PDA

View Full Version : Stereoscopic/Anaglyph Rendering?


jalingo
08-15-2003, 11:38 AM
Does anyone know if Lightwave can render an anaglyph animation - so one can view with 3D glasses (the blue and red tinted kind)?

Or are there other options for generating a 3D glasses (polarized or anaglyph) viewable animation?

I have a client who wants to show a 3D animation using glasses ( so would that be a 3D^2 animation) in an exhibit booth at a convention.

prospector
08-15-2003, 02:50 PM
yes

Zafar Iqbal
08-15-2003, 07:11 PM
http://www.google.com/search?hl=en&ie=UTF-8&oe=UTF-8&q=anaglyph+lightwave+3d

I did some anaglyphs in LW some years ago.. damn it was funny, sitting infront of the computer, wearing those glasses and everybody looking at you like you were some geek ;)

Adrian@Stufish
08-16-2003, 06:43 PM
rendering stereo is easy, just tick the right buttons in camera settings.
viewing is stereo is a pain in the butt.
Anyglyph is 3d monochrome with colour fringes,
Stills are easy - just buy a viewer that goes with one of those simple 3d cameras and lets you print 2 pics side by side. the viewer uses prisms so that the eyes see the images in the right places.
But stereo animation showing is a different thing. you can buy varios proprietry goggles which work with particular video cards, and can probably be made to show stero mpgs. But it is very pricey and there are a lot of dud goods out there.
So first find the display platform - a goggles/glasses/nonitor/card comb that you hav tried and thinks is good enough.
Then work out how to rener the format for it - that'll be the easy bit.

side by side images on a monitor with sheets of polarising material taped to the monitor ONLY wrks if you're Clarence the Cross-eyed Lion and are standing in exactly the right place !

I am sure there is a system out there that displays alternate images on the monitor and uses synchronised shuttered glasses to view it. But my bet is that the IMAX people bought the patents and have stuck them where the sun don't shine.

dwburman
08-17-2003, 11:14 PM
There was an article on stereoscopic rendering in a LW magazine a couple of years back. I think it suggested to add a null for the camera to target. That sets the focal point for the 3D.

In the camera properties there is check box for stereoscopic rendering. When that is checked LW will render out Left and Right images for each frame.

LW comes with an image filter that combines the two images into an anaglyph. I've found that you have to do an F10 render and open the saved frame in order to see the anaglyph rendered.

If you don't use the image filter you'll have to combine the left and right images in some compositing app. The filter ruins any chance for further compositing because the right frame has already been altered.

Also, some things like lensflares might not work properly because they are overlayed onto the finished render and don't exist in 3D space.

Adrian@Stufish
08-18-2003, 05:20 AM
The null targeting is really usefull when you are going in close up - where the distance to object is within about 10x the eye separation, when it helps provide an accurate impression of object scale, but if you use it without some degree of DOF to blur the background it quite succesfully provides some degree of migrane!

Zafar Iqbal
08-18-2003, 02:20 PM
I really wish i could remembers some of the details when i worked on something like this.

I do remember that i could manipulate the renderings in After Effects/Photoshop/whatever, and by doings so in a certain way (something about how close the to pictures (red color channel actually) was to eachother, and even by how much it overlapped, i could control the overall depth feeling.

I could have a rendering that when viewed with anaglyph glasses, looked like true 3D inside the monitor, as if the monitor itself was a window - i could then pull the depth (wich affects the objects) out of the monitor - i tried posing my fingers infront of my, like i was holding the object, and i pretty much felt like it (i mean, by the look of it).. pretty facinating stuff.

jalingo
08-18-2003, 02:24 PM
Thanks everyone for your responses. :)

I think it looks like it might be tricky but not impossible to do it in lightwave. Shutter driven glasses are not an option since my client wants to fill a small room with people and have them view the animation.

Anyway, the info you have given me is at least a start. I did find this - although it's LW 5.6 : http://www.aifx.com/3d_lw_a.html

dwburman
08-18-2003, 03:16 PM
ahhh, good advice. I don't think the tools have changed from 5.6.

BTW I saw spy kids 3D a couple of weeks ago. My eyes did get pretty "tired" watching it. I suspect that a lot of it was due to it being in color. certain colors seem to vibrate when your eyes try to figure out what they are through different color lenses.

JamesCurtis
08-21-2003, 10:21 PM
Just wanted to let everyone reading this thread know that I played around with a method of creating 'shutter glasses' work in Lightwave a while ago. I'm planning on getting in touch with one of the LW mags out there to do an in depth article and make a sample scene available. I used the shutter glasses from a DVD 3D set. The set was 80 bucks and had three 45 minute movies.

Adrian@Stufish
08-22-2003, 06:43 AM
VREX are advertising shutter glasses at $35 which don't look too bad.
Their operation appears to be software-free and frame-rate -independant - they connect between the PC and the CRT with a 9pin through 'dongle'ish thing which apparently blanks alternate lines from the display while switching the L & R lenses. there is a 'hand control' which switches the system on and off and can 'reverse' the lense switching if the stereo is inverted.
you can piggy-back 6 pairs !
They advertise a MAX plugin.

Does anyone have a bright idea for interleaving LW's stereo output onto alternate scan lines?
Simplest would be a 'post' combine of image pairs.
Best would be a render plugin - like the anyaglyph combine.- Actually, thinking about it, one would then need to render only alternate lines for each of the stereo pair images, so rendering should be not much slower.
(would everything go horribly wrong if the viewing software resized the image ?)

Could be what we've all been waiting for.

surferjoe
08-22-2003, 01:23 PM
Originally posted by jalingo
Thanks everyone for your responses. :)

I think it looks like it might be tricky but not impossible to do it in lightwave. Shutter driven glasses are not an option since my client wants to fill a small room with people and have them view the animation.

http://www.aifx.com/3d_lw_a.html

Look into sofTreats program:
3D Stereo Image Factory (tm) PLUS
http://members.aol.com/threedr/

It does every flavor of anaglyphs and does real time adjustments and batch processing.

It has served me well.

Adrian@Stufish
08-26-2003, 07:11 AM
No-one has come up with a handy script but,

Hold everything,

The bath just flooded,


I've got it !

If I create a still image called, say 'BlankingMask' in something like Photoshop, exact frame size - e.g 720x576 for PAL, consisting of alternate rows of black and white pixels, then all I have to do is render my L&R image sequences using LWs stereo camera settings, and then in an empty scene, in Image Editor load the 2 sequences and the blanking still image. Then in 'scene' /'backdrop'/'compositing' I put the L image as Background Image, the R image as foreground image, and the BlankingMask as the foreground Alpha.
As long as I render this scene with the camera resolution the same as the BlankingMask image, we should have Line Blanked Stereo.

Whee!

And while I'm about it, I think I can render the L&R image sequence at half the image height using a tall pixel
i.e. for PAL 720x288, pixel aspect ratio 0.5334.
LW will re-size them neatly for the combining run, and I won't have wasted render time on the initial run.

And of course, if I want to make lower resolution movie files from my original rendered L&R image sequences, I just have to make a different sized BlankingMask image. to suit the camera res.

I'm ordering the glasses now.


Lets Roll !

Arnie Cachelin
08-26-2003, 10:30 AM
I did some field-type stereo stuff in LW on the Amiga, which had cheap and easy field stereo as a result of its video display. This was before LW had built-in stereo. The trick is to use the natural interlacing one gets from field rendering, and have LW compose the frames for you. All you need is a camera motion that oscillates between 2 extreme eye positions. I think I made a three-frame loop with the cam. at the left position on frame 0, and at the right frame position on frame 2. Then I rendered only the odd frames (1,3,5,...) with field rendering on. I think the eye positions should be double the eye sep. you want, since the frames come from positions halfway from the keyed position and the middle.
Parenting the whole thing to a null, and targeting another is quite handy in this case.

Adrian@Stufish
08-26-2003, 10:47 AM
I think eye see what you mean,
Now using non wholenumber keyframes one could set the camera oscilating about it's path so that on the frame number it was at the left position and on the half rame it was at the right position, then as long as you didn't want motion blur it might work.
I'll give that a try too, while I wait for the glasses to arrive.
Assuming they ever get ordered that is - the VREX website is USA orders only.

Adrian@Stufish
10-17-2003, 05:57 AM
Long time later:
I have the glasses, and yes they do work - with reservations.

It only works well with a good monitor and a medium range graphics card. - a cheap or old monitor seems to have too much 'fade lag' on the pixels, so you get ghosting of one eye or the other. and a classy video card (e.g. my perahelion) seems to try too hard with the same result.

Because alternate lines are L & R the amount of information in each 'image' is only half that of the render resolution. i.e. a low quality picture. - a video signal just does not carry enough information for an acceptable stereo experience.
I am getting reasonable results with 2xVGA (1200x960) and using the newest DivX codec (free download from DivX - install the codec and LW seems to find it)
I render standard LW stereo L&R image pairs with the camera set at 1200x480 and 0.5 pixel aspect (this halves the rendertime and gives the same end quality, because the combined image only uses alterate lines, geddit?). I have a 1280x960 'mask image' of alternate Black & white lines. Then in an empty scene I load the 2 sequences and the mask in Image Editor, in Scene/compositing I load the L image as background, the R image as foreground and the mask as foreground alpha, and tick foreground fader (no I don't kow why but it works). Then I set the camera to 1200x960 (pixel aspect 1) & just render to avi with the DivX codec selected and it's default settings.

Anyone else experementing with this?

P.S. if you do try using video resulution, you cannot view the stereo using VT because of the way it handles the display.
The whole thing relies on the video display keepng the fielding intact.

P.P.S. the method I'm using allows you to use motion blur and DOF stuff which fancy camera wobbling prohibits.

OnlineRender
06-20-2010, 11:02 AM
MEGA :bump: ....an old old post , but does anybody know the workflow with stereorender and after effects ..........? Please .