Question about 3D and anamorphic lens... - Page 3

The two that you can't buy as a general purchases - AVATAR and MONSTERS VS ALIENS.

And on the curved screen, my GD is 0% and less than 3% on a flat screen.

I understand that, in a perfect world, digital vertical stretch accompanied by optical horizontal stretch cancel each other out leaving you with an enlarged image. Therefor, it was assumed there would be no reason for an anamorphic lens not to work with 3d.

"How well will it work?" is what I believe Mark was honestly trying to answer. He, nor I (probably a lot of people), did not know for sure what would happen to the 3d image if GD was involved. Maybe it would look great in the center but dublet at the egdes where the GD was it its max? I assume producing a proper 3d image is complicated and requires everything to be just right, if something didnt line up just right then maybe its end result wouldn't look "just right" either. These where things a lot of us did't know for sure since we had no experience with it. I only wish Mark had been able to perform the digital vertical stretch to the image as well, just to make sure there are no negative affects from that either.

Please know I'm not trying to "side" with anyone here on who's lens is best or anything like that, I'm just letting you know how I interpreted the whole thing so that maybe you could see it from a 3rd party perspective.

-Sean

I think you have a valid question as applied to some of the technologies. There is a writeup on Panamorph's site I believe regarding lenses and 3D. If it will work with a Panamorph, it will work with Marks, AB's, or Isco's. If there was an issue, specifically related to the stretch factor, then IMO, AB's would have an advantage over all of them, mine (Isco) included, becasue his does stretch more "evenly" (than an Isco anyway). For the polarized version of 3D (passive), I beleive any of them will work fine, becae both Sim2 and Runco were doing it at CEDIA with a polarized PJ stack. Talk about expensive, 2 Lumii, 2 Isco's, 2 CineSlides .

Were you able to document your results with images?

Last night I projected a green version of my test pattern onto the curved AT screen, then placed and alloy track in front of the screen rig so I had a guide to run a flat vertical panel.

I used a tape measure on the screen surface so that each time the projected single pixel display lines projected onto the tape marking. The circles on the curved screen were 350mm wide each representing a zero grid distortion. The flat panel was slightly closer to the projector, so the circles were slightly smaller. There was a slight difference between centre and edge here as would be expected.

I also captured images in RAW and have pasted them into a JPEG.
I have then imported the images into paint, cut and pasted the circles and drawn a perfect circle in single pixel lines (red) over the white circle of the pattern. There is a small difference, however it so small I seriously doubt anyone would actually notice this in motion picture images.
The top 3 circles are the curved screen and the bottom three are from the flat panel that was moved across the screen to capture the circles.

I measured the over scan on the mk4 last night at a TR of 2.45. It over scans onto a 2.37 frame by a good 50mm on each side. 3% x 2.37 is gives me an aspect ratio of 2.445:1. Or thereabouts, give or take a couple of tenths. This is a 135 inch scope (124 inches across x 52.36 inches high)

Yeah, some might not see it, but it's not the point is it. I, like most, like my image to sit nicely onto the frame. You even saw this yourself at my demo room last year at a slightly shorter throw of 2.15:1, your lens then was also over scanning about 40mm on each side of the velvet frame. This means it must have grid distortion period.

The 5E under the same throw, same projector, same screen (same everything) over scans about 10mm.

The 5E clearly has better geometry or less GD (whatever you like to call it) - which is exactly why AB re designed things to start with. You don't need a tape measure to see the difference, it's very easy to see.

No, there was no need for images on my test. I simply used a simple single pixel grid, relatively much finer than the circles in your image. I used a pair of Starrett digital calibers and measured the distance between lines in the grid at the middle, center, and edge. I think I mentioned the grid size in my review, but from memory the grid width was around 3-4" wide. That test shows more about what's going on at a more granular level. The larger circles would have given more of an average error. IMO.

ozth: While I may be corrected by you know who, I don't think that overscanning and grid distortion are necessarily related. One could have for instance a perfect 1.40x expansion at center, middle and sides and have a larger image. Or one can have a smaller expansion in the middle, and a larger expansion at the size, and a perfect image size on overall width.

Anyway, since I've seen no reason to think either condition will affect 3D, this is OT for this thread.

Back to the regular scheduled programming....

Sorry Mark, but numbers speak louder than screen shots. It doesn't matter whether your images are RAW or medium rare, it's only the geometry that's the issue here.


That "slight difference between edge and center" IS the Grid Distortion. The rest we know: use a curved screen and reduce distortion. That's not rocket science.

Unless you're prepared to quantify the difference between edge and center, on a flat screen, at the same throw and throw ratio as the curved screen, then "a slight difference" is not good enough.

The way to do the test is as follows.

1. Accurately measure the throw ratio.
This gives us a benchmark for comparison with other lenses at the same throw ratio.

2. Measure the widths - edge, 3/4 position and center - of smaller grid panels than you have measured.
The smaller the better. Typically, a distance equal to about 1/16th to 1/20th of the distance from edge to center. Otherwise you're just averaging-out the progressive distortions present in a large distance. That's a no-no. The bigger the size of the grid panel measure the more meaningless are the results.

Something like this:


The full sized version of this consists of 60x60 pixel grid squares. Suggest you make one similar and measure widths as indicated by the white squares.

3. Report your numbers.
Screen shots and statements like "small difference" are, frankly, inadequate for depicting GD, or substantiating any claims about it.

Some immediate observations on your photos are that your screen curve appears to be too steep for the MK-4. The outer projected circles are smaller in width than the inner circle, judging by the "control"outline circles you have superimposed.

If those superimposed outline circles you inserted over the projected circles are all the same size then I don't see how you can say all your projected circles are the same size. The ones on the outer are clearly well within the drawn circles, while the one in the middle is virtually the same size. In other words how can they all be the same size (350mm) if some are smaller than others? It doesn't make sense.

Secondly, placing your piece of flat screen panel out a few inches from the deepest part of your curved screen will make the projected circles appear smaller than they would be on a flat screen. This has a bearing on GD measurement.

The way you have measured this is messy and confused. You are not comparing like with like.

Lastly, I'd be careful of producing photographs (however flawed in execution) that seem to show GD is the same on both a curved and a flat screen in the same photograph at the same throw ratio setup. That is flat-out impossible.

Mark: I'm installing a new Stewart curved screen in my demo room. Due to me on the 24th. I'll revist how it affects the Geometry on the Isco then if you like. Or anything other lens as time permits. I sell a lot of curved screens, so I'm very interested in how this affectd this particular parameter.

Yes and no.

You're right in that a perfect 1.4 - 1.4 -1.4 (Center-Middle-Edge) image is distortion free, but too big for a 1.33x system.

However, in anamorphic lenses the edge distortion is the one that tends to get bigger as the throw ratio gets smaller. Center and middle distortions don't change so much.

So, in that sense, something which was designed to give an AR of exactly 2.37 at, say, a TR of 2.0, will give a slightly wider AR at, say, TR=1.6, mostly due to the increase in grid distortion at the edges.

Hence, a lens which is designed to exhibit less grid distortion overall, especially at the edges will vary less in AR over a range of TRs than one that is not designed to do so.

Interesting that when you posted screen caps they were great and when I produced screen caps of similar quality in the now closed MK4 thread, suddenly screen caps are irrelevant.

Perhaps moving this debate over to a relevant thread like this one below might be the best move..

Cylindrical A-lens owners thread
http://www.avsforum.com/avs-vb/showthread.php?t=1261584

Red herring.

My image screen caps were not produced to "prove" forensically that geometry was correct. They were produced to show contrast, sharpness, etc. I made no claims about geometry by invoking them. In fact I noted that the geometry was all over the place due to the camera lens used.

In fact, the first words of that post were ironic. Perhaps I should have put a smiley in there and you would have got it?

"The obligatory screen shots..."

http://www.avsforum.com/avs-vb/showt...5#post19328065

When some claims were made about geometry, I got someone else to do the test. Numbers plus a test methodology were provided and opened to comment.

Screen caps prove not very much. They're a tradition, that's all. In your screen caps above the red (superimposed) circles are mostly 320 pixels wide. It's clear that the white projected circles are of more varying sizes. I measured them (Width x Height) [in pixels]:

Left to right upper: 312x325, 319x325, 313x323.

Left to right lower: 308x318, 314x321, 311x320.

Yet you claim at least the upper ones are all the same size. They are egg-shaped, taller than they are wide by about 4% at either side, more rounded in the middle. They are distorted. And they are not the same size.

Did you think the claim that: "The circles on the curved screen were 350mm wide each representing a zero grid distortion." wouldn't be checked? They are different sizes, by several percent.

If you excuse this by saying "They're only screen shots", then I think that adds weight my point about the value of screen shots.

Agreed.


Point taken.

...back to the topic of this thread.

It seems that there is a lot of interest in two projector implementations of 3D. Of course this also means that in order to achieve a scope image you would then need two anamorphic lenses, one for each projector. Obviously this would be expensive as a good A-lens does not come cheap.

I was thinking about this and had an idea I would like to through out. At the heart of a 3-chip lcd light engine is a prism assembly. This prism (a cross dichroic or x-cube) is able to combine the three color channels, each entering the prism at 90 degree angles to each other, to create a single image that then travels through the projectors primary lens.

So instead of aiming the two projectors in toward the screen, couldn't you actually aim them towards each other with the proper prism sitting in between them? Basically working in the same fashion as an lcd light engine? The prism would combine the right/left images and allow for a single A-lens to be placed in front of the prism.
See attached diagram.

I realize the prism would have to be 3-4x larger than the one used in a typical light engine, but it seems like it could be easily constructed. For example, it is my understanding that a cross dichroic prism (x-cube) is just 4 right angle prisms that have been fused together. Surplusshed.com has a right angle prism that is 50mm on 2 sides with a hypotenuse of 70mm. Four of these together would result in a cube that is 70mm on a side. The item is PM1045 on their site and is not overly expensive. I suppose cost could climb as one needs to optically fuse them together and possibly add an anti-reflective coating.

I really have no idea if this would work. Obviously, the polarity of each projectors image would need to be maintained as it passes through the prism, and who knows what would happen to image clarity. I only wish I had two projectors to try this out.

The prism idea isn't a bad one, theoretically feasible I s'pose, but you'd have an awful lot of lining up to do. The slightest asymmetry on either side of the prism (I mean down to fractions of a mm or degree) and the whole thing would probably go out of kilter.

I say this because projector manufacturers can make their own standards for prism alignment and they still screw it up, causing misregistration.

I've seen a dual-projector (non-anamorphic) 3D rig up close and chatted to the guys handling it. It took them all day to align two projectors in a jig they had had specially made. Adding a central prism would be a whole new world of pain, I'd say.

The advantage of 3D at the moment is that most of the available program material is awful.

If I'm ever inclined to take it up, I'd wait for 2nd or 3rd generation machines that can vertically stretch 3D-120Hz straight off the bat. Might be something decent to watch then, too.

We must have been really lucky then. When I first bought my BenQ W5000, I got together with another forum member who had just bought the W20000. We double stacked the two projectors on a wooden rack and used lens shift to get them vertically aligned. As I said it was luck as we managed to also get the horizontal alignment right as well to the pixel! I bet that we would never get it that close again.

I liked the rig SIM 2 had for their dual projector 3D stack.

Speaking about the SIM2 stack, the system I saw used the Dolby 3D passive system (under a different marketing name) where they had moveable filters in the light path verses electronic controlled 'wheel' used in the D-Cinema version.

Question is, who makes the 3D processing box to feed the two projectors?

Mark,

The 3D-XL from Optoma should do the trick if it is ever released here in the states. That's what I'll be using for my temp setup. It will work either with a single 720p 3D Ready projecor or using 2 boxes, will feed left eye to one 1080p projector and right eye to the other. I have not seen one up close and I am just going with what I have read.

I will eventually go this route, but I have ONE ISCO III and don't really feel like popping to a second plus an additional projector just yet.

I know I'm probably getting way ahead of myself here. Assuming (yeah we all know what that breaks down to) the X Cube Prism thing actually works, how does one get signal to feed the left eye to the left projector and the right eye to the right projector? Do all 3D BD players have 2 outputs?

The images of that box you mention show only one output, so too me, represents a way of converting a 120Hz TV/projector into a 3D TV. This is OK if you have a single display. It does not appear to be viable if you plan on using 2 x 1080/60P projectors.

As far as I know, you would have to do that with a computer and highend video card. There is software that will split left eye/right eye on seperate outputs. The problem is, most of these cards use dual link DVI, NOT HDMI. You would probably need two cards with the mini HDMI port or use a DVI/HDMI adpater. This will work for passive 3D using filters. My guess is you would place the polarizing filter after the anamorphic lens.

Mark, on the back of the box is a switch that actually allows for either 720p output OR Left or Right 1080p output for passive 3D.

I found this thread on AVS:

http://www.avsforum.com/avs-vb/showthread.php?t=1280393

Post number 2 states the following:

"This would require an up-front 3D processor that could separate the left and right video channels and output them separately to the two projectors. Such a product exists, in the new Optoma 3D-XL box that was demoed at Cedia."

"...two of the Optoma 3D-XL boxes would be needed (one for each video channel) and they would both need to be fed the full 3D stream by an HDMI splitter."

If you used a newer 3D Blu Ray Player with dual HDMI outs, I assume (and I agree with you Mark about "assuming" ) that one HDMI could go to 3D-XL #1 and the other HDMI could go to 3D-XL #2. That still leaves an issue of getting lossless audio into your receiver, unless your BD player has analog outs and you can go that route. I'm sure there are or will be 3D compatible HDMI splitters so that would get your HDMI to each converter box and to your A/V receiver.

For your perusal, here is a pic of the back of the 3D-XL showing the 1080p switch. Of course, until some folks get some things and play with this, well it's all conjecture. And with a dual projector setup, you need polarizing filters for each projector and two anamorphic lenses (if you plan on a CIH which I think we all want on this thread anyway) plus a screen that will work with polarized passive 3D. And the patience of Job to get it all setup and calibrated would probably be needed and I am not aware of any retailers, brick and mortar or online, that sell that!

I attached the pic of the back (at least I hope it attached). It is blown up a bit so not razor sharp, but clearly shows a switch for either 720p or L / R Dual 1080p.

I looked at the image. So is that switch 3 position - 720 dual or L 1080P or R 1080P? If it two position, how does one assign the left and right in 1080P?


Maybe the SIM 2 rig used that as well?


For the cost of AVRs these day, I'll probably upgrade to a new HDMI 1.4 AVR before long anyway. Since going Active LCRs, I now have 3 passive speaker cables routed to the front of the room that are not used. Prehaps the new AVR will offer height channels. Do those Optoma units have HDMI Audio out by chance? If not, the best to source a player with dual outs, use the spliter on the video and run the audio direct to the AVR.

Thanks for that. I might have to have snoop around in the 3D section to see whi else is looking for a passive (flicker free) way to get 3D from dual projection.

So if anyone was really wanting to do this, Pansonic have a dual HDMI out 3D player. It would awesome if the outs were assignable as left and right eye. At this stage they are not.

That pushes me more towards the HD250 than instead of the RS40. At heart I know the RS40 is a better projector, but I am not a big fan of 3D anyway, and if I converted, and still couldn't use it with Cinemascope, I probably wouldn't need that feature


/a bit off topic.

If you can still get the pre order pricing it's not much more for the 40. It's only a matter of time before 3D "2.35 friendly" gear starts coming to market. I know i have a great 2D projection system. I watch Shrek 4 couple nights back in 2D even though I have the Samsung 3D 4 pack. I think Cloudy is my only other 3D 2.35 movie. Have about 8 others all 16:9. When I play 3D video games, I don't worry about stretched picture. Super Stardust HD is insane in 3D 2.35. 3D is still a casual treat for us. I never watch 3d on the plasma and it's a family and friends treat with the rs40. But even though right now it's one out of twenty movies, I'm set for the future. It's an inconvenience but you can zoom for 3D 2.35 for now.

Or get the OPPO Whilst some Java discs won't V Stretch, "Cloudy" did and you need to see that to see the potential 3D Scope really has. I'm sure the Java issue is FW and can be fixed soon as the 2D machine has no issues.

I know. But I talked with an Anchorbay engineer and he wasn't sure if they would be able to scale with a FW upgrade on the Duo and Edge. I like the idea of the streaming with Netflix and Vudu thru the 93. I'm going to sell my 83 and get a 93. It's makes sense that FW would be the way to go but there's a lot of uncertainty without having a definitive answer.

Just to play devils advocate, why would you get a worse 2D projector if the better 2D projector had a 3D limitation, when you don't care for 3D?

Exactly don't utilise the 3D.

I don't think anyone saw this coming. Partly because the whole Scope scene is niche anyway, but also because 3D started out for D-Cinema and DCI does not include the use of A-Lenses. Crap all round then.