Welcome to our community

Be a part of something great, join today!

Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

Capture 2 Streams simultaneously for 3D

Hi Istavan; I certainly know where you are coming from and this ultimately leads into a discussion of whether other more classical methods of stereo projection would be better for 3d cinema than the current RealD system.

In terms of quality at least in my experience for projection screens smaller than 20ft, (that’s the largest I can personally vouch for) I still believe a true frame sequential system is superior to any polarized dual passive system (but the gap is closing). The reason is very simple, in a frame sequential system using active eyewear you don’t get any cross talk or ghosting (or at least substantially less than polarized systems). In such a system there is no polarizer in front of the projector. As you know, when the left frame is projected to the screen, for a 60th of second ONLY the left frame is on screen, the shutter glasses have the right eye go dark and the left eye is clear to view what is on screen. Then in a frame sequential system when the right frame is presented it is ONLY the right frame that is on screen for the next 60th of a second, the active eyewear switches to block the left eye and the right eye goes clear. In stark contrast with a polarized system you always have two images on the screen at the same time and each eye still receives a little bit of light from the wrong image, especially with subjects on a dark background. The clear density of CrystalEyes is less than having a polarizer in front of the projector and having another polarizer in front of your eye, but you also have to take into account the effective time the image is on the screen. So the argument about using two projectors VS one to get more light to the screen in this scenario is not clear cut (as you might think) especially when you have to consider an effective loss of contrast ratio for various parts of the image from cross talk/ghosting. [Older VR systems that use CRT projection and now 120HZ projection also look great because you can look at the screen from wider angles than polarized systems.]

We also found for smaller screens of the order of 10ft or so that when using dual projectors that were over 5000 ansi lumens (each) that the linear polarization started to break down and the stereo effect got weaker. This is because of the ratio of scattered light vs. polarized that hits the screen (remember that the polarized filter still lets through a certain amount of diffuse/non-polarized light), it seems that the more amped up projectors increase the ratio of scattered light so we were advised to keep it under 4000 ansi lumens (this was for a linear polarized system). So it’s not strictly true that more light = better stereo for linear polarized systems, but I agree that many 3d projection rigs that one sees are often a bit dim and feeble.

The RealD system is different from both dual passive projection and also different from a frame sequential system that uses active shutter glasses. Here (as you know) the active shutter is placed in front of the projector (Lenny Lipton’s/RealD’s Z-screen) and the special panel creates alternate circular polarizations for sequential left and right frames, clockwise for the right eye and counter-clockwise for the left eye. So in this type of system there will be substantially weaker imagery and more cross talk as compared to a dual projection system with linear polarization. However, it offers the only practical solution from a business perspective for cinema owners to go digital and implement 3d. If 3d turns out to be a real hit for the long term then maybe in the future RealD and others will provide dual projection systems for cinema owners for an enhanced stereoscopic experience (assuming sufficient demand).

I think one of the biggest factors for good stereo is high resolution (this is something I have been pushing for, for a long time). This includes image acquisition as well as display. The human stereo visual system relies very heavily on very fine high frequency image detail to generate a clear 3d percept; if the resolution is not there to represent these fine 3d features then the stereo percept will be weak. Stereoscopic perception of a subject (when projected) is greatly improved even when you go from 1024 x 768 to 1280 x 1024 (our source imagery far exceeds these resolutions).

So as you say (ISTVAN) when you saw twin 4k projectors linear polarized (at NAB) it must have been practically a religious experience! At such resolutions (assuming good source imagery), it is almost a new or different level of reality. I don’t get out much, and have not seen the 4k projection rigs yet, but have worked on multi projector VR systems that have a similar effect.
 
So in my personal (current) ranking of methods of stereo projection systems purely in terms of stereo image quality, for medium sized set ups, and this does change a bit depending on the specific application, but I will still go

1. Frame sequential + (good) shutter glasses.

2. Linear polarized dual projection

3. RealD/Z screen frame sequential circular polarization.

I understand your concern that 3d cinema may die out because RealD does not provide the rich psychovisual experience of a dual projection system with linear polarization. However RealD is the only practical solution (from a business perspective) that will pave the way (I believe) for a permanent 3d cinematic industry (hats off to them). If it sticks, better systems will come, so people need to make really good 3d movies now!

On a practical note if you are driving dual passive systems from a graphics card and have a frame sequential monitor you have to choose one mode or the other, as a graphics card cannot do both unless you use a CYVIZ/Christie convertor box ($10k). A total frame sequential system allows you see both your projector and monitors at the same time. The beam splitter monitors are very pricey, Planar Systems (approx $6k on up) and frankly I am a little dubious as to whether perfect alignment of the two monitors with a sheet of glass is really possible. I suspect that there is a significant amount of Y parallax generated in this system, which is not good if you have to spend many hours doing S3d work. At least with a frame sequential monitor the left and right pixels are where they are supposed to be (no Y parallax), and if you use a black background behind your monitor flicker is not apparent either.



So Istvan what are you using to drive your stereo?
 
I've seen a RealD presentation two years ago and the quality was pretty impressive. Keep in mind that it was not permanently installed and setup pretty quickly I guess. But still very good 3d. They then showed Monster House in 3D. I walked out after half the movie, but because I already knew it and the 3D wore off after a while. I also had a front row seat in the corner which didn't help either. But the overall impression of the system was good.
I watched Bolt in 3D a few days ago in a cinema here. They used a shutter DLP system from XpanD. They claim that there's no ghosting with their system, but there is in high contrast areas. Even active shutters can block only so much light. And too much brightness can cause exactly what Eric stated.
In the end it's hard to say which system is superior. I guess RealD has the long term advantage with their cheap glasses. But they need to work on the issues of ghosting, brightness and resolution in order not to lose the momentum of their current success.
Also, the active glasses are much more uncomfortable after 2 hours.
I've read that Sony and RealD are collaborating for a 4k 3D system right now. That's promising if they get it right.

The other and even bigger issue is: Filmmakers really need to learn how to use 3D properly. When watching Bolt, half of the time I didn't even recognize the 3D because of the fast editing. They should consider releasing two different edits of a movie for 2D and 3D.
Like you need a beat or two to get a sense of what you're looking at in a shot, you need one or two seconds to really get a sense of space. And freakishly moving cameras don't help either.
 
Its always very cool to chat on line with Benni and Istvan. Istvan does an amazing amount of leg work to go to nearly all the relevant trade shows and conferences and has had a lot of face time with key technologists that make this stuff. And Benni I don’t think that there is anyone else alive that has more S-3d links at his finger tips (some really funny and obscure ones too) and has a really good feel and understanding of the technology. As I am usually very focused on getting our first software(s) and cameras out the door I don’t very often get a chance to really scout around to see what everyone is doing in this industry.

So the XpanD stuff is real eye opener for me, (I have been asleep at the wheel), I went to check out their website www.xpandcinema.com , will go back to really take it in.

It’s a very interesting business model and clearly one of the main issues is that with the XpanD system there is potentially a much lower risk to cinema owners as they don’t have to install a silver screen. For smaller cinema screens the price/risk may not be a big issue but for larger screens it is certainly a consideration.

They claim the glasses add only 2.5 cents per viewing, that they have the fastest shutter glasses for the brightest image, and that DC projectors can be set to their normal lamp settings, where as other systems require higher lamp settings, the XpanD system therefore increasing lamp life by at least 80%.

They say “X6D Limited is a European company whose products and services are marketed under the XpanD brand name. In March 2008 X6D Limited has formally announced that the active 3D cinema glasses assets of NuVision Technologies LLC, a widely recognized manufacturer of superior stereoscopic delivery systems with production facilities in Oregon, have been acquired by an Oregon corporation controlled by X6D Limited. Subsequently, all of NuVision's 3D cinema products are now produced under the XpanD brand name.”

So its basically (new) NuVision glasses.

So clearly their business model is to offer a less risky solution than RealD, they even provide financing. However I think RealD will probably be more dominant in the long term as they have an excellent incentive package for Cinemas to go digital in the first place and the 3d is offered as a sort of carrot (very smart).

With the RealD and Sony 4k effort it will be interesting to see what actually emerges. I have heard much talk of the “ghost busting” algorithms that RealD want to implement to prevent cross talk, but have not actually seen it or read any papers (as yet) on how the various algorithms are actually implemented. Never the less I think the market is big enough for RealD and XpanD, but I think RealD will have the Lion share of the US market.

Benni I’m glad you pipe in with the problems of stereo and fast cutting. It really does take a bit of time to adjust to a new scene with a complete jump in the range and disposition of parallaxes. Some key people in the industry have tried to poo poo this Idea but it is very real. I think this insane fast cutting is a fad that will probably subside in the next ten years anyway. However I think this issue really needs to be fully tested. A useful set of experiments could be devised to test view accommodation + convergence times for switches between different ranges of parallax with an audience of 40 or more observers. This time taken to accomdate and converge is something that the military have studied quite bit for fighter cockpit simulations, when the pilot has to look at his instruments in the cockpit and then look at a large 3d VR immersive screen with a simulation of the battle space. Basically they found the reaction times to be unrepresentative of what would happen in a real field situation so instead they use very large (15 meter) compound curved, confocal mirror screen(s) where the eyes really focus on a distance set at infinity focus and then back at the instruments in the real cockpit. They basically found that switching from a large range of parallax to a small close range of parallaxes did indeed take more time for the pilot to read the instruments correctly and therefore simulated real world reaction times far more accurately.I think your idea of experimenting with different test edits of the same scene is a really good idea, somthing you should do sooner rather than later (hint, hint, get your 3d on Benni).

The other thing that interests me is the difference between CG stereo movies and Live action. A lot of CG movies use smooth rendering on the characters and sets, so sometimes there is not much 3d high frequency texture to lock onto appart from the occasional texture map. Live action can provide more often very rich 3d texture, but is a much more difficult environment to control (although our software will eventually help with this). I think when they/we are able to engineer more precise camera systems then the stereographic quality of live action will improve dramatically and will give synthetic CG movies a run for their money in terms of 3d clarity.

I was really interested in your comment "They then showed Monster House in 3D. I walked out after half the movie, but because I already knew it and the 3D wore off after a while." (Front row viewing for 3d should be banned). Really interested in the 3d wearing off effect?
 
Eric, I envy you for the long contributions you are always ready to give. As you remember I have less patient so I will be much shorter.:))

We agree on so many things, but not on the dual vs sequential solution. You described the differences perfectly, only I get as conclusion to the opposite preferences. By the way I think the screen material has much more impact on the quality of the image. I recently started to use the Harkness screens with incredible improvement compared to other type of silverscreens. http://www.harkness-screens.us/

10ft screen and dual 5000Ansi lumens, well I haven’t tried it yet, but I can follow your point and believe you probably are right with it. (whereas I still would like to test it with the Harkness-screens, they are surprising in the quality). Anyway “more light” was not intended as more versus infinitum, but for sure more then 1200 Ansi Lumens as single projector. I will try to find out what Harkness is saying on the 5000 Ansi lumens.

RealD/Zscreen: I’m not so sure if they will be the winner. The present solution is week, eats much too much of the lights and reduces dramatically the benefits of the stereo experience. If RealD, then it should be the Zscreen XL. I saw it in work during a short presentation I think at Sisgraph (but I’m not anymore so sure if it was there or in some other event) This is a good improvement, but as far as I know they are still far from implementing it into the theaters.

The alternative is Dolby. I never liked the Infitec filters they use, and I was convinced they would never really work for cinema, BUT Dolby found a software solution to rebalance the colors after they passed the filters. I have no clue how they did it, but they did it and the light-throu-put is much better then with the Zscreen. Businesswise remember, Dolby is already present worldwide in the theaters, they have a much stronger sales position then RealD. To convince the theaters by telling them that they can use the existing screens is a strong argument in their favor. So the game is not yet won by Zscreen, they only started it.

Somewhere later you wrote about XpanD. They are around since some time but didn’t had so much success. The present glasses cut off even more light then all the systems we talked about. The newer models, the red ones which we can see on their web-pages should improve something and is supposed to come out later this year. I don’t remember when. Anyway it happened to me allready twice during a show that the batteries went down (even if they claimed on a conference that this never can happen)

BTW, I don’t advocate that in theaters (!) there should be put the dual systems. The dual system I consider perfect for a filmmaker who wants to study the s3D possibilities. In theaters I would prefer in short term if the existing systems would be installed in smaller theaters, to give the benefits of higher light throu-put. On long term I hope to see in 2-3 years the first laser projectors to appear. Then we will really start to dance in the cinemas.

On one thing we always both agreed: one of the most important part in a good s3D experience is the empathical perception the texture of a materials is giving us. To get to that indeed high resolution is a conditio sine qua nonl! But, to see the s3D results of that high resolution you need light, light, light. The possibility to feel the texture in s3D reduces dramatically as the light lowers. Most people are focused in looking for the s3D effect on the objects and actors and think “thats good enough”. But I believe for a filmmaker the real improvement for the storrytelling will come when we can freely control the texture in s3D (and this means high resolution and much more lights)

BTW, I think you do agree that more light helps also to reduce the conversion-accommodation missmatch, don’t you?

Beam-splitter monitors. Indeed Planar is asking a lot of money, but Benni is living in Germany and Mr. Stahl in Hamburg is offering a solution for around 2000.-€ with an excellent beam-splitter. You are right,this solution is not so perfect, but 10 classes better then all the frame sequential monitors I saw till today (except maybe the prototype of Kerner Opticals). I have 6 Hyundai s3D monitors staying around and nobody in my company is willing to use them. They prefer to wait till or the screeningroom or the beam-splitter solutions are free. I will have to give the monitors to some kids of my employees as present for gaming.

OK, enough, I wrote more then I should
Istvan
 
I’m glad I squeezed a few more words out of you! [very interesting, Z screen, Dolby, XpanD, laser projectors and last but not least Kerner Optical]. Admittedly on other forums I don’t post embarrassingly long posts, but here there really are some technical issues that need to be addressed and discussed that you won’t find in a book, journal article or on the web. (length of post = too much coffe in the morning!)

I own and use for six years now a Harkness-Hall silver screen about 10ft wide. I normally use dual passive projectors Epson EMP 9100s with CYVIZ convertor box. I normally use CRT’s frame sequential for our 3d/stereo photogrammetric software and are looking to switch or test the new Viewsonic monitors as they are the first LCD monitor commercially available that is true full frame, frame sequential (i.e. not ½ res horizontal line interlaced stereo) [I agree with you that up until very recently commercially available frame sequential (flat panel) monitors are visually not very good due to the horizontal res problem, we still use our SONY and Viewsonic CRTs and the image quality is really good, hoping that the Viewsonics will be comparable.]. I will definitely be purchasing a 120HZ Dlp projector, as well as some XpanD glasses (to compare with CrystalEyes, Xpand glasses seem very reasonably priced). I do expect to find less ghosting (mainly from my experience with CRT projectors/shutter glasses), than my dual passive system. For small exhibits the 120Hz active shutter solution seems cheaper and more portable to me which is important if I want to start “pushing” my software.

When I have purchased my first 120HZ projector you can definitely come over (some time when you are in L.A.) and I will set up a side by side comparison of dual passive with active single (bring your 3d movies!). The polarized systems do have a more “magical” or “romantic” feeling (ghosting perhaps) than a true 120hz/shutter glasses system that has by contrast a crisper but visually more complete look that may appear to some to be a bit clinical or “computery” looking. In the work we are doing, we are going for absolute realism at an affordable price.

Whether we use dual passive or single 120 hz active really depends on the size of the audience for us depending on the expense of the glasses we have to provide or rent; it's a very practical issue with us.

Cheers,

Eric
 
About the following

Istvan wrote: “BTW, I think you do agree that more light helps also to reduce the conversion-accommodation missmatch, don’t you?”

I think that would only be an issue if the projection is weak and the ambient lighting conditions of the exhibition space are bright (as is often the case at trade shows), i.e. the surface of the screen is apparent and the other objects in the room are also visually apparent and compete with the projected stereo scene. With my own set ups I go to rather obsessive lengths to completely block out all extraneous light and wherever possible only display in complete darkness, this dramatically improves the 3d experience. S-3d in a board room environment with moderate ambient light will always be weak. However once you have a lighting situation where you only see the image on screen (and nothing else) the amount of light hitting the screen does not affect the geometric realities of the “view accommodation vs. convergence miss match". I.e. once the imagery is correctly presented, cranking up the light output does not improve or have the capability to affect the stereoscopic geometry of view accommodation and convergence. Never the less, as you may remember our software does have the capability to resample all of the parallaxes to rein in the range of parallax closer to the notional plane of the screen, without compressing the fine 3d detail (unlike convention methods), in other words you can achieve visually rich and comfortable viewing.
 
Eric,
as you wrote it: coming from the photogrammertry you are used to look for accutate measuring tools, I'm more interested in the emotion the s3D can bring to us. This is why I prefer a 3 chip LCD projector to a one-chip DLP version. (I'm curious how the rainbow-bug will effect the s3D experience, let me know:) I believe once the s3D will be more established, the adjusting work on the texture perceptions will get developed similar to the color-grading we do today on our films.

(I discussed last week with a professor of the University of Venice how expensive it could be to develop together a sofware which calulats the normals of the texture elements from a photo-stereo-pair. Once those are know it would be possible to develop a grading-station for adjusting the texture image. It was interesting, but with the given economical crises and the fact that today on the RealD system you don't really see the s3D fineness of the texture it would mean throughing away money for nothing. But somebody will do it one day, I'm sure)

BTW, did you ever used the SPAR Polarizers? They achieve on LCD projectors 70% of transmission. http://www.silverfabric3d.de/html/sf_polarizers.htm. I have a couple and I like them (even if I never misured if the 70% is true:)

Relation between light quantity and conversion/accomodation missmatch: well on last years SD&A it was discussed by John O. Merritt how the depth of field effects this missmatch. I totally agree with you that external lights disturbs a lot the s3D experience, but I also found out that indeed the DOF of the projector has to be an important considaration in the balance of the set-up.

One last argument for the dual-system: you can use it with a Macintosh too.
Ciao, Istvan
 
Yes I agree the rainbow effect (even at 120HZ) could become quite interesting especially if you dance around in front of the screen, a sort of fully immersive VR simulation of an acid trip (could become popular in clubs). The filter wheels and signal processing can only go so fast, but I’m hoping 3 chip DLP may come down in price in the next few years (but probably not). The Mitsubishi TVs looked O.K.?

The software we have developed is much closer to what you described, and does it very cleanly. It has the capability to control parallax in new ways such as being able to adjust macro and micro parallax and really came out as an industrial accidental by product of a new type of VR model we were developing. It requires pre processing but once carried out the adjustments can be made in real time (so you probably won’t render an entire movie using this technique but just tricky shots). Regardless of resolution and quality (RealD or not) there will always be the need to control or “grade” parallax in special ways as long as people are shooting live action stereo scenes with a substantial range of parallax. Our patents are doing well but should be able to spill the beans in the next few months.

I think Benni and many others find ourselves in the dilemma of whether or not to invest in certain hardware now or just wait for better more affordable technology to come along. Sometimes the newer technologies lack in quality or the new technologies never come; so the age old question becomes, when is a good time to jump on the “stereo” bus if you are on a budget?

The thing that has been maddening for us, is that over time there have been huge gaps that have opened up in the technology that are only recently being bridged. So for example we have been using CRTs for stereo (great!), but when our software ships (hopefully in 2009) it would be unthinkable to ask our prospective customers to use antiquated CRTs, but it was only in January of this year that the first competent and affordable replacement for CRTs came along in the form of the Viewsonic 120HZ LCDs. We have been sweating this one for several years.

Similarly if you are on a budget you can experiment with weird hybrids of industrial equipment and deal with all the weird “mubo Jumbo” as Benni puts it, or you can wait till the scarlet(s) come out. We had shelved engineering special equipment for the RED ONEs to organize our own stereo shoots because of the Scarlet announcements. Now that these are likely to be delayed, and a new RED Rental outfit has moved to New Mexico we are now gearing up to engineer test systems for using RED Ones to further test our software.

I think the Scarlet(s) could enable an explosion of S-3d applications. The question I put to you guys or anyone else; is there a market for quasi turn-key stereo systems for folks such as Benni or others wanting to learn the technology? I.e. High quality (above amateur) but not million dollar systems and work flow either. Is this “discovery” market for technically minded enthusiasts and experimental film makers valid? Would the RED ONE model (in terms of price performance and access) work for Stereo 3d applications as well, or is this only of interest to three guys hanging out on the 3d thread of REDUSERNET? [I note that this rather closed thread (my fault) has over 700 views, so I feel that someone else might want chime in as well?]
 
Sorry to be late to the thread.
Rave2K offers up to 2 streams of dual stream (4:4:4), up to 2K UNCOMPRESSED
RaveDi will offer up to 4K (releasing soon)

Rave is one of the only products that offers dual, dual stream, using sequential files (native DPX) and offers both slave and master VTR emulation. Rave is much more than just a recorder/player.

To answer your R3d support question that I know your going to ask. Yes we are rolling this in.

Another exciting product is our new 3d Live, which brings in any 2 sync'd SDI streams (HD) and whala 3D images, there on-set. check out the other tools for this product
Here is the link. http://www.spectsoft.com
You can contact me direct for our free version.

Everything we design is geared for professional use (check out who's using) and although we are priced very good for what we do, a couple thousand is not doable. Hardware alone, far exceeds that :)

Hope that helps
Cheers,
Ramona
 
Benni Diez said:
Is there a capture board that can take 2 streams uncompressed video simultaneously and in sync?
Or can I use 2 regular capture boards at the same time. But then, what software do I need to sync the capture?
Click to expand...

IFX's Piranha Stereo-Edition with the proper DVS card, can ingest dual HD-SDI streams from the RED camera, and present them together in real-time for on-set calibration and preview of footage from a 3D rig. This was demonstrated at NAB in their booth. Piranha can also capture, playback, and do real-time color correction w/ LUTS, etc. on the live feeds.

Check out their site at www.ifx.com
 
I missed this :( I was focused only on the rig. Did they say any prices?
 
Having used the Codex portable, Piranha and Rave products I can easily say that Spectsoft is the only one to truly address the complete 3D solution.

The on set 3d live product is indispensable, feature rich and ever evolving with improvements. Ramona is absolutely correct when she said the Rave is much more than just a recorder/player. The 3D live/Rave combo is hard to beat.

In the end as usual your budget and needs will ultimately dictate your decisions. We have used other less expensive solutions to record dual streams but this is done at the expense of creating many post headaches. So in a way you spend money either way.

The codex is great if you need a portable solution, but they don't offer a live viewing system. Thier VFS system can be good thing in many cases but it can also cause some issues. The drive packs are now available with solid state and are easily swapable.
 
You must log in or register to reply here.
Back
Top