-red stripes occupying the full field), then perhaps you need to get your eyeglass presrciption checked. :-)

The point is that your eye, in moving, "breaks up" what would otherwise be perceived (through persistence of vision and the like) as a static object. Again, the effect is nowhere as obvious as with a single-DMD system, but it is still there. Why don't you call your TI guy and ask him to contribute to our discussion?

Atlant 13:35, 29 Mar 2005 (UTC)

Human eyes perception is around 85Hz, few might see around 100Hz, given that those 4X models of the single-ship aren't creating rainbow effects anymore, I wonder how one can perceive that in three-chip DLP? It doesn't even have color wheels. Remember that our brains interprets the signals from the eyes, and by the rule of Persistence_of_vision I found your argument "technically sound" but not "realistically" possible, because if we can actually see the Rainbow Effect in 3-chip DLP at 5kHz, then there is just no way we can settle with 24 frames per second as we will be seeing 24 frames of photos instead of a movie. Kenimaru 09:29, 13 October 2006 (UTC)
1. The rainbow effect occurs when an object is moving relative to your retina. This can be because the object is moving on the screen or because your eye is moving relative to the screen.
2. Because the light source from a DLP projector is binary (fully on or fully off) pulse width modulated, it is not continuous. Furthermore, for certain colors (I used pink in my example above), the duty cycle of the R, G, and B components is not identical. Therefore, in a three DLP projector projecting "pink" light, there are times when the "pink" light source is actually saturated red and other times when it is unsaturated white.
3. Therefore, if the object is moving relative to your eye (so that persistence of vision transforms a pattern in time into a pattern in space), your eye will therefore see areas of two different colors: some red and some white.
Feel free to argue that my example is pathological, but there is no doubt that if you try my example you will produce some degree of the rainbow effect, even on a three DLP projector; it's a natural fallout of the technology. Three DLPs helps compared to one DLP and a color wheel. Better (smarter) PWM algorithms helps also, but there's an upper bound on how fast you can cycle the mirrors and it's only in the low KHz range. Given a carefully-designed experiment, you'll still see the rainbow effect.
There's another good way to demonstrate the effect I'm describing. Take a set of LED Christmas lights. These flicker at the mains frequency (so 50Hz in most of the world and 60 Hz in North America) and spend about 50% of their time "off". Sweep your eyes across the lights and see if you don't get a pattern of dots.
Alternatively, go to the Boston's Museum of Science and go look at the line of red LEDs in the Thompson Electrical Theatre. That is, they look like a plain line of LEDs until you sweep your eyes across them rapidly. THEN you find that you're looking at dinosaurs, fish, and other objects. (I think the Exploratorium has an even better version of this in red and green.) This is because the motion of your eyes provides the X-axis scan that allows full fields of light to be painted across your retina. The DLP experiment I'm proposing exploits exactly this same effect.
Atlant 13:41, 13 October 2006 (UTC)

[edit] Marketplace

I really like this segment, could people who know add it to other television types. Especially the pros/cons. I'll start researching it, but it may be a while before I get it done.

[edit] Pros/Cons - DLP Lag?

An item on the Cons list says, "Lag time between input and display with video consoles such as the PlayStation 2, making video game play nearly impossible."

Do we have a citation for this? - I can't see how this problem would be a consequence of using a DLP element in a display. My guess would be that since most DLP displays are very high resolution (and most consoles very low) the lag would be coming from upscaling the video. Either way I suppose the point could use some clarification?

— Mobius 08:32, 8 July 2006 (UTC)

That entire thing was worded oddly. I hope I did a decent job rewriting it, but I ended up removing the statement to the effect of "TI's new stuff is better" because I can find no supporting statements anywhere. This article was the first return on one search on the matter. If anyone can find a citation, feel free to re-add it.
Also, I put in a citation to a faq posted on a forum. I don't have any clue as to its permanence, this might be a bad citation.
- GTMoogle 03:21, 16 August 2006 (UTC)

- I think it needs to be added that this was only an issue with the 1st generation DLP chips, and hasn't been a problem in DLP sets for 2 years now.--Indiearmy 03:19, 6 November 2006 (UTC)

DLPs buffer the entire frame and then display it all at once, as opposed to CRTs (and maybe LCDs?) which start displaying the frame as soon as it is recieved. Therefore there is at least an additional half frame of lag on average.4hodmt 17:37, 29 November 2006 (UTC)
I'm not sure that's a technically-accurate statement. DLPs are organized just like Dynamic RAM chips and are accessed in much the same way, so the "write" doesn't take place "all at once". And, of course, since they are using pulse width modulation to achieve grey scales, any given pixel can be written many times in a given display frame. It wouldn't surprise me if the video processors only delay the data as much as they need to finish the processing algorithms (so the initial RAM write takes place a few video linetimes after the actual video data first appears).
Atlant 17:55, 29 November 2006 (UTC)
That's not accurate. Each frame is loaded into a buffer within the DMD chip and then loaded onto the mirrors themselves all at once by removing the bias charge. Also, video processors need to buffer a frame or two to do de-interlacing/scaling/motion interpolation properly. --Dtcdthingy 18:20, 1 December 2006 (UTC)

[edit] Are there actually cons?

The Cons section essentially states that "LED is the end of all troubles." Does someone have a citation for this? Supposedly according to a different part of the article LED DLP projectors are not even released yet. Iamthebob 02:17, 28 July 2006 (UTC)

LED DLPs are released. I just got my Samsung HL-S5679W yesterday. It is the end of all of those troubled listed in the CON section, but my particular set introduced some new problems in Color banding which probably has more to do with the image processing than the fact that the light source is LEDs. Eric.frederich 16:31, 24 August 2006 (UTC)
The article makes a zillion references to LED driven DLPs eliminating rainbow effect, but is this actually true? All the LEDs do are replace the color wheel with a sequence of flashing lights. The image is still an additive process with different colors being projected at different times. I'm sure that the LEDS can flash faster than a color wheel, so as to even further minimize the rainbow effect except for those with superhuman vision, but the way the article is worded implies that the rainbow effect is just not even possible with LED DLPs. I don't believe that to be true. 216.85.117.123 00:22, 12 September 2006 (UTC)
The preceding comment is actually me before logging in. adavidw 00:23, 12 September 2006 (UTC)
I added that material. Your understanding is basically correct -- you'd need superhuman vision to see a rainbow effect with a LED DLP. Feel free to edit if you feel I didn't get that across clearly. Jerry Kindall 20:50, 25 October 2006 (UTC)
That's really excellent wording. Thanks! adavidw 02:32, 26 October 2006 (UTC)

[edit] How do the damned things work?

I came here hoping to find something more than the marketroid stuff written by TI/DLP themselves, after discovering that practically everything on the web is just a regurgitation of that: was disappointed, of course. Well, there is additional material, including the pro/con stuff, but what's missing, at least for me, is some basic information, like: how does this technology work? What operates the hinged mirrors? Some kind of tiny electrostatic motor? I'm very curious to know this. +ILike2BeAnonymous 00:36, 10 September 2006 (UTC)

To partly answer my own question, I did find some stuff on the How Stuff Works page on the subject; it has an intriguing illustration of the mirror assembly, but a totally inadequate explanation. Apparently, it does operate electrostatically as I had guessed, but I need details, man, details! +ILike2BeAnonymous 00:47, 10 September 2006 (UTC)
Yes, electrostaics drive the tiny mirrors. The mirrors themselves are made from a layer of aluminum metalization which is then etched into the required shape and then under-etched to make the space for the mirrors to tilt. The mirrors are suspended by tiny portions of the aluminum metalization that are left to remain as a sort of "taut band" hinge (like a taut band-suspended d'Arsonval meter movement). Because the suspension bands are very thin, they have an adequate life before metal fatigue sets in. Because the distance between the mirrors and the substrate is so small, the voltage required to flip the mirror is also miniscule, so when we say "electrostatics", we're still only talking about IC-level voltages.
By tipping a few degrees left or right (and banging up against a mechanical stop), each mirror simply steers its portion of the light to the screen or to a "light sink". The switches are purely binary; grey scales are achieved by pulse width modulating the mirrors, with the mirrors switching thousands of times per second.
All of this is documented in technical papers that you can find somewhere on the the TI site; Google found them for me once and can probably find them for you as well.
Atlant 01:46, 10 September 2006 (UTC)


The details you're looking for are in Digital Micromirror Device --Dtcdthingy 18:16, 1 December 2006 (UTC)

[edit] NuVision LED spam and forward-looking statements

All the contributions of User:65.118.181.114 [2] are about a not-yet-out NuVision DLP projector with LEDs, and therefore probably classifiable as commerical spam. That's why I marked some of them with citation needed (fact) tags. They should probably just be removed unless somehas has a published source. Dicklyon 21:42, 12 September 2006 (UTC)

[edit] LED?

Looks like the Samsung LED stuff was also all forward-looking spam, since these still don't exist. Please put correct info back when it becomes verifiable, not before. Dicklyon 13:53, 24 September 2006 (UTC)

wasn't me in the 1st place, but looks like there are some. [3]

[edit] Lights Dimming With Age

I've done a few searches on this online. I've found a few pages here and there that say that Halogen/incandescent lights do not dim with age and a few that say they do dim with age. This, to me, means that the evidence is very inconclusive. Now, I've read on a few forums that people complain about dimming lights after they've used their bulbs for a few months (the same complaint holds true with rear projection), but that dust collects on the bulb and wiping the dust off on a regular basis will alleviate the dimming problem. For the time being, I've removed "* The picture dims as the lamp deteriorates with age." statement from the cons, because I've found no concrete evidence to suggest that the halogen bulbs used in such TVs do, in fact, dim with age.Kakomu 12:22, 27 September 2006 (UTC)

Plus, that "con" also applies to other projection technologies such as LCD. Not specific to DLP. Dicklyon 12:17, 27 September 2006 (UTC)
info:
do not dim
http://wi.essortment.com/headlightcar_rxkh.htm
http://www.burnyourbonus.info/alt.energy.homepower/thread75.html
do dim
http://www.safetyissues.com/magazine/2004/11/16th/cnsmr5Star.htm
http://www.christiedigital.com/AMEN/technologyExplained/theProjector/projectorBasics/Lumens.htm
So, as you can see, there's a general disagreement. Especially considering two websites, talking about halogens in car headlights, saying two completely opposite things.Kakomu 12:22, 27 September 2006 (UTC)

[edit] Colour-absorbing screen

One con that seems to not have been mentioned is the DLP's poor performance when there is a lot of light in the room (eg from windows). Apparently this can now be fixed - Sony's developed a screen which absorbs all colours except for red, green and blue, which means that almost all of the ambient light (only a small part of which consists of those exact wavelengths) is absorbed into the screen. Link: [4]. Anyway, I just thought that maybe this info and the link could be added into the article somehow. Esn 12:08, 23 November 2006 (UTC)

This problem is not unique to DLP projectors, of course. Jerry Kindall 00:18, 28 November 2006 (UTC)
Since light is only made up from Red, Green, and Blue, how does this magic screen work? The ambient light would be made from the same three colors unless your neighborhood uses an alternate form of physics.
That is not correct. Light comes in all colors, but because our eyes are only sensitive to RGB, we can 'fake' all colors with just RG&B. Check out the wikipedia articles on spectrum or Emission spectrum for more details.

No. These screens are not color selective. The technique is based on retro-reflection. Do your research! Developed by DNP [5], not Sony. Marcus The box (talk) 01:04, 6 March 2008 (UTC)

Um, how about you do your own research! Sony have indeed developed a 'black' screen which uses a different technology to the DNP screens. This is all already mentioned in the Projection Screen article. 122.107.20.56 (talk) 01:41, 10 June 2008 (UTC)

[edit] PIP

Do TV's offer picture in picture anymore? I had heard that DLPs have the capability and do offer it. True? —The preceding unsigned comment was added by Happinesswiki (talk • contribs) .

PIP is done by the video signal generator; it doesn't matter if you display it on DLP or something else. Dicklyon 06:35, 29 November 2006 (UTC)

[edit] Fan noise

A recent editor changed the "con" "Fan noise" to "Some devices may have fan noise"

Are there any DLP devices that don't have fans and fan noise?

Atlant 17:40, 21 December 2006 (UTC)

I'm sure all devices have fans (although I guess it's possible for the LED driven ones to not use one). However, it's entirely possible to have a fan without noticeable fan noise, depending on how they build the cabinet. Additionally, whether or not the noise is a con is a subjective judgement that would depend entirely on the noise level and one's own tolerance. adavidw 07:12, 22 December 2006 (UTC)