This is topic DLP projectors and rainbow effect in forum Digital Cinema Forum at Film-Tech Forum ARCHIVE.
To visit this topic, use this URL:
https://ft-forum.com/ft/cgi-bin/ubb/ultimatebb.cgi?ubb=get_topic;f=16;t=000236
Posted by Thomas Pitt (Member # 4322) on 07-14-2008, 03:58 PM:
Something that's been on my mind for a while now - when I went to see Speed Racer, it was a digital 'print'. I couldn't help noticing that if my eyes darted back and forth across the image, there was a noticeable 'rainbow' round sharp edges as if the three colors were splitting up. It was particularly noticeable on Speed Racer because of all the scenes of people's heads floating across the screen ![[Wink]](wink.gif)
I also saw Journey to the Center of the Earth 3D today, in a different cinema. That too exhibited some signs of the rainbow effect if my eyes were moved rapidly around the screen, though it didn't seem as severe as on Speed Racer.
I know you often get the rainbow effect with consumer DLP projectors* (using a single chip and a color wheel), but I thought all theater-grade DLP projectors would use 3 chips.
Anyone know why the rainbow effect would crop up in digitally projected movies?
* The rainbow effect in the theatrical presentations was nowhere near as severe as I've seen on consumer projectors
Posted by Jon P. Inghram (Member # 4125) on 07-14-2008, 04:14 PM:
The mirrors on a DLP chip rapidly toggle between "on" and "off" to simulate different levels of brightness, so that might be the cause.
And, not knowing how the projectors are specifically designed, I can imagine it might be that they load the new frames into the chips sequentially which might introduce enough of a delay between chips to cause a mild rainbow effect, but that's just a wild guess.
Posted by Bruce Hansen (Member # 281) on 07-14-2008, 04:30 PM:
I know that Journey is 3D, was Speed 3D as well? Could this effect have something to do with the polarizer that is in front of the projector switching back and forth?
Posted by Marco Giustini (Member # 4544) on 07-14-2008, 06:15 PM:
Strange, rainbow should nor affect at all 3DLP projectors...
Posted by Mark Gulbrandsen (Member # 72) on 07-14-2008, 07:57 PM:
You may have been seeing some moire effect. Where either the pixels of the DMD's or something in the image itself was the same size as the perforation in the screen. You then get moire effect... a form of artifact. NTSC television also suffers badly from this artifact.
Mark
Posted by Tristan Lane (Member # 1169) on 07-14-2008, 10:44 PM:
Normally, this type of problem is only reserved for single chip DLP projectors, and is associated with the color wheel.
The Dolby 3D wheel only has a slight tint of red and green, so I doubt it's the cause. Either way, Thomas stated that he noticed the effect during Speed Racer, which everyone knows is a 2D movie.
It could be a convergence issue. Convergence is the actual alignment of the chips to each other. Usually convergence problems are visible regardless of eye movement. It's possible that it's only slightly out of adjustment, and by moving your eyes you are able to notice the problem.
All purely speculation of course. It could be as simple as a color setting, or playback issue.
Posted by Mark Gulbrandsen (Member # 72) on 07-15-2008, 07:42 AM:
Actually there are a few test patterns that will generate moire on the right screen surface. The first time I saw this generated from 3-DLP was in the Dolby Screening room during a training class. I also noticed it once at a local theater during Ratatiollie. So what ever type of screen surface they are/were using at that time the moire was more pronounced.
Mark
Posted by Ian Leach (Member # 3382) on 09-26-2008, 08:54 PM:
I also noticed this rainbow effect during Eagle Eye. It was most apparent in brown or orange objects that were bordered by dark objects. When my eyes darted from one side of the screen to the other, vertical red and green stripes could be seen in the dark area.
I've seen the rainbow effect on single chip DLP's before, and this was nowhere as severe.
Funny thing was that it was showing on a dual-projector Digital IMAX system.
My only thought was that since there were two DLP projectors shooting the same picture, maybe there was some latentcy between the two? The picture was sharp and clear, no problems with convergence.
Posted by Steve Guttag (Member # 268) on 09-26-2008, 09:57 PM:
I've noted that RED is more difficult to converge on DLPs...that is, you can get a red fringe on either side of the pixel...I've seen this show up as a color fringe, if you are looking for it.
Steve
Posted by Mark Gulbrandsen (Member # 72) on 09-26-2008, 11:18 PM:
That is because red is ever so slightly out of focus as per green and blue... due to the way T.I. wants the prism blocks built. I woulda prefered that blue be out of focus as it only accounts for about 12% of what we see in the image... our eyes are not as sensitive to blue.
Mark
Posted by Joe Redifer (Member # 3) on 09-27-2008, 12:55 AM:
Mark, the proper term is: "Our Eyes Suck At Blue".
Posted by Mark Gulbrandsen (Member # 72) on 09-27-2008, 11:08 PM:
Joe,
My eyes don't suck at anything! The funny thing about the focal planes of the DMD's is that the lens manufacturers could easily take this into account when designing the lens. This sort of thing was done for the Technicolor Camera lenses where the focal planes of the three colors all HAD TO arrive at different focal points but arrive in absolutely perfect focus...
Mark
Posted by Danial Simmonds (Member # 4638) on 09-30-2008, 09:46 AM:
We played that movie Journey 3D for like 2 months..
Didnt see any rainbow or colour problem..
Mark has good eyesight :-) "Thumbs up" haha...
It most likely could be the colour setting.. i read that somewhere i cant remember. Should use a colour meter when doing this.
Regards
Dan
Posted by System Notices (Member # 2357) on 10-03-2009, 04:28 PM:
It has been 368 days since the last post.
Posted by Jon P. Inghram (Member # 4125) on 10-03-2009, 04:28 PM:
We've been running CP2000-ZXs for a while now and just the other day I noticed this effect, although not directly on screen.
I was sitting to the rear and side of a projector facing away from it when noticed out of the corner of my eye that when I turned my head the light from the lens (reflected off the port glass and then off my glasses) would break into separate colors.
The easiest way to see it is to take a mirror and wobble it rapidly while viewing the reflection of the light from the lens (preferably reflected off the port glass unless you want to irritate the customers and blind yourself.)
Posted by Steve Guttag (Member # 268) on 10-03-2009, 07:15 PM:
That is a completely different sort of rainbow. You are merely seeing the unfocused three DMDs...you'll never see that on a focused image.
The rainbow effect from single chip DLPs will appear on all focused images...particularly ones that have your eyes moving and even more emphasized on B&W images. This comes from the fact that the colors are presented in a time-shifted fashion due to the color wheel.
The CP2000-ZX, like all DCinema projectors are 3-chip DLPs and present all three colors at the same time.
Steve
Posted by Julio Roberto (Member # 4976) on 10-03-2009, 09:45 PM:
Well ... to say all 3 colors "arrive" at the same time, is probably not entirely correct for this type of subtle effect, as the DLP are basically pulse-width binary modulator devices (unlike the Lcos, which are continous liquid modulators even if digitally driven) and the intensity of each color is timed to the amount of time the micro-mirror stays in the "on" position compared to the "off" position.
Of course, to render a particular color, the mirrors must be on-and-off a different amount of time for each primary color, thus, i.e., a 10% red, 70% green, 20% blue color would mean that the red DLP chip remains "on" for ~10% of the time (and off the rest of the time), 70% of the time for the green and 20% for the blue. For a (incredibly small) split of a second, the colors are subtetly off-phase while the chips complete the PWM scan cycle. Our brains then time-fuse the pulses of on-off light into a continuous "gray scale" for each primary color. You know: our brains become a cheap (free) digital-to-analog converter
I don't know if this effect is visible by the human eye at all (happens too fast and only on colors with significant difference in components). If your retina moves too fast (i.e. move the eye with respect to the imager, or like moving the projector of effectively moving a reflected image off a mirror), then one part of the retina would've collected so-many red photons on a red-predominant color while the blue and green chips had not had the chance to reflect enough amount of light (on-cycle) to make-up the correct shade when they "landed" in a different part of the retina (since the "head" or the "image" moved).
Again, this is suppossed to happen so fast that, at worst, could only happen on colors with over 50% of difference in intensity between one of the 3 primaries for less than 288th of a second, perhaps an order of magnitude less depending on how the DLP's are driven (i.e. bitsplitting is used), so I don't think it should be visible as a rainbow, but perhaps as a slight artifact (slight color fringe) at worst.
So the (d)effect is (theoretically) neglible, but it's not "perfect" by the time-based nature of the DLP to produce different shades of "gray", which are different for each individual color. With DLP, there is always SOME (miniscule?) time-based artifacts.
But the effects noticed here are probably a lens aberration/optical, specially if watching the image through viewing glasses (and even more if reflected from non-first-surface mirrors), as all the light is not perfectly color-collimated. Hell, all I have to do is watch my computer monitors through my viewing glasses at an angle (i.e. not straight on 90º) and I can see the subpixels of the LCD going in different directions ...
Posted by Mark Gulbrandsen (Member # 72) on 10-05-2009, 09:37 PM:
Don't forget that at least on Christie projectors the Red (I believe) is just a tad bit out of focus. This is not a design flaw in the prism but just the way physical constraints made things happen in the prism itself. So it is technically impossible for all three colors to be in focus at the same time!! You can set them up individually to be in focus when you look at them but never will you see all three in focus at once...
Mark