When small interaxials just don't cut it! - Page 33

Jeff- on your post above about frame 375. Had you considered this could be the function of a very short duration flash from an explosion with short decay combined with the shutter speed I was using? As you know the frame rate was 60 progressive but you did not know that I was shooting manual shutter speed of 1/250 second. Maybe the better approach would be to make a longer duration exposure ( 1/60) to allow the shutter to remain open long enough to expose both cameras since the idea of syncing the timing perfectly without genlock is not likely to happen.

It appears that the reflection across the water came from a bright flash reflection on the glass of the building from the other Barge. See the white spot on the building? that's the one additional spot in the frame that is also missing from the right. I think all the other explosions were longer duration.

You should find many of these left not right or right, not left exposures near the end where there were lots of flashing of very short duration. Here I also get a bit disturbed by the lack of smooth light. Maybe using longer exposures is indeed the answer? ( still get the flash but the 3D integrity would remain) I will shoot the next Fireworks at 1/60 shutter speed to test.

I assume that that type of device is passive (doesn't control the cameras) and matches to the nearest frame (e.g. 60p = 60fps, 60i = 30fps). If more sophisticated, it could match to the nearest interlaced field by extracting and combining fields from adjacent interlaced frames. Anything finer than that would seem to require motion interpolation, which would be ok for uniform motion...

I can appreciate that a Lanc Shepherd for asynchronous cameras would be a very useful measuring device. If operating two cameras the worst possible mismatch if capturing "blind" at 60p is 1/120th second [if it otherwise greater, the relative alignment of the clips can be slipped a multiple of 1/60th sec in edit]. But with the Shepherd you can achieve much better than that. By chance, there is a 50% likelihood the error will be no more than 1/240th second. And I think a 25% chance it will no worse than 1/480th second. The Shepherd device allows a good "roll of the dice" to be identified.

A little project I'd like to do would be to capture a railway yard using wide IA, and edit so as to speed up the rate of motion by a certain factor to have the trains move as fast as they appear to in a model train set...
Edited by MLXXX - 7/8/12 at 9:23am

Yes I did consider that and I think using a longer exposure time could indeed help to even out the visibility of these very short duration, but very intense, flashes.

Using a longer exposure time could also:
(1) Reduce the apparent brightness of very fast moving spots of light to the point of invisibility
(2) Mask the precise position of moderately fast moving specks of light [probably a good thing, to avoid the need for millisecond precision in the capturing sync; could - paradoxically - give a firmer feel to the 3D effect].

Just as 24fps traditional cinema used a 180 degree shutter to help smooth the apparent motion, there might be advantages in deliberately introducing a little "smudge" at the time of 3D capture, particularly if there is any likelihood of the capture being rendered at 23.976p. I'm not sure about this.

Let me chime in with how much I enjoyed your video, too, Don. I was so burned out with regular 2D HD video editing. This 3D hobby makes it all fun again, doesn't it? It's a whole new world.

Frank,

Are you starting a new business or moving an old one?

BTW. I love this LG passive monitor more every day. Along with Edius, it's made 3D editing so much easier. The first couple of days, I was sitting outside the sweet spot, and I wasn't that impressed because of the ghosting. The 47" LG is just right from my normal seating position. It's amazing how much of a productivity boost it gives. I double click the Edius screen, put on the glasses and watch what appears to be a full resolution/full color preview. Outstanding! I had pretty much talked myself out of buying it until you convinced me otherwise.

Jeff-

The frame store synchronizer will line up video frames accurate to the microsecond! Remember, video camera's local oscillator is very accurate and keeps the starting point of each line of each field at very precise intervals, down to the microsecond. So, if you have two videos that are not locked to each other, then the frame store synchronizer can lock the two videos so that the frames on each are started at precisely the exact same moment in time. The first line of an interlaced field starts at the top center of the video field. The synchronizer will line up the same frame at that same top center position so the two videos will be in sync exactly. There is no modification of either camera. The original purpose of these devices was to synchronize an incoming video signal demodulated from an RF signal off a microwave link or satellite down link with the video being generated in house so the station could mix and blend the two sources in editing. There is also a device called a Time Base Corrector. This is different in that the TBC must feed back an advanced sync signal that will shift the timing of the start point of each horizontal line trace in the video. These devices do not work unless they can be connected to the local oscillator of a video source. Plus they only adjust the start point of a couple lines at a time. Common application of these are to stabilize the imperfect timing of a tape player to correct the quiver of the start point of each line as picked up by the servo motor spinning the head in the player. Motors are not very precise so this makes the output of a tape player as stable as a camera, but this would never work with consumer tape players. They worked mostly with broadcast type equipment that have advanced sync and Black burst inputs. Genlock is only good for non-motor driven video generators like Cameras and character generators. Simply put genlocking allows the different devices to adjust to one clock timer called a house master clock.
The Lanc shepherd and other LANC controllers just send a power up signal to each camcorder that starts the clock ticking. This is a very crude way to get close to a match in frame timing. But the errors here, measured in a few milliseconds, the genlock is right on the money and a frame store inaccuracy measured in fractions of a microsecond. Quite a difference in scale. What is interesting is the meter in the Lanc Shepherd for letting me know how off I am with the error. But, I don't trust it and prefer to use the old fashion clap board. The reason is I can work much faster. You see, the Lanc Shepherd at best will be off by a frame or 3 and that is when the LS gives me a 10 ms timing error which is common. So, I really don't know how trustworthy the LS is for it's metering. Clap board works and is a very simple and accurate way to line up the clips as easily verified by the video movement onscreen.


Thanks Joe- It was a good learning experience doing my first project in 1080 60p. Amazing how much additional rendering time, and drive storage this took over 1080 24p. Overall I'm not sure it was worth it as I've been pretty satisfied with 24p for most of my work.


Frank- you retired too soon! We're about the same age and I have no intention on going back to meeting deadlines and ridiculous client demands and threats. Until someone is ready to hand me money and stay out of my way to do this 3D work, I will remain doing it as I currently do. As a hobby.

Don,

I doubt they start the first field at the top center of the frame anymore. They did that when there were 525 lines in an SD frame (and the math dictated that the field start in the center of the screen). It caused glitches when the first and second fields lined up incorrectly. Since there are only even numbers of lines in 1080 and 720 video, they'd have no reason to introduce that can of worms into HD production.

Now we're really showing our age.

The point I was making, Joe, was that the timing precision of video does not vary from one frame to the next. I don't care if the example of timing reference I used was in place for RS-170 standard or ATSC HD video. I used the RS-170 standard because it was simple and as basic as it gets not to mention my age makes it my mother tongue in TV engineering.

I don't mind showing my age either. I received my BE degree and First Class FCC license in 1971 which is why I often think in the basics. FWIW.

Yes I appreciate that with such a device the content of an incoming video stream can be stored and then clocked out to match the frequency and phase of a local frame rate. But the device appears to have no power to change the time at which the camera shutters of the two cameras operate relative to each other. [If it did have that feature, and one of the cameras could be controlled,one camera could be slaved off the device.]
That seems more the sort of device that might come in handy. Again it needs at least one of the cameras to be controllable.

Before acquiring my Sony TD10, I'd been looking at the Go Pro Hero2 3D system which has a short cable running between two cameras to synch them, but the max frame rate specified at 1080p is only 30fps, and there is no zoom.

I don't know much about camera lens quality but I can tell from inspection of the images that the geometrical precision of the image from my TD10 is far superior to the geometry of the lens of my earlier camcorder (a Canon HV20). So although the TD10s lack any genlock type of feature they do offer a good standard of lens geometry, making them a good choice for pairing for wide interaxial shots.

If I had a Lanc Shepherd I would certainly use it for the more demanding subjects so as to reduce mismatch. A 1/480th sec discrepancy (2.0mS) is much better than a 1/120th sec discrepancy (8.3mS). It would definitely be worthwhile I believe for fireworks. Not sure about water from fountains, viewed from a reasonable distance, but it certainly wouldn't hurt. To get much below 1/480th sec could be too much of a hassle, and make little if any noticeable difference I would guess.

Jeff, I don't know of any device than can sync shutter speed. But then I haven't really researched it. The closest to controlling a camera's settings remote would be a CCU or Camera Control Unit. These work with studio cameras and can adjust various video settings but I'm not sure about shutter speed. I've never used a CCU with any of my cameras.

The Go Pro is an interesting system with the link cable. It's possible it is a genlock design but I don't know that. Likely if it switches both cameras on with the control of a master camera then it probably is just like Lanc and it is only a way to slave the on off switch. If you turn each camera on separately but just have the cable connected, then it is likely a genlock design. There are also rules to shooting genlock camcorders. The cameras must be recording time code in free run, drop frame. But, that can be done internally and set when that cable is connected. My (Broadcast) cameras require the system be set in the switches or menus to genlock.

When the Lanc Shepherd turns on the two TD10's the timing is normally 1/120. If I switch on and off over and over, I can often do double that. I've never seen 1/480. It's like throwing dice with diminishing odds for the higher sync.

I note that a full frame at 29.97fps corresponds to 1001/30 mS = 33.367 mS

So a raw discrepancy of, say, 35.000 mS in the time line would become an error of only 1.633 mS, after slipping the frame alignment in edit by one frame.

It seems that one camera might generally be slower to respond to a "start capture" control pulse than the other, making attainment of near zero timing disrepancy problematic.

Don, these questions arise in my mind:

Originally Posted by Don Landis View Post

You see, the Lanc Shepherd at best will be off by a frame or 3 and that is when the LS gives me a 10 ms timing error which is common.

And that's why you need to use the clap board or some audible marker so you know whether to slip 1, 2 or more frames and in which direction. You must trim the excess before pairing.


Lanc controller has nothing to do with pause and record sync. It only sets the sync for power up. Hitting record button will start recording. If you hit record on one of the cameras locally it will record a file in 2D on that cameras memory, then you can pause and fire up both using the Lanc controller. The only problem here is in post when you try to assemble all the clips you shot and match to the clips from the other camera there will be that one lone clip in the list that you will have to remove from the timeline to get alignment of the remaining ones.

I don't know. I had not seen this before but will look into it. I see two issues for me and first I have to see if my LS has the right firmware as this appears to be a feature update for new models. Second, the process would need to be supported by the camcorders.

Overall, the LS precise sync is not a terrible problem to deal with in Vegas as the lineup of the clips is pretty easy. When shooting twin cameras, geometry alignment is a harder issue to achieve than adequate frame sync. But, I've told you this before.

I've experimented using two TV sets side by side to view 3D cross-eyed. A small amount of vertical misalignment or of non-matching zoom (picture size per square inch of screen), as between Left and Right would pass completely unnoticed for my vision. The brain is remarkably flexible in being able to fuse two images. (And colour balance could be quite different. The brain seems to just "average out" the colour.)

My brain will complain if part of the image contains over-demanding parallax, or if there is a sustained windows violation, particularly if it affects a large part of the border.

The 3D stereoscopic adjust plug-in in Vegas will often do a good job in compensating for geometrical alignment discrepancies, but this will I suspect depend on the amount of data the algorithms have to work from. I suspect that fireworks using a narrow view of the sky, with a few points of bright light, pose challenges as there are no readily identifiable objects in the far distance for the software to use as reference points.

In any 3D workflow a limit has to be drawn in terms of how much time and effort will be put into adjustments for geometrical alignment. However, one thought that came to my mind was to briefly point the pair of cameras first at the skyline and then tilt the rig upwards to shoot the night sky fireworks. If changing the zoom setting, the cameras could again briefly be pointed to the skyline. I'm not certain, but I imagine that the view of the skyline could provide better data for Vegas to use to calculate its geometric adjustments.
Edited by MLXXX - 7/14/12 at 9:12pm

I have shoot some footage with the side-by-side rig and the two TD10s now. Some impressions:

- definitively, it is better to adjust the two camcorders in full zoom. After adjusting the two ballheads and adding the two camcorders, it is necessary to align the camcorders in a more precise way. Both horizontal and vertical alignment worked perfect with a building in a distance of 70m, with clear horizontal and vertical lines. I have also checked tht with the horizon and the sea, and it was in line.

- the zero angulation is best done with a very far background, that was more then 1.5 or 2 km away. The far backbround is then in perfect superimpose. Here the separate adjustment screws at the bottom of the ballheads were really helpful. It is not enough to adjust that at 70 or 100 m - doing so and checking that for a far distance shows that the axis are not in the zero angulation then really! So I decided to use the adjustment in the very far background - and then you end up with a great adjusted system.

Doing some testshoots and using the automatic adjustment in Vegas showed vertical offsets of -0.006 to -0.003, a zoom of 0.003 to 0.013, keystone of 0.001 to 0.005 and a rotation of 0.05 to 0.13.

So especially the important vertical offset is very low, lower then the losses will be due to the horizontal offset that has to take place due to shooting with the parallel axis. That is great so far.

Controll of disparity is a tough exercise. The issue is always that I do not have a 3D field monitor. So especially the nearpoint is dangerous. I use the iphone calculator Rittai Calc als recommended by Don - that is great. I have my laser to measure distances - fine. But sometimes it is hard to identify the nearpoint - and then the calculation can be wrong too, and the disparity can be too large. I think one has to be aware that shooting depends from your screen size - and you must not go beyond the 6.5 cm eye-distance to avoid a divergence of the eyes. For my 50' HDTV that means that I have to be below 6% (or 6.7 mm = 2.64 inch).

A drawback is, that the TD10 does not show you the actual number of the focal distance f for the different zoom ranges. So you know that you are somewhere between 34.4 mm (full wide angel) and 344 mm, f calculated to 35 mm. I have not yet calculated f for the more then 20 zoom positions, even if that can be done.

Shooting in the full tele of 344 mm of the TD10 pair gives you limited range and requires sound adjustment to avoid to go beyond that limit, otherwise eye-divergence will occure. With a farpoint of 150m or 492 feet, and the minimum IO of the system of 11cm or 4.33 inch, the nearpoint must not be below 16m or 51 feed, to achieve 6% parallax on the screen. Sometimes that is hard to check, since you may have the floor in your video. If somebody wants to stay below the 2%, the minimum near distance would be 51m or 167 feet with f=344 mm!!!

Also the background screen must be checked carefully.

With the wide angle of 34.4 mm that is less critical, here you come down to 1.7m or 5.6 feet.

The toughest points are a good adjustment of the two cameras, the clear identification of the nearpoint, the consequent use of the calculator and the clear setting of the rig. I wish the rig would allow IOs below 11mm or 4.33 inch - but for that I would need other cameras. Fortunately, I have my Z10K too.

I have not been able to check the results now on my s3D editing place or 50 inch HDTV yet, but I have Vegas here with me, on an old laptop to check the results using the anaglyphic preview. I love the grid overlay, especially the horizontal grid that can go down until 5% - unfortunately not below - that is adjusted in the preferences, video, horizontal grid setting where the maximum input are 20 what are 5% in the result.

I knew you would come around to figuring out that alignment of the cameras in full zoom is the best, most accurate way. Like Frank, I rarely do any adjustment in the field these days as the quick release dovetails get the vertical close enough for my satisfaction. My typical vertical is adjusted between .006-.015. Not as good as your manual adjust but then I haven't done an alignment in months! I still need to verify the horizontal angle the cameras are so they are parallel. A couple times the little quick connect mounting foot gets slightly out and needs to be reset in the field. I usually spot this when looking at both cameras and see an object dead center on one camera while it is barely in the shot off to the side on the other. Usually a twist of one camera is good enough to fix. The bigger the vertical adjust the more your Vegas zoom will correct but that loses resolution so I agree it is best to have that number low as possible.

The near object minimum distance is really critical when you are shooting extreme ultra wide stereobase ( SB ). I was editing a couple shots on Bryce Canyon yesterday where the camera was to the edge of a 2000' cliff. I was using 28" SB and didn't notice that a part of the cliff jutted out into the shot during a pan in wide angle. So, I just set some keyframes in the pan crop and as the pan saw that part only 20 feet away, I just pushed in on the shot framing it to the upper part. Otherwise the 28" SB had this rock completely double image while the rest of the scene was gorgeous. Another good rule of thumb for knowing if your near objects are too close for the SB is when shooting you see an object well in view on the right side of the right eye camera but it isn't in view on the right side of the left camera.

BTW- I also found your method of using the stereo adjust filter in the track too restrictive. I continue to set the keyframes in the event. I just have more freedom of movement in the project editing that way. Actually, since I have so many clips with different settings I use the two Vegas system to prepare the pairing and adjustments. Here I work with the entire clip length and after paired, I copy the clip and paste it into another launch of Vegas for the program. Then I review the smaller parts of that clip for the edit and slice it out and place the clip in the movie. Now I do another auto adjust setting keyframes for the project compatibility, especially the Horizontal for the depth placement. I use some zoom in my shots so here I have to set a key at the beginning of the zoom and again at the end of the zoom so the ramp will adjust the calibration linearly with the zoom. That was a trick I learned from Gary Rebholz at Sony. Finally I add any color adjustments and pan crop keyframes. I use track effects and filters more for audio than I do video. If I shot some bad video for a whole session and all the clips needed the same correction, then I would dedicate a timeline to those clips and set the correction for everything on that timeline.

Don, I really think the solution is to use the wonderfull calculator in a consequent way! Even if I see that sometimes it is hard to identify what the nearpoint is - then the calculator will not help! If something happens like your example with the part of the cliff, the question is if it can be used at all. I have seen examples with the sand on the beach mainly - even that can show a large disparity.

So to my opinion, with a rig with such a high IO it is really critical to align the camera axis to parallex axis - then - and only then - you can end up with limited numbers for the disparity. Even if you shift everything behind the zero window, you should be able to stay within the calculated disparity - if everything is fine.

To my opinion, there is the special situation for hyperstereo. Shooting with huge IOs will require convergent cameras really. Maybe especially important for you, but with our sbs-rigs we all tend to shoot with large IOs. To my understanding hyperstereo is everything beyond the average IO of the human eyes (so > 6.5 cm).

And you will depend strong on the screen size that you produce for - the lager the more important it is to bring the disparity down. I have decided to preduce for my 50' HDTV only, that is demanding enough - here it is possible to allow something about 4-5%.

The figures that I posted are mainly for zoom - I have not shoot here a lot of footage in the wide angle. That is something that I still have to test. But what I see with mine system as it is adjusted yet, the vertical offset with f= 34.4 mm will not be too huge.

Don, can your maintain the keyframes when you copy/paste events - if the filter has been applied within the event? That would be an advantage. I still like my system to apply the s3D filter at the track level, since it allows me a great overview about the total project. That is something that I have not seen in Edius, and that is something I miss there.

Actually, speaking from lots of experience it is pretty easy to know what the offending near object is. You will see it on your monitors. Then just zap it with the range finder to measure the distance. Even a near spot in the sand you could see and spot for measurement. That has never been a problem for me. In some of the shots I set up that way, I intended to create the look of a sandy beach that had a long range from just in front of me to way behind the screen frame. Yes, this does tax the eye muscles a bit but it gives me the look I was after.

What can be a problem is the example I mentioned with a rock jutting out into the shot because it is located 40-50 ft away and the main scene is at 1500-2500 ft behind it. Now we have a serious problem. It doesn't work.

Hey- I've been meaning to ask you if you know of an easy way to screen grab a place on the timeline and output it to an MPO file for 3D still? Vegas can't do that, only jpg and png. I wonder if I render a short clip to a T/B clip and then import to PD10 if PD 10 has that capability. Can you suggest to Sony team that be added as a feature in the next version of Vegas?


Oh, forgot your question-- Yes, the keyframes and any other filters remain intact for the copy / paste. Only that bug where the "pair with next clip" gets shut off. I have to turn it on again but then if I copy and paste that clip a second time it is ready to go.
Edited by Don Landis - 7/31/12 at 1:10pm

Today, I'm excited!

I just completed ordering the pieces to build out my second 3D twin camera system. Most of you know the direction I've been heading. I will be using twin NEX 5n's with 16mm + .75 wide angle adapters added. The main purpose of this rig is to have fixed lens wide angle options and 1080 60p x 1920 with more glass and pixel density real estate than the TD10's can do. My test in June with my nephew's camera went well and I was able to sync the NEX 5 system manually good enough to get some really nice wide angle images. I have now 5 different short projects for this wide angle - wide Stereobase system in mind. Two of these are in Disney.

Contgratulation to the new system.

I think about to go into the direction of a beamsplitter rig. My experience with the sbs-rig ist good - for some purposes. The large IOD of >10cm is a disadvantage for some shootings, even with smaller cameras one is not able to overcome that. So mabe that is my next step.

My idea is that we will end up with different systems that we use for different situations - and combine them according to the situation.

Yes indeed! I found the use of The NEX 5n for extreme wide panorama 3D plus the TD10 and twin TD10, plus the Panasonic 3D1 combination of 4 systems having their unique advantages. Beam splitting can combine the TD10 with the Twin TD10 purpose but it is complicated to use and align not to mention extremely expensive and awkward to haul around and set up. I just don't see it in my deck of cards. The system from Cyclopital 3D that Joe has, now, is a good possibility. If out for a days shoot in the park, the SBE is no more bulky or attention gathering than a twin TD10 rig and much simpler to use, I think. If I hadn't modified my TD10's I would probably have purchased one already. I was thinking more in line of budgeting for one with a Z10k kit some day. ( Not in the near future, however. )

Yes, but the advantage of the beamsplitter is the variable IOD - down to low ranges where our sbs-rig cannot compete really. I will continue to use the sbs-rig, it was a great learning point. The beamsplitter - maybe it is more complicated, but I think the beamsplitter is an ideal learning point to understand quite well how to adjust the beamsplitter too. Still thinking about it ...

The low end is about 100mm I.A. Frank was experimenting with mixing the right cam image with the left cam image on two cams to get a closer I.A. but then had to extract those from the 3D of 4 cameras which reduces the high quality of the 3D settings we can do in the TD10.

- My Bryce Canyon documentary is uploaded to my server. The iso is about 6 GB. just add a /BC to my domain name used for the Valley of Fire for the folder to download. It ended up being 37 minutes long. Hope to get the YT uploaded soon.

Yes I know - the minimum with our TD10 is about 10cm. A lot more then 6.5 cm for the average human eye distance.

Great that you have the video ready. Will give that a try, but later.

I don't have a lot of experience with benefit / liability issue with making small increments of I.A. change. So, when I shoot with a single TD10, vs a pair of TD10's using minimum IA, is there a significant benefit to having an increment in between setting? I do see one advantage in that the single TD10 vs twin TD10's demonstrate a superior image quality in detail. I don't know why that is but I just see image quality of two TD10's in 2D mode paired being sharper, crisper, and more detail than the MVC 1080 60i.

The general issue is that with a huge IO, both the depth impression is increased - an effect that is welcome to some extend if you compare that with the Wallys. But more important, the disparity tends to become too large. So a sound calculation where you can adjust the IO in a perfect way is the way how to come up with an improved s3D. I have done some test shootings playing with the IO between 50cm and 10 cm. You can see the increase in depth if you go to the larger IO - quite clear.

Another effect that I have seen now, shooting sea-lions in full zoom, is what they call cardboard effect. Here you need an IO that is larger - to avoid that.

There is another factor where I am not so sure how to see that - roundness. Seems to be calculated by different tools, but playing with one of them showed me that it seems to be hard to come up with good figures for roundness if you use consumer cameras? Cannot imagine that this is right. I have seen that with the free version for the iphone, of IOS Calc jun.

My impression is that the calculation tools deliver different figures - the results from RittaiCalc3D seems to be different from IOS Calc jun, and the Rittai does not even calculate the roundness (but seems to deliver figures to controll the disparity, here is is great). So I am looking for a better calculatoin tool too.

You shoot with the two TD10 in 1080 60p? That is a superior format, compared with 1080 60i. I shoot in 1080 50p too, but the format is demanding in the postproduction. Both Edius and Vegas are not able to playback a paired clip with 50 fps, here I see a drop in performance compared to MVC.

Most of the scenes in this video were shot with 12" to 28" IA. The signs were shot with the Panasonic 3D1 as MPO's I used UV filters on all but the 3D1 camera.

The video was shot in the highest TD10 2D mode for the Twin cams and at 24p frame rate. Single TD10 was shot at 60i in 3D of course. The location Piracy Point was shot with the 3D1 in SBS mode. The Google Earth 3D maps and tour had a set of lines turned on by default and I just figured out how to shut them off and still keep the roads and signs turned on. Google Earth does not have very good documentation on how to do stuff. The NEX 5n was used to make the 3D panoramas in a sweep panorama mode. The older video I shot of Bryce Canyon was in Hi8 mm Sony camcorder. The underwater shots were done back in 1990 with Hi8 and regular 8mm cameras with my own home made housing and my own 500 watt underwater lighting for the underwater caves. I also have a short clip of Bryce Canyon shot in 2003 with my first HDCAM.


If anyone here would like to download the iso file to burn a 3D BD, I can give you the link but I have to warn you the files are 6 GB so be prepared for a long download even with an extremely fast connect speed. Send me a PM and I will give you the link.

If I may say so, Don has produced an informative, professional documentary, including some breathtaking 3D shots. As reviewers say, "Recommended".
.

Deleted by author - sorry irrelevant
Edited by ComPH3D - 8/8/12 at 1:06pm

Don, thank you for the nice video. I think it is great - and I love the pictures.

I do not know what ComPH3D means to tell us.... ?

Glad you enjoyed it, Wolfgang. I have to admit I used the Valley of Fire video as a practice piece for Bryce Canyon. Practice with not just the 3D aspects but also the Google earth. I'm still not 100% with GE yet. It is a challenging piece of software to learn. I could have used the 1 meter slide rail on that one but the 28" length is what fit in my luggage for travel. When I do the Grand Canyon and Monument Valley, I will be sure to make arrangements for the wider IA.

COMPH3D- Why did you delete the link? I enjoyed his work. I am working on my own rig to do a better job of shooting the Bellagio Fountains and found his work to be quite educational. There are some others up there but not in wide base 3D. So far all I have done is single TD10 on the fountains. He did 100 cm which was about what I had in mind. But I will be using 12mm lens assembly on twin NEX5n's. Hopefully, I will be receiving all my hardware this week and get ready to shoot a local small fountain to get some practice.