$100 DIY Beamsplitter Rig at DVinfo.net

Jesse Blanchard · May 9th, 2011, 05:33 PM

I'm working in the DIY 3D world. Here's a quick overview of my latest build. Works pretty well for the price
It's only got about 6 pixels of vertical discrepancy between the two cameras.
Here's phase 1

and phase 2

Frank Stearns · May 10th, 2011, 07:07 AM

Pretty cool. I enjoyed watching the video.
I started making a beam splitter rig myself but unlike you, I was a little too ambitious at the time and tried to include way too many bells and whistles. such as servos to adjust the convergence,IO,remote video output in stereo, etc.
So, back to the drawing board/Cad system it went and still sits.

Jesse Blanchard · May 10th, 2011, 09:43 AM

I certainly understand that temptation. This is my third design. And for this I tried to simplyfy it as much as possible. I found the more stuff I added the more the images were off. Next, I'm gonna try a follow focus type IO control. I figure it will be better to have something that I can manipulate by hand as opposed to trying to achieve the right feel electronically.

Pavel Houda · May 10th, 2011, 11:44 AM

Wow, the alignment is impressive. With a sensor of say, 15mm vertical, your alignment has to be within .083 mm to get 6pixel misalignment worst case, between both of the cameras, tolerances of the mirror mount, etc. - that is amazing, given that plywood will ever so slightly warp with temperature and humidity, and all the tolerances involved. This is way better than what I achieved continuously using cameras with total adjustment, on SbS rig. Even the sensors and optics are generally not that well aligned on 2D cameras, they don't need to be. What is your secret of aligning the two sensors? Do you have any solutions for the genlock? So far the best I've heard of for Canon DSLRs was to take a still shot just before movie shot, on the Canon's, but that only lasts for a short while. How do you know you are in sync? Can you monitor the camera alignment while you are filming? How do you do that? I am very interested in mirror rig, but the best one I've seen so far that is reasonably priced is several orders of magnitude more expensive than yours. Congratulations. This must be the best price/performance ratio for mirror rig I ever heard of.

Jesse Blanchard · May 10th, 2011, 10:19 PM

It took me a lot of fudging to get the alignment that good. If you look closely at the side of the rig, it's riddled with holes.

That said, I think part of the strength of the design is that I tried to really make it simple so that there would be as little error as possible. (and I must say that the alignment isn't always that good. It does goes off by 9-12 pixels or so pretty often).

For syncing, I just use a clapper to set a sound sync and the remotes to trigger both cameras. I have seen someone use the photo trigger and I've thought about trying to do the same here. But so far, it hasn't been that big of an issue. The sync is off by, at most, 1/60 of a second which is pretty minimal. I know people will argue otherwise but I did some speedbag 3D and it worked pretty well.

I do have a 3D monitor in very rough shape that I can actively monitor through. I'm still tweaking the design but you'll be able to do live HIT convergence when it's done and it will cost less than $500 for all the parts. It's ugly but works alright. (kind of like my car!)
I'll do a video once I have a system worth showing.

Tim Dashwood · May 11th, 2011, 02:08 AM

Pavel and Jesse,

For syncing on the Canon DSLRs it is important that the shutters are perfectly in sync for motion or you will always have a tiny bit of temporal disparity. I have also heard of the trick Pavel mentioed for the Canon cameras that may or may not work.

Step 1: Start the video recording using the IR remote as you normally would. The electronic shutter cycle will be based on the internal timing established at power on.

Step 2: Use a hard wired remote shutter release to trigger both cameras simultaneously to take a picture at the same moment in time.

That's it. The video after the still photo is taken should have perfect shutter sync and hold well enough for at least the video time limit on the camera.

I think Chris Hurd has a photo or two of my $40 DIY beamsplitter for DSLRs?

Jesse Blanchard · May 11th, 2011, 08:40 AM

Cool! You can do a hard-wired remote!? I'd only seen this work around with wireless remotes which was too expensive. I'm definetely going to try the hard remotes now. Do you know if you can use a radio shack type splitter for the cables or do you have to wire them? Is there anything more elegant out there? I'm going to try a set-up for this that runs both cameras off the same battery. I'm curious is there's a sync solution there as well. We'll see. Thanks for the tip.

Tim Dashwood · May 11th, 2011, 09:39 AM

You could do it the "safe" way and use a Pocketwizard on each camera and one to trigger (3 x $170) or go with the

cheaper version

(2x$40)

Or you could make your own double N3 wire and DPDT switch that connects the top pin with the lower left pin (N3 has three pins in a triangle.) You would have to buy a couple

remotes

and cut the cables off in order to wire them the way you want... or maybe just splice one cable into the other so you can still use the nifty remote handle controller.

This will work for 7D and 5DmkII but not T2i.

Pavel Houda · May 11th, 2011, 12:41 PM

I saw the article on this issue here: Genlock Canon 5D's & Shoot 3D ? .

BTW, on slightly different topic, nother interesting approach for Panasonic camera genlock is described here:

"Hi, All

Recently I completed making of 3D video rigs based on Panasonic GH2, a DSL(R) HD
video camera with 60fps sensor output and 60i AVCHD recording.

3D sync is almost perfect after post editing in a Premiere CS5 60P sequence as
far as judging from audio waveforms of R/L clips, more quantitatively
approximately 0.4 mS measured by means of external flash. My particular pair of
GH2s with adjacent serial numbers shows relatively small drift of 0.05 mS/min.
Therefore 12 min of 3D clip can be shot in sync less than 1 mS.

3D photos of various kinds of rigs and flash sync test result is shown here;
Stereo Flash Viewer

The syncing method is very simple with two additional components, an external
battery (9V Sanyo rechargeable battery for musical instruments use) for just
power-on of both cameras simultaneously to start internal clocks which governs
REC start sequence and a spliced remote shutter (for 3D photo shooting) to
trigger video recording. Because of the unidentified procedures in the camera
before starting REC, from exactly zero to several frames of 1/60 sec difference
exists between both cameras. however it can be easily fixed in the post edition.
Half-push to complete the Auto-Focus before full-push for REC ON and setting
shutter speed to faster than 1/60 sec is the necessary condition for the sync.
No sync failure have been observed so far. A drawback of this method is some
clue e.g. clapper sound etc. is required for the post sync, however because of
the discrete nature of frame shift of a unit of 1/60 sec it is not difficult to
sync R/L without any clue by the anaglyph monitoring of the 3D scene.

Here is a test 3D video for sync and also for EX Telecon function ( x2.6 zooming
with keeping Full HD resolution of 1920x1080, one of my favorite functions of
GH2) used for shooting 3D telemacro video up to 1560mm equivalent.
YouTube - 3D Cherry Blossoms in Spring Gust 2 (Telemacro HD Video with twin GH2)

inoue_k3D"

Neil Richards · May 11th, 2011, 04:19 PM

Thanks for that Pavel, that's really useful.

I have the GH2 and have wanted to do some stereo with it so I can try this to get the sync sorted out. I have two SD700's as well so I'll see if it will work for those too.

Karl Kozak · May 11th, 2011, 07:30 PM

Rig looks cool... nice job. The thing to be wary of is the glass. You need a high-grade beam-splitter glass to get quality 3D.

Best of luck !

Jesse Blanchard · May 11th, 2011, 07:49 PM

Yes, I know you guys at 3D Film Factory are always going on about the glass. But I haven't noticed anything in the final images. Where do you guys source your glass? Do you have some specs on why it better? Do you have certain scenarios where the glass doesn't work? I'd love to know more. Here's some macro footage I shot without issue.

May 9th, 2011, 05:33 PM	#1
Jesse Blanchard Regular Crew Join Date: Apr 2011 Location: Portland, Oregon Posts: 97	$100 DIY Beamsplitter Rig I'm working in the DIY 3D world. Here's a quick overview of my latest build. Works pretty well for the price It's only got about 6 pixels of vertical discrepancy between the two cameras. Here's phase 1 $100 3D Beamsplitter Rig on Vimeo and phase 2 $100 3D Beamsplitter Rig Phase 2 on Vimeo

May 10th, 2011, 07:07 AM	#2
Frank Stearns Regular Crew Join Date: Sep 2010 Location: Superior Wisconsin Posts: 62	Re: $100 DIY Beamsplitter Rig Pretty cool. I enjoyed watching the video. I started making a beam splitter rig myself but unlike you, I was a little too ambitious at the time and tried to include way too many bells and whistles. such as servos to adjust the convergence,IO,remote video output in stereo, etc. So, back to the drawing board/Cad system it went and still sits.

May 10th, 2011, 09:43 AM	#3
Jesse Blanchard Regular Crew Join Date: Apr 2011 Location: Portland, Oregon Posts: 97	Re: $100 DIY Beamsplitter Rig I certainly understand that temptation. This is my third design. And for this I tried to simplyfy it as much as possible. I found the more stuff I added the more the images were off. Next, I'm gonna try a follow focus type IO control. I figure it will be better to have something that I can manipulate by hand as opposed to trying to achieve the right feel electronically.

May 10th, 2011, 11:44 AM	#4
Pavel Houda Major Player Join Date: Mar 2009 Location: Laguna Hills, CA Posts: 352	Re: $100 DIY Beamsplitter Rig Wow, the alignment is impressive. With a sensor of say, 15mm vertical, your alignment has to be within .083 mm to get 6pixel misalignment worst case, between both of the cameras, tolerances of the mirror mount, etc. - that is amazing, given that plywood will ever so slightly warp with temperature and humidity, and all the tolerances involved. This is way better than what I achieved continuously using cameras with total adjustment, on SbS rig. Even the sensors and optics are generally not that well aligned on 2D cameras, they don't need to be. What is your secret of aligning the two sensors? Do you have any solutions for the genlock? So far the best I've heard of for Canon DSLRs was to take a still shot just before movie shot, on the Canon's, but that only lasts for a short while. How do you know you are in sync? Can you monitor the camera alignment while you are filming? How do you do that? I am very interested in mirror rig, but the best one I've seen so far that is reasonably priced is several orders of magnitude more expensive than yours. Congratulations. This must be the best price/performance ratio for mirror rig I ever heard of.

May 10th, 2011, 10:19 PM	#5
Jesse Blanchard Regular Crew Join Date: Apr 2011 Location: Portland, Oregon Posts: 97	Re: $100 DIY Beamsplitter Rig It took me a lot of fudging to get the alignment that good. If you look closely at the side of the rig, it's riddled with holes. That said, I think part of the strength of the design is that I tried to really make it simple so that there would be as little error as possible. (and I must say that the alignment isn't always that good. It does goes off by 9-12 pixels or so pretty often). For syncing, I just use a clapper to set a sound sync and the remotes to trigger both cameras. I have seen someone use the photo trigger and I've thought about trying to do the same here. But so far, it hasn't been that big of an issue. The sync is off by, at most, 1/60 of a second which is pretty minimal. I know people will argue otherwise but I did some speedbag 3D and it worked pretty well. I do have a 3D monitor in very rough shape that I can actively monitor through. I'm still tweaking the design but you'll be able to do live HIT convergence when it's done and it will cost less than $500 for all the parts. It's ugly but works alright. (kind of like my car!) I'll do a video once I have a system worth showing.

May 11th, 2011, 02:08 AM	#6
Tim Dashwood Wrangler Join Date: Aug 2005 Location: Toronto, ON, Canada Posts: 3,637	Re: $100 DIY Beamsplitter Rig Pavel and Jesse, For syncing on the Canon DSLRs it is important that the shutters are perfectly in sync for motion or you will always have a tiny bit of temporal disparity. I have also heard of the trick Pavel mentioed for the Canon cameras that may or may not work. Step 1: Start the video recording using the IR remote as you normally would. The electronic shutter cycle will be based on the internal timing established at power on. Step 2: Use a hard wired remote shutter release to trigger both cameras simultaneously to take a picture at the same moment in time. That's it. The video after the still photo is taken should have perfect shutter sync and hold well enough for at least the video time limit on the camera. I think Chris Hurd has a photo or two of my $40 DIY beamsplitter for DSLRs? __________________ Tim Dashwood

May 11th, 2011, 08:40 AM	#7
Jesse Blanchard Regular Crew Join Date: Apr 2011 Location: Portland, Oregon Posts: 97	Re: $100 DIY Beamsplitter Rig Cool! You can do a hard-wired remote!? I'd only seen this work around with wireless remotes which was too expensive. I'm definetely going to try the hard remotes now. Do you know if you can use a radio shack type splitter for the cables or do you have to wire them? Is there anything more elegant out there? I'm going to try a set-up for this that runs both cameras off the same battery. I'm curious is there's a sync solution there as well. We'll see. Thanks for the tip.

May 11th, 2011, 09:39 AM	#8
Tim Dashwood Wrangler Join Date: Aug 2005 Location: Toronto, ON, Canada Posts: 3,637	Re: $100 DIY Beamsplitter Rig You could do it the "safe" way and use a Pocketwizard on each camera and one to trigger (3 x $170) or go with the cheaper version cheaper version (2x$40) Or you could make your own double N3 wire and DPDT switch that connects the top pin with the lower left pin (N3 has three pins in a triangle.) You would have to buy a couple remotes remotes and cut the cables off in order to wire them the way you want... or maybe just splice one cable into the other so you can still use the nifty remote handle controller. This will work for 7D and 5DmkII but not T2i. __________________ Tim Dashwood

May 11th, 2011, 12:41 PM	#9
Pavel Houda Major Player Join Date: Mar 2009 Location: Laguna Hills, CA Posts: 352	Re: $100 DIY Beamsplitter Rig I saw the article on this issue here: Genlock Canon 5D's & Shoot 3D ? . BTW, on slightly different topic, nother interesting approach for Panasonic camera genlock is described here: "Hi, All Recently I completed making of 3D video rigs based on Panasonic GH2, a DSL(R) HD video camera with 60fps sensor output and 60i AVCHD recording. 3D sync is almost perfect after post editing in a Premiere CS5 60P sequence as far as judging from audio waveforms of R/L clips, more quantitatively approximately 0.4 mS measured by means of external flash. My particular pair of GH2s with adjacent serial numbers shows relatively small drift of 0.05 mS/min. Therefore 12 min of 3D clip can be shot in sync less than 1 mS. 3D photos of various kinds of rigs and flash sync test result is shown here; Stereo Flash Viewer The syncing method is very simple with two additional components, an external battery (9V Sanyo rechargeable battery for musical instruments use) for just power-on of both cameras simultaneously to start internal clocks which governs REC start sequence and a spliced remote shutter (for 3D photo shooting) to trigger video recording. Because of the unidentified procedures in the camera before starting REC, from exactly zero to several frames of 1/60 sec difference exists between both cameras. however it can be easily fixed in the post edition. Half-push to complete the Auto-Focus before full-push for REC ON and setting shutter speed to faster than 1/60 sec is the necessary condition for the sync. No sync failure have been observed so far. A drawback of this method is some clue e.g. clapper sound etc. is required for the post sync, however because of the discrete nature of frame shift of a unit of 1/60 sec it is not difficult to sync R/L without any clue by the anaglyph monitoring of the 3D scene. Here is a test 3D video for sync and also for EX Telecon function ( x2.6 zooming with keeping Full HD resolution of 1920x1080, one of my favorite functions of GH2) used for shooting 3D telemacro video up to 1560mm equivalent. YouTube - 3D Cherry Blossoms in Spring Gust 2 (Telemacro HD Video with twin GH2) inoue_k3D"

May 11th, 2011, 04:19 PM	#10
Neil Richards Regular Crew Join Date: Feb 2011 Location: Somerset, England Posts: 147	Re: $100 DIY Beamsplitter Rig Thanks for that Pavel, that's really useful. I have the GH2 and have wanted to do some stereo with it so I can try this to get the sync sorted out. I have two SD700's as well so I'll see if it will work for those too.

May 11th, 2011, 07:30 PM	#11
Karl Kozak New Boot Join Date: Nov 2009 Location: San Diego California Posts: 13	Re: $100 DIY Beamsplitter Rig Rig looks cool... nice job. The thing to be wary of is the glass. You need a high-grade beam-splitter glass to get quality 3D. Best of luck !

May 11th, 2011, 07:49 PM	#12
Jesse Blanchard Regular Crew Join Date: Apr 2011 Location: Portland, Oregon Posts: 97	Re: $100 DIY Beamsplitter Rig Yes, I know you guys at 3D Film Factory are always going on about the glass. But I haven't noticed anything in the final images. Where do you guys source your glass? Do you have some specs on why it better? Do you have certain scenarios where the glass doesn't work? I'd love to know more. Here's some macro footage I shot without issue. 3D Spring Flowers (red/blue glasses) on Vimeo