The software used on this project is mainly Maya, Photoshop, ImageMagick, and perl scripts on both Linux and OS X.

We agree. Our current system does not allow us to release multiple formats in a simple manner. We chose red/cyan anaglyph as a test case. We are quite glad to see that there is demand for a side-by-side version as well as a YouTube version. For the slashdot crowd, conversion of formats and/or download and play through special software is no problem. The general public is the vast majority of our visitors and they generally just want to push play in the browser. What I suspect would make our website folks happiest is if we could employ something like the YouTube 3D player on our web pages. Your comments will help us make arguments for better 3D support. Thank you.

Thanks for the info, Joseph. You clarified the most important point - that the TV is switched into 3D mode separately from Media Play opening the file and starting it. We will have access to a 3D HDTV for testing shortly (it is also a Samsung, which is why I had looked into Media Play). When ready, I will be glad to post a version for you and others to download. I'll post the link as a comment to this thread and also see if the folks who do the YouTube channel will post it there. Thank you very much for your feedback.

We are interested in producing files that folks could watch directly on their 3D HDTV via USB, but have not figured out the encoding to use. In looking at the Samsung Media Play supported video formats, none of them jump out at me as being side-by-side, top-bottom, or other obviously 3D aspect ratios. They are all listed as 1920x1080 (with a couple 800x600). Hence, one could do a 1/2 width side-by-side or 1/2 height top-bottom, but not a full resolution or the Blu-ray 3D format of 1920x2205 at 24 fps. Have you done 3D off a USB stick on your Samsung? What is the encoding that tells the TV to activate 3D mode? How does the TV know a file is 1/2 width side-by-side 1920x1080 vs a normal 1920x1080 movie?

No, we did not attempt to include relativistic effects. If we were creating a video to explain special/general relativity, we would include them. Otherwise, such effects would be too confusing for the average viewer, especially since this movie has no narration.

Yes, of course the camera separation is much wider than human eye separation. The camera motion is also probably faster than the speed of light. As you correctly infer, scientifically accurate visualizations of what the human eye would see moving at currently achievable speeds would look no different than the original Hubble image. What would be the point in releasing such a visualization?

Thank you for your comments on 3D formats. We did not feel that enough of the public has 3D hardware today, but a reasonable number might have anaglyph glasses. If we do future projects, we will increase the formats as appropriate.

The visualization does uses separate left and right cameras. However, I forgot to mention in the posting that the "3D" is mostly "2.5D". We have no information about what the backside of the nebula looks like, so we could only do full 3D modeling if we artistically created volumes and pixels that Hubble does not observe. We did some of that for the "Hubble 3D" film, but did not invest such time on this project. We did sculpt the front side of the clouds in the nebula into landscapes, but the camera path stays mostly in front, so the sculpting is not that obvious. We are testing to see how much effort is required to get "enough" immersion.

Yes, that line in the post was intentionally ironic. We did not trumpet the 3D in the press release for the general public, but if the post made it to slashdot we would be loud to the tech savvy audience who could give us the best feedback.

The 2D version of the movie is available as well: http://hubblesite.org/newscenter/archive/releases/2010/29/video/a/

OK - Here's the math ...

100 light-years = 1 quadrillion kilometers -- You want a 1 meter resolution at that distance, so you need an angular resolution alpha, where tan(alpha) = 1 / 10^18 --> alpha = 5.7 x 10^-17 degrees

Let's use Hubble as a scaling proxy. It has a 2.5 meter mirror and 1/20th of an arc second resolution. Converting units, that resolution is 1 / (20*60*60) = 1.4 x 10^-5 degrees. Now, simply scale to get the desired resolution and you have the diameter of the mirror = 2.5 * 1.4 x 10^-5 / 5.7 x 10^-17

The diameter you want is 614 million kilometers, or more than 4 times the distance between Earth and the Sun. Good luck building that.

Take a look at the ATLAST concept for a telescope that could be launched in the 2025-2030 timeframe. It comes closest to what you are looking for.

Adaptive optics works great at infrared wavelengths. It does not (yet) do well at visible wavelengths. Even the 2.5 meter mirror of Hubble has better resolution at visible wavelengths than the 10 meter Keck mirror due to atmospheric blurring. Further, adaptive optics is only effective over small fields of view (such as a single star and planet). One can not take a wide field view of a nebula or a galaxy and get a high resolution adaptive optics view over the whole field.