MTBS in 2011

[Fredz]

Well I was thinking about a Flash plug-in simply because I am good at Flash and could do it easily. It is also well supported on all browsers, unlike something like WebGL which won't be relevant for years to come.

WebGL won't be relevant for years to come ? You must be kidding, every major player in the browser field is implementing it, it's going to be the only truly open standard for 3D acceleration on browsers for years to come, just like OpenGL ES is right now on mobile phones.

Flash 10.2 is adding 3d hardware acceleration (as in polygons), not native support for stereo 3D. You also cannot link with C/C++ libraries, so it still wouldn't be possible to support non-frame-compatible 3D modes like shutter, iz3d, etc.

I've read several articles talking about the stereo 3D capabilities of the future Flash version, but I didn't find anything to confirm this on the Adobe site unfortunately. In this article it's said that Adobe and NVIDIA did test streaming of 3D videos using an alpha version of Flash 10.2 with 3D Vision though :
http://www.theinquirer.net/inquirer/new ... ming-video" onclick="window.open(this.href);return false;

Nvidia does have a 3D Vision plugin, that works with Silverlight (I think) but I am not sure if there is an API for this. I am also not sure how much better off we would be with a custom plug-in, what would we gain?

When 3D acceleration will be available in browsers with Flash by giving access to Direct3D surfaces, adding 3D Vision support would be quite trivial though thanks to their API.

I mean, even the iz3D developers for all their years of experience cannot get shutter output to work with total control of a Windows service. I highly doubt we would have better luck doing this in a browser.

It seems the iZ3D developers don't have much knowledge in low-level graphics programming at all, solutions for page-flipping and shutter glasses have existed for years in the VR field beside NVIDIA as I've already said on other topics. I could even implement a hardware page-flipping test myself in only two days under Windows and pilot VGA DDC glasses with NV40 and NV50 graphic cards under Linux. Doing all this in a browser would not be any different than in a standalone app either.

For viewing natively I figured Flash could just open a dialog box to save/open a JPS file. Ideally you would already have something like the Nvidia Photo Viewer as the default application. So then you just select "open" and it would open in 3D Vision mode.

That's probably the best short-term solution before other techniques exit from their beta state (Flash 10.2, WebGL, etc.). Using an existing Flash-based photo viewer like those I mentionned could also be a quite easy solution to use, giving functionnality equivalent to YouTube 3D for images.

[cybereality]

WebGL won't be relevant for years to come ? You must be kidding, every major player in the browser field is implementing it, it's going to be the only truly open standard for 3D acceleration on browsers for years to come, just like OpenGL ES is right now on mobile phones.

Its not even a standard yet, let alone fully implemented by all vendors. Most importantly, it is not supported in IE, even in IE9. So we will have to wait until IE10 to *maybe* have support for WebGL. How many years will that be? How many more years will we have to wait for all the people to upgrade their browsers to the latest version (keep in mind a good chunk of users still run IE6). It has a long long way to the 99% penetration rate on Flash Player 9.

[Fredz]

Its not even a standard yet, let alone fully implemented by all vendors.

Yes it's still a work in progress, but it's still a cross-platform, royalty-free web standard. It's not yet fully implemented by all vendors but it's supported by the browsers that make more than 70% of the market as of november 2010. See : http://www.w3schools.com/browsers/browsers_stats.asp" onclick="window.open(this.href);return false;

Most importantly, it is not supported in IE, even in IE9. So we will have to wait until IE10 to *maybe* have support for WebGL. How many years will that be?

PNG has been a standard for lossless images for a long time and IE started to support it correctly only in IE 9. Would you call PNG a failure ? Internet Explorer is no more relevant in the browser market, just like it was some years ago.

How many more years will we have to wait for all the people to upgrade their browsers to the latest version (keep in mind a good chunk of users still run IE6).

That's just plain wrong, IE6 is used only by 4.1% of the users, other versions of IE account for 24.5% of the usage. If you look at the stats for all the browsers, you'll see that the latest version is always the most used and by a very large margin.

It has a long long way to the 99% penetration rate on Flash Player 9.

70% is not bad for a start for WebGL considering its work in progress state, and I really question this 99% penetration number of Flash on the Web, the only numbers that can be found on this subject all come from Adobe.

[cybereality]

That's just plain wrong, IE6 is used only by 4.1% of the users, other versions of IE account for 24.5% of the usage. If you look at the stats for all the browsers, you'll see that the latest version is always the most used and by a very large margin.

I am not so sure about this. The numbers from W3C are skewed because the people that go to that site are all web developers, of course using the latest and greatest browser versions. From what I am seeing, IE6 still accounts for around 16% of the market (all versions of IE are about 50%), which is sizable:
http://news.softpedia.com/news/Chrome-L ... 9200.shtml" onclick="window.open(this.href);return false;
http://en.wikipedia.org/wiki/Internet_E ... sage_share" onclick="window.open(this.href);return false;

That would make IE6 market share over 50% more than Chrome and over twice that of Safari and Opera combined. This is not a negligible number. Of course, this market is also mostly corporate running ancient computers and legacy 16-bit software. Probably not the type of people with $300 video cards or stereoscopic 3D monitors. However it is important to take into account when you are developing any sort of web applications. Of course, you can just drop support and request that these users upgrade to a modern browser. However if you can find a way to support them, or provide acceptable fallbacks, then this widens the potential market. If you are only developing for the bleeding edge, unstable nightly build, then you are only going to get a certain kind of customer. As a developer I would much rather create something that the vast majority of users can enjoy, rather than catering to a small 10% of the "elite". Of course, you have to come to a compromise somewhere, I am not interested if my software will run on Windows ME. But it is important to look at the numbers and make an informed decision.

It has a long long way to the 99% penetration rate on Flash Player 9.

70% is not bad for a start for WebGL considering its work in progress state, and I really question this 99% penetration number of Flash on the Web, the only numbers that can be found on this subject all come from Adobe.

Dude, believe it. Show me one person with a computer that has never been to YouTube and watched a single video. They would need to have Flash installed to do this (beta HTML5 support notwithstanding).

I agree that WebGL is going to be very important in the coming years. I imagine even we might see some experimental stuff in 2011, and commercial projects in 2012. But I don't think it will be ubiquitous within that timeframe, and I doubt any major sites will rely on it as their sole medium. I could see gaming portals as the first early adopters of this standard, and that will happen sooner rather than later. But as it is today, it is nowhere near "production-ready" and the penetration rates are far too low to consider it a viable platform. I am not trying to discount it. Starting developing on it today would be very helpful so you have the experience when it is actually a standard. But shipping a product using it today is naive and not really an option.

[Fredz]

I am not so sure about this. The numbers from W3C are skewed because the people that go to that site are all web developers, of course using the latest and greatest browser versions. From what I am seeing, IE6 still accounts for around 16% of the market (all versions of IE are about 50%), which is sizable

Ok, my reference was a little bit biased, nontheless the market share of 16% you gave for IE6 dates back from August, the number for November 2010 is only 13.72%. But the stats of this site (Net Applications) look really dubious to me, according to them Netscape 6 which is no more maintained since 2002 has more market share than IE 9, how on earth would this be possible ? I don't know how they get their stats, but I really question their validity.

Anyway the other sites referenced by Wikipedia give very low market share to IE6 (W3Counter : 4.93%, StatCounter : 6.44%) so this still confirms what I said before, IE6 market share is marginal at best.

I was wrong about the other versions of IE though, I thought its market share was much lower. Still, WebGL could be made to work on IE thanks to the Chrome Frame plugin, but that's not an ideal solution though.

It would probably be a good idea to have browser stats for MTBS3D to know which browsers are the most used to view the site though.

Dude, believe it. Show me one person with a computer that has never been to YouTube and watched a single video. They would need to have Flash installed to do this (beta HTML5 support notwithstanding).

You are quite right concerning PCs, but there are still at least over 80 million people that have never watched any Flash video with their machine while browsing the Web (using iPhone/iPad). Adobe didn't include these types of devices when they claimed their 99% market share.

Anyway what is important to know is what technologies are available for 3D image viewing on browsers for a maximum of 3D equipements. At this time you can choose between Flash and WebGL and maybe Silverlight (3D Vision Live). Flash and WebGL can support the same type of outputs (frame compatible, interlaced, anaglyph) and there are ongoing tests with WebGL to add frame sequential 3D. Maybe Flash 10.2 will be able to do that too.

Starting developing on it today would be very helpful so you have the experience when it is actually a standard. But shipping a product using it today is naive and not really an option.

Tell that to the guys behind 3DF33D.tv, they are using WebGL right now for 3D pictures and video streaming. They're currently testing page-flipping, which makes them the only site with this type of functionnality on the Web that isn't tied to a proprietary vendor (ie. 3D Vision Live).

[cybereality]

To tell you the truth, I couldn't care less about IE6 and if it will be supported by any MTBS3D plug-ins. I should hope users of stereoscopic equipment were also interested in keeping up with the times in terms of their web browser. I just brought it up to illustrate the issues you get into when developing cross-browser compatible web applications. The flip side of the coin is like the site you linked to: http://www.3df33d.tv/" onclick="window.open(this.href);return false; . When I go there I see a bunch of black boxes and a pop-up telling me to download beta software. That is great that they want to be on the bleeding edge, but with it you are reducing your userbase significantly, in an already small niche market. I guess someone has to be the front-runner, so I don't blame them for trying. In terms of the MTBS3D photo-viewer, I would like to appeal to the greatest audience as to be accessible by new or inexperienced users. This basically means supporting a platform they already have installed, like Flash. I think Silverlight is also a contender, since it has support of Nvidia 3D Vision, but I am not really too familiar with developing Silverlight apps (though I am sure I could learn if necessary). WebGL, while technically maybe the best solution, just does not seem ready for prime-time at this moment. I would almost be more interested in just developing a custom proprietary MTBS3D plug-in that could work across all major browsers (only needing a quick 30 second install) rather than developing something in WebGL that could only be used on unreleased beta software.

[Fredz]

Another technology to keep an eye on for stereo 3D on the Web is the Unity Web Player. It's a browser plugin for 3D rendering (not stereo for now) that supports Internet Explorer, Firefox, Safari, Chrome, Opera, Mozilla, Netscape on Windows 2000/XP/Vista/7 and Safari, Firefox, Mozilla, Netscape, Camino on Mac OS X 10.3.9 or newer : Unity Web Player

What is really nice with this plugin is that it's lightweight (571 KB), that it does work without restarting the browser after installation (tested on Google Chrome) and that it's able to show real-time interactive 3D, which could be a great solution for 360° panoramas. You can find some examples here (but not in 3D stereo) : http://www.ruined.com/" onclick="window.open(this.href);return false;

BigPoint has announced stereo 3D support with shutter glasses for two of their titles (Ruined Online and ToonRacer) which will be available in spring 2011. I don't know if the stereo 3D version of the plugin will be available before this date though :
Bigpoint Brings Stereoscopic 3D Action to Browser Games

For now there is still support for interactive anaglyph rendering as shown here : Stereoscopic Images in Unity3D

I guess it won't be difficult to adapt it to use other anaglyph modes or to render frame compatible content like in YouTube 3D (side-by-side, interlaced, checkerboard, etc.). And since it's interactive, it would also be possible to add control for convergence and separation too I guess.

[cybereality]

Yeah, I've played with Unity a bit. Its pretty cool. It does work perfectly fine with existing 3D driver solutions (Nvidia and IZ3D) but you have to launch in full-screen for certain outputs (I can do windowed 3D on my Zalman w/ iz3d which is pretty nice though). So in terms of stereo 3d gaming on the web, it is here today and works great. Supporting native 3D solutions might be a little more difficult, but I suppose frame-compatible methods would not be an issue. Not sure what type of file access you get w/ Unity though (ie loading JPS files from a server and parsing that to local bitmap objects, saving/opening files in a native OS dialog, etc.). But yeah, its cool technology.

[cybereality]

Just a little update: Work has been dead slow this week, so in my downtime I was able to start development on the MTBS3D Stereo Photo Viewer. It can load JPS or PNS 3D images (actually can be JPG, PNG, GIF or BMP if in the proper full-res side-by-side R|L format). The 3d outputs currently supported are as follows:

- Anaglyph (full color or grayscale) in Red/Cyan, Green/Magenta and Yellow/Blue for use with cheap colored glasses.
- Side-by-Side (half-width frame-compatible) can be used with modern 3D HDTVs or Vuzix Wrap HMDs.
- Over-Under (half-height frame-compatible) can be used with modern 3D HDTVs or shutter glasses with "sync-double" capabilities.
- Horizontal Interlace (row interleaved) can be used with Zalman, Hyundai, Miracube, JVC, LG passive-polarized displays. Also works with shutter glasses in "line-blanking" mode.
- Vertical Interlace (column interleaved) can be used with older auto-stereo displays (Sharp, SeeReal) or even DIY parallax barriers.
- Checkerboard which can be used with "3D Ready" DLP HDTVs from Mitsubishi and Samsung.
- Free-View (crosseye or parallel) can be viewed without any glasses or special hardware.
- Wiggle which basically just cycles back and forth really quick. Looks a little cheesy, but doesn't need any hardware.

So far that covers a pretty good chunk of stereo display solutions. I'm also working on a mirror-split mode (side-by-side or over-under) but it wasn't totally finished. I would also like to expand Anaglyph support to include optimized/dubois and official support for ColorCode and TriOViz solutions (although I did test ColorCode glasses with my Yellow/Blue mode and it was close). SENSIO and RealD formats could also be supported but it would take a performance hit and I am not sure how needed it is when Side-by-Side is compatible. I also think I have a work around to get dual display modes supported. This means that dual passive projection and some HMDs could work. I think native IZ3D mode could also be done (may be processor intensive though). Beyond that I am not sure Flash could handle any other formats. So native support for stuff like Nvidia 3D Vision or HDMI 1.4 is probably not in the cards. However there is the ability to open the image in a native OS dialog box, thus allowing use of the Nvidia Photo Viewer, etc. Not sure what other formats people might be using. If there is something that I missed here that could be implemented in Flash, please let me know and I will add it. After that is done, then I will have to integrate it with the MTBS3D gallery (which I hope won't be to hard). Shouldn't take too long, though. I will post a beta version sometime soon, maybe in a few days.

[Fredz]

Apart for the corrected anaglyphs, mirror and frame sequential modes I don't see anything that would be missing. Congrats on this job !

I'm not sure if you know it but you can use the Dubois algorithm for any type of anaglyph glasses (red/cyan, magenta/green, blue/yellow), it would be a great addition to your viewer. You could also add half-color for anaglyph too since it is used in most viewers and in the NVIDIA driver for 3D Discover.

Did you try to implement frame sequential ? From what I've read the rendering is synchronized to the vertical retrace on Mac OS X so it may be possible to at least have a solution for shutter glasses on this platform (with some sort of activator for the glasses though). You could also try the Flash 10.2 beta to see if there are updates concerning synchronization for other platforms since it supposedly should include Direct3D support.

Any chance you can put a test website online also so we can test the interface ?

[cybereality]

Yes, I certainly want to explore what can be done with anaglyph (dubois, half-color, etc.). You know, I have not bothered with anaglyph much recently, but I bought a bunch of paper glasses (just for testing the image viewer) and anaglyph is not really as bad as I remember. I have a bunch of different high-end red/cyan glasses already (Pro-View, Pro-Ana, Ana-Chrome, etc.) but I haven't used the other colors much. So I got some red/cyan, red/blue, green/magenta and ColorCode glasses. With the proper content they can actually look pretty good. With the ColorCode glasses on this one specific level in UT2004 (mostly gray colored), I swear it looked better than the Zalman. The resolution was certainly a step up. Most levels, however, had bad color loss or retinal rivalry though. Some content is also too dark to see properly. But it leads me to think with the proper optimizations we could have very acceptable quality from these cheap $2 glasses.

In terms of frame-sequential: the "wiggle" mode is basically software pageflipping. I have a delay value to how many frames it stays on for each view. If the value is set to zero, then it alternates between left and right each frame. The issue is that Flash in general cannot hold a steady framerate. Its set to 30fps, but can go as low as 20fps or higher like 35fps. With stuff like video I know Flash can utilize the hardware directly, so in that case I think it can reach smooth 60fps framerates. Not sure how much access I have to that stuff but I can look into it (I just started using Flash 10 right now, so maybe 10.1 or 10.2 have some features I haven't seen). I also thought about maybe using blue-line encoding or something along those lines (that would work with an external activator). However, even if the sync was perfect getting 15Hz per eye would be horrible. Best case scenario would be like 30Hz per eye (since browsers tend to throttle Flash). This is going to change in 10.2, so we can always add support at that time if its possible.

Yes, I plan to put it online soon. I'm trying to get my website back up this weekend, should only take a day or two for the DNS to propagate.

[Zerofool]

Hi, cybereality, I'm happy that you are working on that, and I hope we'll see some results soon.

I just want to say what I found out few days ago, while I was searching for update for all my firefox plugins. I noticed one plugin that I do not remember installing - "sView Browser plugin" . And I found some information about it at these places :

http://forum.iz3d.com/viewtopic.php?t=2134" onclick="window.open(this.href);return false;

http://www.mtbs3d.com/phpBB/viewtopic.php?f=3&t=5231

It appears it's installed together with the player itself, which is a good thing. And it works great, it probably also can play embedded video that way. So I guess that this easy method could be used as as second option (beside the flash viewer), and a link for such formatted page should be available below the flash viewer (for users with devices, unsupported by the flash version). That will solve the compatibility problems with iz3d displays (if it turns out to be hard to implement in flash), as well as page-flipping based ones (although it depends on the hardware when the software mode is used, as usual ), I think it's 3D Vision compatible as well (unless "NVIDIA Direct3D stereo-driver" means the legacy driver ), but I suspect that it will support HD3D in the future as well (at least I hope so). I hope you guys find this idea useful.

example gallery where it works:
http://www.stereo3dgallery.com/cgi_bin/ ... ?PicID=386" onclick="window.open(this.href);return false;

Try on those links, to be sure the image is correct:

http://www.stereo3dgallery.com/cgi_bin/ ... ation=NORM" onclick="window.open(this.href);return false;

http://www.sview.ru/ru/embedded_demo" onclick="window.open(this.href);return false;

Cheers all (and especially sView's dev Kirill Gavrilov)

[Likay]

Really cool! Looking for it's implemention!

[cybereality]

Ok, so I have the alpha version online now. There is not really any GUI, so some functionality cannot be tested even though it works (like swapping the eyes, black & white anaglyph, etc.). Press the arrow keys or spacebar to cycle through images (may take a long while to load, no progress bar yet). Press the "O" key to change output.

http://www.cybereality.com/view3d/" onclick="window.open(this.href);return false;

The sample images I am using were taken directly with the iz3D driver, I did not process them in any way.

[Chiefwinston]

cybereality thank you for your time and efforts in this endeavor. It is appreciated.

I'm surfing to Mtbs3d with my playstation 3 browser on occasion. Sony has a decent not great 3D picture viewer. It seams to only support MPO format images. If this could be incorporated, many with a playstation 3 unit would be able to enjoy th 3D goodness. Heck I don't even know if its practical or possible. But once agian your efforts are appreciated.

cheers everyone

[Chiefwinston]

I should clarify that last comment a little. I'm not trying to view the 3D material in the PS3 browser. I'm saving pictures and trying to view them on my big screen. Once saved I can view them with sony's software. MPO is the format required for some reason. I don't know if thats a common standard or a very proprietary one on sony's part. They have examples of huge panoramic stereo 3D high def. photo's that are pretty nice.

cheers everyone

[Fredz]

The MPO file format is what is used in the Fujifilm cameras, it's also supported by the 3D Vision Photo Viewer and it's been proposed as an open standard by CIPA (Camera & Imaging Products Association).

[cybereality]

Yeah, MPO is proprietary, but luckily I have wrote a utility that will convert them into the JPS format:
http://www.mtbs3d.com/phpBB/viewtopic.p ... 19&p=31457" onclick="window.open(this.href);return false;