When small interaxials just don't cut it! - Page 34

Sorry Don/all. I must admit that I didn't read the whole thread, and my input didn't make sense to Wolfgang, I assumed it was irrelevant. I don't want to violate any forum rules, so I withdrew the input. I met with Inoue-san recently and talked lot about his rig and experiments, some of them with pretty extreme inter-axial distance. He just informed me about the new video with 40 cm stereobase, so I listed the link. He has done some lot more extreme stereobase work, some of it shown at the 3D-convention last month, so I wasn't sure that my input was relevant. I should be more thorough, was quite busy lately. My apologies. Here is the link: http://www.youtube.com/watch?v=LjpsMt7MmzA .

I did not state that you should delete the link - I only did not understand what you want us to say.

I like the extreme stereo base, for the reason that you end up with a great depth. But I am also aware of the danger of using that - to my experience the nearpoint can be too near, and so that must be handled with care. I run myself the same side-by-side rig as Don uses - with up to 160 cm (but frankly spoken, that is experimentell up to now, as a lot of things here).

I have also seen a lot of hyperstereo of different guys. And I love it, but I think it has to be created carefully, to avoid too high disparity and diverging eyes (painfull!).

Again, sorry. I know you didn't ask to delete the link. I should be more careful before posting.
Edited by ComPH3D - 8/8/12 at 3:52pm

This is my favorite hyperstereo base YT clip. Cameras are said to be spaced 25 meters apart! I'd like to know how he goes about aligning the cameras to the scene.

Between that and his slomo the work is unique.

http://www.youtube.com/watch?v=w5EWMTZlccM

Vivid 3D effect. The true shape of the various fireworks is revealed, well beyond what unaided human vision could perceive. Quite something!

Fireworks as God would see it--quite amazing.

You simply stepped into a threaded conversation by two posters. Don't lose your spontaneity for sharing...it's what makes this forum great.

True: in this thread a limited number of user has been involved in the discussion - 3-4 user only. But other uses are welcome to participate for sure. So I do not think that this is a reason for anything, but my point is simple that I did not unterstand the message. For sure ComPH3D is welcome to participate here! And there is no reason for any kind of apologize.



Back to the topic: I have ordered yesterday a beamsplitter-Rig. There is an interesting beta-offer by Jesse Blanchard,

http://www.dvinfo.net/forum/3d-stereoscopic-production-delivery/509519-new-beamsplitter-market.html

and I have decided to give that rig a chance. I will run that together with my other systems that I have now - what is a lot: 3 units of TD10, 2 of them can be combined on sbs-rigs up to 130 cm. And one Z10K. I will try to use the best system for the specific purpose, and at the moment I will use my paired TD10 with the beamsplitter too.

The price of the rig are US2500, but for beta testing there is a special offer - and he still has one unit left for that purposes, as far as I know.

http://www.ebay.com/itm/290713904931?ssPageName=STRK:MESELX:IT&_trksid=p3984.m1555.l2649

I have become aware of one major limitation of the consumer TD10 cameras, that is the format but also the chip size - what seems to result in a very limited roundness. But since I do that for fun at the moment, I will stay with the TD10 cameras for the rig at the moment. Beside that I have not tested the 130 cm sbs rig yet really, but I am keen to do that.
Edited by Wolfgang S. - 8/10/12 at 5:12am

Good luck with that rig. It looks good and is about 1/3 the price of the traditional units. I'm sure you will find the beam splitter challenging to align.

In pairing the clips, the 3D adjust effect in Vegas has some flip image settings that you can set which should make the process easier but what I am curious is how you do the initial mechanical alignment. Wide angle will be especially tricky.


I got all but one wide angle adapter in now for my NEX 5n rig. Doing the alignment work today. I may need to add some set screws to my slide tables for more precise camber control on the NEX 5n's The mounting base area is quite small and I want to have an easy alignment process in the field.

BTW- the NEX 5n rig will have a minimum I.A. of 125mm. If I ever get into macro s3D I guess I'll be needing to think about a beam splitter.

Nice to get out of the edit suite and back in the shop after these past 6 weeks on those two big projects.

Well, I think it will be fun to explore the mechanical pairing in detail. I hope that the sbs experience will be helpfull.

125 mm sounds to be a lot. Especially the nearpoint disparity will be something that you have to watch with care.

My near project object will be about 150' and shooting at 12mm wide angle so if you run the numbers through your calculator, it should be fine.

The example COMPH3D put up was no problem and he was working with a close object distance of about 60 ft. And his I.A. was 400mm! The secret here is that we're using very wide angle imaging. If you do some examples through your calculator, you'll see that wide angle is what makes these large IA's forgiving from the disparity problem. But, of course even this has it's limits.

There are only two advantages I see with using the NEX 5n and that is higher pixel density and fixed wide angle lens options. Otherwise I would just use the TD10 with 1080 60p. and have zoom options. That is the tradeoff.

I you shoot with 12 mm wide angel, 150 feet as nearpoint 2000 feet farpoint, assuming a parallax of 2% and producing for a screensize for 50 inch I end up with a stereo base of 9.8 feet - what is a lot?

That's right and using 12" IA will be a good safe wide stereo to achieve the depth for my projects with full width and pushing the background to about 750 to 1000 ft. The 12mm will give me a miniature effect of the background so that tall buildings will be squeezed into the shot frame height.

I just found a couple of cheap fisheye lens attachments that will work with my 18-55 zoom and my 16mm lens. My current wide angle is .75x which is giving me about 85° measured capture angle with very little curvature distortion. The cheap adapters are rated at .42x so there will be considerable barrel distortion on the 16mm but using them with the 18-55mm I may find the happy compromise with full frame and some curvature. The price is so cheap I couldn't resist as an experimental cost.

The Sony Fisheye is much higher quality, is .62x but costs $110.00 Sony says this adapter on the 16mm gets 180° full frame. We'll see. ????

I'm currently searching for a local picture spot to do some testing next week that is more interesting than my backyard. Considering that water and flat land will be my lower foreground for my projects, I may just use 16" IA to get plenty of depth and roundness ( volume) in my subject. The testing will zero in on what looks best ( to me).

Don's video above if downloaded direct from YouTube at 1080p side by side results in a 1.5GB file and includes personal commentary about the 3D photography. The BD iso link Don has kindly offered provides a more business-like "cut" without some of the commentary on the photography. The 3D photography looks even better in the iso version.

I don't think Don is looking for nit-picking comments. I can honestly say that the 3D effect for my often critical eyes was very satisfying. [Any slight strain I occasionally perceived in the 3D was momentary and minor.] I was impressed that Don did the commentary himself.

Thanks Don for sharing this very interesting video.

I've watched this and the Valley of Fire iso's offered, pretty cool stuff at both 65" & 110" ... Thanks Don!

This video was made with two Sony Camcorder HDR-CX130, with stereo base 20cm and 28cm. Originally recorded in 1080p60. Edited with Sony Vegas Pro 11.




Thanks
Milton
Edited by Powerplay4 - 1/25/13 at 3:02am

It is sad that both these videos combined have 700 views while hitting someone in the nuts with a vacuum cleaner in 3D nets 5000+. Idiocracy isn't a comedy, it is a biography of our near future. Great work Don and Powerplay4 as well.

Reality TV has become very popular. But a common theme of the popularity is not watching someone like me every day sitting here editing or building or cleaning the pool but rather people who get into arguments and fights and use moronic language. Content or story is king! quality of production is not so important. Why do shaky cell phone videos of the same event get far more views than a quality stable production. That's one question I'm still puzzled over.

Anyway, thanks for the compliment.

Milton- Nice work! I see your setup is locked to specific I.A.. Do you think that is better than having a slide bench for full variable I.A.?

Don, I bought "Dual Camera Bracket Mount 2 Cameras Stereoscopic 3D" and I can vary the interaxial from 8cm to 30cm. What I have shown in the photos is just one example (8, 20 and 28cm). I am also using Manfroto 323 RC2, as you suggested. I want to use my slider 100cm with two 323 RC2 for landscapes far away as you did.




Milton

Milton,

nice rig, very simple to what we run here. I have not seen your video yet, but will do so.

How satisfired are you with the quality of the adjustment of the two cameas, that you can achieve with your equipment?

Do you use a controler to sync the two Sonys?

And how fine was the editing of 2x1080 60p in Vegas 11? For 2x1080 50p I find it a little bit tough - my PC is an 4.4 Ghz overclocked 2600K, I have something about 30-40 fps (instead of the 50 fps that I would like to have). How is your situation?

Very Interesting, Milton. Wolfgang and I are using the igus slider for the wide stereobase bench. I actually bought three as sold by Amazon.

Then with 3 rails I can make different length benches convenient to travel kits for what I need with the longest being 1 meter. The Amazon price of $79 each USD, is much less than buying separately from igus. Once the quick releases are properly mounted to the igus slider the alignment holds permanently and only the small dovetail mounted to the camera needs a quick alignment when attached. calibrating camber and tilt is a non event with my setup now. The plan for the three slider tables is to put two wide angle NEX5n's with 10mm lenses at each end with about 50-60 cm I.A and a single HDR TD10 in the center to shoot the same scene but the TD10 for zoomed in closeups or medium closeups.

Here are some images of my parts used to make my modular system:

IMG_0355.JPG 1098k .JPG file IMG_0327.JPG 1564k .JPG file

IMG_0334.JPG 1282k .JPG file

IMG_0338.JPG 2192k .JPG file

Hey Don,

How did you cut the Igus slider? What tool did you use? I cut one piece but with the tool I used, it took almost forever.

Wolfgang, answering your questions: I am very satisfied with two cameras because they are 1080 60p, same model and serial number sequential. I put white balance = outdoor (for external shots) and steadyshot = off (on tripod). When white balance = automatic there was a little diference between left and right images, with outdoor the images is the same.
For now I don't use any device to control the two cameras. I bought a Sony RM-VD1 ($ 50) and wanted to connect three wires (GND, LANC and +5V) in one camera and the other two wires (GND and LANC) in the other camera, but it did not work. I plan to buy a LANC Shepherd (~$ 400), but to import to Brazil, usually we pay almost 100% of taxes. This makes it very expensive. For now I'm snapping my fingers next to the cameras at the start of the recording and then I sync by the audio tracks in Vegas. How do I record in 60p, one frame will fit a maximum of 16ms. I still can not use the zoom of the camera. In this address http://3dvision-blog.com/forum/viewtopic.php?f=8&t=807&p=3059 the user says it is possible to use two RM-AV2 to control both cameras using a single control. He says to connect all the wires from a control in another. I'm afraid it does not work.
My PC is Core i7 2600 with 16GB of RAM. I haven't any trouble to render my videos 60p. I usually create my videos MP4 60p and MVC 24p. For example, the video I posted here has 19min and took about 3 hours to render.

I'm sorry my bad English.

Milton

Trick- I have a pretty good metal shop in my garage. I have a metal band saw, a drill press with a x slide vise and set of end mills. The metal band saw is equipped with a diamond tooth blade. It cuts through most metal like butter. I also have a diamond cutting disk for the drill press too. Right tool for the job.

Powerplay4-

The audio sync method is quite adequate for near frame sync. I've used it here as well. The main advantage I see with the Lanc Shepherd is having zoom sync capability.

You are correct that simply paralleling the wires won't work. You would need to add two buffer amps to isolate the cameras and set one as master to slave the other. The sync is done at the moment of power up as that begins the clock function for the digital switching. There is also a 0-4 frame slop in the sync of record start and stop too. So, even with Lanc Shepherd you need sound sync. I think the 0-4 frame slop in record start has more to do with the master / slave errors in the TD10 camcorder since Wolfgang has the same error as I do and we use different controllers.

I have had no problems with sound syncing my clips 3 minutes to 15 minutes in length using a clapboard for 1080 60p and no controller. I make a sound and visual sync a few seconds after rolling video and again just a few seconds before pausing. The part frame missync is pretty close to the same even after 15 minutes of recording. It's not genlock but close enough for my amateur work.

Don, in Austria it is possible to order the igus parts in every length you want them to be, up to 400 cm I think. You pay a little bit for that - but maybe that is not possible in the states if you order them from Amazon.com.


Same findings here: I used the white balance automatic when shooting on the sea some sea-lions - and ended up with a terrible difference in the colors. Ok, can be corrected in the post, but I have also learned that it is a nice idea to work with fixed white balance. And sequential serial numbers are great! That makes it very very likely that the optical and sync drift behaviour is very similar.

Sure, with 60p the difference between frames is small. But I would tend to use such a controller if possible, since they allow to see if the readout from the chip starts synchrone. That is something different what Don mentions. The start point of recording of the two AVCHD cameras when you start to shoot differs always by something like 0-4 frames, I think it is because we are talking about consumer cameras here and not a professional Genlock. The difference in the start point of recording has to be adjusted in the postpro, what can be done in Vegas in a nice way. And yes, for the synchronisation - so the differene in start of recording - I have purchased a clicker. Cheap nice solution, you know such a part that they use for pet trainings, and that makes it easier to see the correct point in wave form in the timeline.


But even if one camera starts 3 or 4 frames later, the second question is if the startpoint of the readout of the chips of both cameras is similar. Without that, it may happen that one camera starts at line zeor, while the other is in the mid of the chip. A difference in startpoint of readout of the CMOS chips is smaller then the startpoint of recording - at the worst it is the length of a frame, sure. BUT you never ever can correct that again, if you have not got that right during shooting. That is why it makes sense to use controller for the sync drift, to see where you are.

I use a ste-fra lance, but it is not cheaper really: http://www.digi-dat.de/produkte/index_eng.html#stefraLANC

For sure, with 50p or 60p it is less worse compared with 24 or 25p. And if your motives tend to be static, it is less an issue. I understood it is expensive at Brazil.

My question was more about the playback behaviour of two 1080 60p clips, not about rendering. With two 1080 50p clips I see with my i7 2600K 4.4. Ghz and 16 GB ram a dropdown in the preview speed to something about 30-40 fps (100% would be 50 fps), with preview settings at "preview/full". We had the idea to the the project to 720 50p or 720 60p to overcome that, what works fine in Vegas too. Sure, you also could go to "preview/half resolution" in the preview.

Another question: how satisfied are you with the mechanical adjustment of the rig? Do you see a strong difference in vertical adjustment, rotation, keystone, zoom?

Wolfgang, in fact the mp4 file is SBS (side by side) 1080p 60fps, not 1080p 60fps twice. I didn't render 2160x1080p60. The Blu-ray MVC is 1080p 24fps. In this case I don't have problems to playback. I never tried to render and play SBS full.

About the mechanical adjustment of the rig, I don't have problem. The vertical adjustment for me is not necessary because is always aligned horizontaly. I don't use zoom yet because, as I said, I don't have remote control yet. I'm still learning.

Milton

Milton,

your English is not so bad, and I am also not a native speaker.

Ok, you use sbs? How do you get that from shooting with two CX130?

I had the idea that you have to 2D streams in AVCHD 1080 60p, that you pair in Vegas. I shoot two 1080 50p streams and there I saw the playback issues.

Yes, I use SBS 1080 60p (mp4) and Blu-ray MVC 24p (iso file). I don't have issues to shooting with two CX130.
Yes, I pair two streams 1080 60p in Vegas timeline and render MP4 and MVC. I saw with no issues from my computer plugged with GeForce GT-430 HDMI 1.4 on my Panasonic TV 58VT20. The TV, display 1080 60p 3D when I use Media Player Classic Home Cinema. In this case, I need to set manually SBS on TV. When I use PowerDVD 11, the TV display 1080 24p 3D only, for both mp4 60p and Blu-ray MVC 24p (automatically).

Milton

Today, I completed my modifications to my SBS rig to get ready for a two camera 3D shoot project. It will use a single TD10 in the center with it's own pan tilt head and two NEX 5n cameras with fisheye wide angle lenses ( 130° capture angle). I started testing the shoot settings to determine the best frame rates for the NEX 5n to match against the TD10's fixed 60i rate. I'd like to use 60p on the NEX5n's but experiment will determine how good that is, otherwise I can switch the NEX5n's to 60i, same as the TD10.

Nice pool..... And nice rig off course.

Milton, if you shoot with your camcorder, you will have 1080 60p footage from each of the camcorders. To my opinion, you bring this two files to the timeline, pair them - and then you render that to side-by-side half maybe. But the camcorders will produce 1080 60p with 1920x1080 - each of them.