3D stereoscopic convergence distance test anaglyph glasses

Apefos Adapter

Apefos Adapter

Well-known member
maybe dvxuser needs a specific thread for stereoscopic...

I did this test: http://www.youtube.com/watch?v=ynfI1ORdA5k

setup: two D8 camcorders TRV120, two camera distances, 11cm (4,4 inch) and 28cm (11 inch), two convergences, sometimes in foreground, sometimes in background, using same wireless remote control to start/stop and zoom both camcorders. using thumb clicking to sync, got half frame outsync sometimes due to different starting in camcorders mechanism, but not a problem for me because the thumb clicking sound and movement keeps in same frame, shot 30p over 60i using slow shutter feature, iris was auto due to slow shutter, wb was auto due to camcoredrs does not have manual wb. exported as mpeg2 at 30p.

Conclusions:
1) camcorders must be as near as possible from each other, same human eyes distance if possible to help the brain to get the 3d image. 11 cm, 4,4 inch works great.
2) convergence must be in the point of image you want people to look at to help the brain to get the effect.
3) shallow dof can help keeping convergence point in focus and the other points slightly out of focus forcing people to look at convergence point.
4) objects to much close to camcorders makes the effect more dificult to the brain (anaglyph conclusion, dont know if polarized or shutter glasses is like that)
5) anaglyph is not so good, polarized glasses and shutter glasses are the way to go (everybody knows that).
6) sync the camcorders is easy and works great even with half frame difference in the footage because the sound and movement from the thumb clicking keeps in the same frame even with the half frame difference in time.

trying to test 3D titles also.

this is edited version. there is raw footage from camcorders just synced in post also.
 
I've been playing with some 3D tests too. I found that 2.5 inches is even better if you can get it. The best way is to use two mirrors placed at a 90 degree angle, with the angle itself "pointed" towards whatever you're filming. Only issue with this set up is that convergence can be tricky, but it makes for a good "eye distance".
 
I've heard that shallow DOF is actually quite distracting...as you end up with these giant out of focus blobs sometimes if you have things too close in the foreground. Perhaps using lighting to your advantage would be a better way to go.
 
RyanT said:
I've heard that shallow DOF is actually quite distracting...as you end up with these giant out of focus blobs sometimes if you have things too close in the foreground. Perhaps using lighting to your advantage would be a better way to go.
Click to expand...

I concur with that statement as well.

Also, the interocular distance should be around that of people but may easily be exaggerated in either direction (further apart or closer together) to achieve artistic results. Thus it does not have to be exactly as people's eyes are, although that is pretty much the starting point. This is not a matter of feet or such but simply small amounts. Probably why the other poster liked the different interocular distance... ya know? I'm guessing, without going back to look, it was greater then what was originally proposed that therefore had greater depth.
 
dof

dof

I was talking about a very little out of focus, not to shallow, but yes, the lighting and the action too can help keeping the eye in convergence point
 
camera distance

camera distance

the D8 trv camcorders i used are not so small, so the smallest distance I could get was 4,4 inch and it was good for objects at 1,5 meters or more.

the other distance i used, 11 inch, was not so good for foreground objects at 1,5 or 2 meters from camcorder, but maybe it can be good for objects far far away.
 
CRT x LCD

CRT x LCD

The anaglyph video I did works very great in the 17" CRT analog monitor, but in the old sharp xvh37 3LCD projector the image suffers from ghosting, the green from the cian eye goes inside the red eye and ruins the effect.
 
anaglyph tips

anaglyph tips

some tips i got with my tests:

the anaglyph can be good if

1 the foreground objects are 2 meters away or more
2 the camcorders distances are as small as possible, limited only by the camcorders body
3 the convergence is very important for a good anaglyph, so the principal object or person must be in the convergence point.
4 I found this best calibration for red/cyan glasses in editing:
red eye: red=80 / green=0 / blue=0
cyan eye: red=0 / green=80 / blue=100
I found this using a 17" analog CRT monitor and the imersion was great, i do not know if this works in lcd screens, the 3 lcd projectos suffers from ghosting with anaglyph.

anaglyp is a good cheap way to do editing, a realtime preview in the timeline using two layers with the settings in item 4
 
results in different screens

results in different screens

CRT analog computer monitor 17": great result, no ghosting, excelent imersion.

LCD television 42": ghosting in the image, bad result

3LCD projector: ghosting in the image, very poor result

CRT analog old fashion television: ghosting in the image, bad result

I bought two 3D DVDS, Shrek and Sharkboy/Lavagirl, come with glasses, waiting arrive to compare with the anaglyph i did
 
Well, anaglyph is horrible. why not try polarized or field sequencing.
If you shoot your own movies, it will be a lot better .
 
better results

better results

I got better results in the LCD screens doing a better vertical alignment in the red/cyan images and doing a better horizontal alignment to get perfect convergence in the principal object. ghost was very reduced, but still present. i got a better imersion, good to look at.

this learning in anaglyph will help a lot if I work with polarized or shutter glasses in the future. anaglyph has a great learning potential due to be very dificult to do. anaglyph just give good results if everything is perfect done, so lessons from anaglyph experiments are very usefull.
 
anaglyph is just a way to look at video, the rules to shoot are the same whatever method you use to display.
IMHO , anaglyph is just a disappointing way to look at sterescopic content.

if you want to shoot nicely 3D, use a video mixer to superimpose the picture from both camera. It is very helpful for checking convergence, scale, keystone etc...
I think there would be a real market for a little box with 2 unsychronized composite input and a mixed out . cheap if possible.
something very close
http://www.colorado-video.com/mixer.html
I am pretty sure these guys can build that in a small compact format for cheap.
(ideally something like pack of cigarettes)
they have a killer product here , and probably they don't even know.

With modern chip, it is easy to build the schematics, but small SMD IC require very high precision printed circuit and are the hell to solder, almost impossible at DIY level.

I use 2 sanyo FH1 at 1080p60 to shoot 3D, firmly mounted on a DiY frame. There is nothing to adjust for perfect 3D , except convergence when subject is very close (distance betweence center of lenses is 7cm , about 2-3.4" ). Such device would be very helpful.
i use one remote control for start/stop/zoom and camera are perfectly synchronized.
Unfortunately this camera is working poorly in low light but picture is great when light is ok.
less than 800$ for a pair of FH1, you cannot beat that.
the great stuff would be to hack one of these 3D photo frame from samsung to get 3D display in the field, without glass. thechnically it is easy to obtain the proper signal format, but i do not know if the
screen can accept video from external input.
 
Last edited:
I think a big issue too we'll need to figure out is how to edit 3D, since we're dealing with two sources. I currently edit the left eye in Vegas, render it, then have Vegas replace the clips with the right eye equivalent. Render it. Now I have two versions of the flick that I can combine via anaglyph, polarized, or whatever.

How do you guys do it?
 
mixer and editing

mixer and editing

thats a great idea to use a mixer to preview both images togheter in field.

i am thinking about how to edit, i think a editing software to do the cuts for left right eye and render and after effects to do vertical, horizontal, rotation, keystone, zoom or crop, color correction.
 
editing and delivering

editing and delivering

El Director said:
I think a big issue too we'll need to figure out is how to edit 3D, since we're dealing with two sources. I currently edit the left eye in Vegas, render it, then have Vegas replace the clips with the right eye equivalent. Render it. Now I have two versions of the flick that I can combine via anaglyph, polarized, or whatever.

How do you guys do it?
Click to expand...

My hopes are in a 3d blu-ray drive recorder for PC and in a software for encoding the stereo video pair in a 3D AVCHD-MVC file to record in the 3d blu-ray disc

first edit each video pair in premiere or vegas, after this go to after effects.

there are some plugins for after effects for 3d compositing. even without plugins you can composite the stereo pair in AE using color channels settings for a anaglyph preview.

left eye video (layer 1):
R=100
G=0
B=0
opacity=50

right eye video (layer 2):
R=0
G=80
B=100
opacity=100

use a red/cyan glasses for preview

this is just for preview the composition and do the x/y/z axis corrections, after done, remove this color settings to export each eye video. after this, with each eye video done I do not know how to encode it for a 3d blu ray, but this solution must come soon.
 
i use same method with Premiere CS3, (clip swapping).
I think if you really need to edit in 3D with prewiew (does not look really important to me, 2D is ok to edit a story), you can just get all of your clips trimmed to start at same frame and get same frame count. When they are all ready, you should be able to create a field sequenced version of it with a script in avisynth.
Then you could use Nvidia glasses.
Or you can also create a line sequenced version to use with polarized glass and Zalman monitor.
The last solution has the advantage of having 3D possible even when scrubbing or video paused and is probably less expensive.
You can even do that in the NLE, just create a picture with all even (or odd) lines as colored lines, then use the picture as a mask oved the left picture, so colored lines will show the right picture.
The best would be to have a plugin doing that automatically, it must be very easy to write.
for instance with my setup i have no need for vertical, horizontal, rotation, keystone, zoom or crop, color correction, and got stunning result right out of the SD card. It is just painfull to convert AVCHD to a workable codec (in my case cineform) then check all clip to make sure they are in sync and same length.
 
Last edited:
learning too much

learning too much

Oh my God, I am learning too much these days reading all the forums, websites and so on...

I found another important points: the toe-in / keystone... , file formats, dysplay methods, window violation...

And after tests I realised the images for anaglyph must be very clean without grain or compression artifacts to minimize ghost

anaglyph is the only way i have to see 3d now and it is helping a lot to studying 3d.

premiere motion and basic 3d effects can do all the corrections needed with great quality for progressive video.
 
You must log in or register to reply here.
Back
Top