How is it when Peach Open Movie will be created in two versions? One a normal one and the other a anaglyph movie
more infos on http://en.wikipedia.org/wiki/Anaglyph_image!
To see the 3d movie in 3d and not in 2d. Maybe someone can devolop a new camera, which will takes pictures from 2 horizontal lying positions(7cm) and save them as a movie or picture.

All you really need to do for this is parent two extra camera's to your
original camera and place them correctly.

Then just render your animation 3 times.
If your cheap you could probably just parent one additional one.
So The "Normal" video is just the left eye (or right).

To tweak things you may also want the two additional cameras to track to
an empty so you could change the focus point. I don't think adding an
additional camera type will simplify things though.

I do think though it would be cool to have Peach done in 3D.
I really liked seeing meet the robinsons in 3D,
and now that the technology is out there why not take advantage of it.

http://en.wikibooks.org/wiki/Blender_3D:_Noob_to_Pro/Add_some_depth_with_stereo

Rendering stereo isn't that simple in today's pipelines... a lot of stuff is being composited, and you have to make sure this doesn't give a flat appearance in stereo. (Yep, just saw beowulf in imax 3d!).

For an open movie project in stereo I'd like to get a sponsor who helps us financing the extra work involved for it...

Ton,

it would be worthwhile to render out Rabbits Revenge and or ED in stereo (assuming our rendering time is 'free'), and then see which spots need work to improve the look for a stereo project.

LetterRip

SirDude wrote:

All you really need to do for this is parent two extra camera's to your original camera and place them correctly. Then just render your animation 3 times. If your cheap you could probably just parent one additional one. So The "Normal" video is just the left eye (or right)...

Completely agree SirDude, except that I can't see a reason to render the animation 3 times. You only need two. Either the left or right will work fine as a flat. I use a 3 camera stereo camera setup. The center camera is used to control the left/right cameras but I only use it to render when making quick low res single image test renders.

kidb wrote:

http://en.wikibooks.org/wiki/Blender_3D:_Noob_to_Pro/Add_some_depth_with_stereo

The authors of the Wiki are right on in discussion of the manual creation of a stereo camera and the problems of adjusting convergence (aiming the cameras at an object rather than keeping them parallel). Adjusting the convergence can make the background impossible to view so it should be used with caution and limited to scenes with short depth (close background like a wall).
Convergence will also set the "stereo window." This window is the zero depth location that coincides with the plane of the screen. As the Wiki suggests, to avoid distortion (often called keystone distortion) the cameras should be kept parallel.
However parallel cameras set the stereo window at infinity so everything is in front of the screen. That often is a problem because object get cut off by the edge of the screen (window violation) or are too close to the viewer to be seen comfortably. To adjust the stereo window you can render your images wider than the desired aspect ratio, shift the centers and trim the excess. Ideally Blender would have a full featured stereo camera for this purpose. Hash Animation Master has such a built-in stereo camera and can render in anaglyph, cross-eye, parallel, and interlaced stereo formats. I think someday Blender will have these features. If I were rich I'd pay for their development

In the meantime, I think it is best to use a manual stereo camera rig based on 3 cameras. Render the left and right images about 25% wider than the desired aspect ratio, then use these free programs to adjust the stereo window and convert to any desired stereoscopic format:

http://stereo.jpn.org/eng/

The programs are free but I don't think they are open source. Use StereoMovie Maker for animations and StereoPhoto Maker for stills.

Sirdude,

rendering the image twice is a bad idea since it is wasteful of processing time - instead both cameras should happen at the same time - then the image only takes 10% longer to render compared to a single camera.

Actually we should set it up so as many cameras can render at the same time as desired - then you could render your 3D and flat movie at the same time for only a 20% increase in render time.

LetterRip

LetterRip wrote:

Sirdude, rendering the image twice is a bad idea since it is wasteful of processing time - instead both cameras should happen at the same time - then the image only takes 10% longer to render compared to a single camera. LetterRip

I doubt that a full stereo pair can be rendered at only a 10% increase in render time over a single flat image. True, a stereoscopic camera could be more efficient than two separate renders, but can't imagine it would be that fast. Is that estimate based on an existing stereo rendering system? Maybe analgyph could be that efficient since it renders the same number of pixels as a flat image, but not full color pairs that you would need to produce quality 3D animation as seen in Beowulf or Meet the Robinsons. My 2 cents.

LetterRip wrote:

Sirdude, rendering the image twice is a bad idea since it is wasteful of processing time - instead both cameras should happen at the same time - then the image only takes 10% longer to render compared to a single camera. Actually we should set it up so as many cameras can render at the same time as desired - then you could render your 3D and flat movie at the same time for only a 20% increase in render time. LetterRip

Uh, where are you pulling these 10% and 20% figures from? Sounds like thin air to me...

The other thing is that stereo rendering could also limit what sort of work you can do in comp. I'd presume you'd pretty much have to keep it all procedural, or things like manual painted masks could complicate things somewhat, maybe even also other pixel based things like blurs popping at different times due to the slightly different images, etc. etc.

The 10% number was from a recent post at CGTalk from a company that was rendering their movie in 3D, and by switching to a renderer that did both cameras at the same time it was only 10% longer per frame as opposed to a single camera render.

The 20% is a crude extrapolation based on the above

I finally found statement I remembered, but it doesn't make any reference to percentage (just a 'substantial reduction in render time from rendering both at the same time)

http://forums.cgsociety.org/showthread.php?f=59&t=572530

So perhaps the 10% is a bogus memory...

LetterRip

Sirdude,

Quote:

This paper presents a GPU based implementation of an imagebased rendering method for reducing the cost of stereoscopic rendering to the cost of rendering a single monoscopic image plus a smaller fixed cost. Our approach is to use the color and depth information from the rendered image of one eye to produce a reconstructed depthsprite which is rendered for the other eye. A GPU hardware accelerated technique for producing and rendering this depthsprite at rates above 60 Hz is presented. Our technique enables the real time stereoscopic display of complex and data intensive objects, which are currently constrained to monoscopic rendering technology.

http://ieeexplore.ieee.org/Xplore/login.jsp?url=/iel5/11055/34910/01667668.pdf

LetterRip

Why don't Blender just make a stereoscopic camera? (Just wondering)

Gollarks wrote:

Why don't Blender just make a stereoscopic camera? (Just wondering) :?

Answer: It is not so easy to optimize a 3D camera. The average size between 2 eyes is about 6.4 cm although it can vary quite a bit between individuals. Also the size of videos varies from You-tube small window to large cinema screen. Moreover the distance between the eyes and the screen can vary between a few centimeters to several meters. If you want to make video objects "come out of screen", you have to be careful about which audience (cinema theater or you-tube or other) is your video directed to, so that you can optimize cameras postilions and angles and object distance to make your audience less tired. In conclusion, in any case a multipurpose camera cannot be made.
In the link below you can see some results we got after optimization of camera positioning in Blender. If you need help about camera positioning, send us a mail, or we can consider making a tutorial.

The Pitiless Senses Hunters


http://www.youtube.com/watch?v=7qdLz7PQfWs