What we have, what do we need

[J.D. Frey]

Howdy,

Our gear list is below, and I would like to shoot some rudimentary 3D shots. Not looking to pull focus or do anything super fancy (because you know 3D is so easy to do these days ). I'd like to just see if it is possible to do some testing with what I have:

1. Red One and several lens options
2. Epic M and several lens options
3. Panasonic 3D mixer (AG-HMX100- HDSDI in and mix to HDSDI L/R 3D out)
4. Marshall 3DR-HDSDI (Signal converter from 3d HDSDI to consumer 3D HDMI)
4. Off the shelf consumer 3D tv and glasses

I would like to use the Red One and Epic for binocular acquisition, then monitor the footage live using the other equipment. We would then post the two Redcode streams in AE/Premiere. My main questions is do you think it's possible to physically get the RedOne and Epic mounted in a way they can be used together (i.e. is there leeway regarding mounting distance etc.). Secondly is it a pipe dream to think that my monitoring streams would actually work and look good enough to monitor on set (This is important since the EVF for the Epic M is not widely available and because we happen to have the panasonic mixer)?

I'm planning on just getting out there and trying it anyway, but wanted to know if anyone sees any potential pitfalls.

Thanks!

 

[Keith Putnam]

You need a way to precisely match the field of view of both cameras (and also allow for horizontal offset to set IO), and ensure that both sensor planes are precisely the same distance from the subject for matched focus. Most often this is done with a beamsplitter camera rig, though side-by-side is also possible (with attendant alignment problems). What is your plan for mounting the cameras?

 

[J.D. Frey]

I was thinking of having something machined and hard mount after picking a lens set. I'm assuming if you get close then you need one axis for convergence and some method of measuring?

It's not impossible to think we would eventually upgrade the Red One to the Epic X (but I really really love the Red One), so in the future I guess we could go with one of the many packages that seem to be available.

Really I was just wondering if I could get two images on the mixer and then look at the monitor and play with different set-ups in the studio to "see what it looks like". I've never done anything remotely close to 3D and I'd like to play around with it- so I don't need a rig that is portable, or really even useful on location.

 

[Jason Goodman]

You need to get two of the same cameras. The EPIC uses a larger area of the imager than the RED ONE so even with the same lens on each camera you will get a different field of view and your 3D will be screwed up.

 

[J.D. Frey]

You need to get two of the same cameras. The EPIC uses a larger area of the imager than the RED ONE so even with the same lens on each camera you will get a different field of view and your 3D will be screwed up.

I was thinking that zoom lenses set at different distances might work. Sounds like we need to upgrade if we want 3D- although I'm guessing that the Epic-X and Epic-M are different as well. Maybe time for a rental unless X and M are the same size and image the same lens area.

Thanks for all the help by the way.

 

[Doug Horton]

The FOV is the same if you set the EPIC to 4k HD, as well as the R1 MX. An important hurdle is ensuring that since you're mounting the cameras parallel, that your lens height matches perfectly between the two cameras. After that you can worry about getting your alignment, which using a set of matched primes would make this primitive shooting experiment a lot easier on yourself rather than guessing and second guessing matching focal lengths on zooms.

As a side note, two separate zooms of the same manufacturer, can still have different wide/tele end stops, meaning, you can use two 16-42's, and put them both on 16 just to make it easier on yourself, however their FOV can have a disparity of 5 microns or more, so even though you think by setting them both to 16mm and you're lenses are matched, you're actually just chasing your tail because of the false security you gave yourself.

Not trying to make 3D sound super frustrating or complicated, it's just these little things you don't know at first until you start running into problems, then you have to begin digging for answers/ways to resolve these issues.

 

[Jason Goodman]

It is a valid and good point that setting each camera to the same format would overcome the inherent FOV differences. However, as mentioned, the lens height is not the same on the EPIC as it is on the RED ONE, and many other aspects of each camera differ as well. No matter how smarty pants you want to be about it, you really should just get two of the same cameras. There are enough hoops to jump through getting two identical units to work together, why set yourself up for failure by starting out with two different cameras?

Really for the best results, I would go as far as saying you should seek out cameras with sequential serial numbers and equal operating hours. At very least have the same software build on each. Maybe I'm superstitious but I like to try to stay in front of the 8 ball and stack the odds in my favor as much as I can.

I must also disagree with the assertion that primes would be easier to match than zooms. A slight FOV difference in primes (there always will be) is something you can only deal with in post and can frustrate an inexperienced stereographer during alignment. Ultimately, mismatched primes will give you less precise alignment and lower image quality at the end of the line. If you are renting primes, keep in mind lenses of differing ages or stored / used in differing climates may exhibit differences not only in terms of FOV, but also contrast, color, blooming and other aspects that are affected by the lens coating and other factors.

Forgetting about the focal length markers, it is quite easy to nudge a zoom lens one way or the other to precisely match the FOV of two factory matched zooms. Many hardware and software tools exist that can help you in this process if you do not trust your own (or your stereographer's) eye. Angenieux does a very good job of factory matching their zoom lenses, not just in terms of focal length, but aperture and focus as well. The added advantage of zooms for 3D is the time savings of not having to constantly change lenses. Mere seconds to go to a new focal length with zooms (audition many, then decide). Primes take a few minutes to change out on a 3D rig and require realignment with each change (more minutes). It is also worth noting that purported "automatic" rig alignment systems do not currently work with prime lenses and do not always work faster or better than a skilled human.

21st Century 3D has been in the stereo business for 18 years. All of our cameras and lenses are matched, sequential and have been used for the same number of hours. Our rigs are completely tool free in operation and can be aligned by a skilled rig tech or stereographer in seconds. Our beam splitter mirrors have a custom designed coating that beats all others in the industry in terms of color, brightness, contrast and pola match from left eye to right eye across a wide range of FOVs. These things are critical in terms of matching and many of them cannot be easily addressed in post. Please get in touch to arrange a visit with us in NYC or Hollywood so we can demonstrate all of this.

 

[Doug Horton]

There's probably no concrete answer as to which is better. We have internal arguments, and tests over and over and over as to which is better than the other. An inherent quality of zooms are the exaggerated barrel distortion compared to primes. I'm not dogging Optimo, love the company, love the zooms, have had SO much time with them and constantly learning and changing methods. But in my opinion, which may not be correct, (and ignoring vignetting), is that once you start stretching to bigger and bigger IO's, we'll say 3+ inches, suddenly you have a noticeable barrel distortion overlayed on top of another image with barrel distortion, but converged so this distortion is now giving you what looks like a bad alignment, so you begin compensating, it's a complicated theory/argument. And bumping the zoom to slightly change the FOV isn't going to do anything about barrel distortion. If you've never noticed BD (not you Jason, I know you're the master here), but general public, shoot something you know is flat, take a still from footage, go into photoshop and squeeze it horizontally (makes it easier to see when your eye focuses on a narrow image) and then add a thin ruler and compare it to what was flat in your shot... you'll see a very noticable bend.

It's all a matter of project, resources, what tools you have to align, etc... and as far as "time" is concerned, it again depends on the system and prep. I've changed primes in under two minutes and be back to shooting, which isn't as fast as bumping a microforce, but some DP's love their prime lenses, and I will never lead them to believe that it'll slow them down if I can have lenses changed in almost the same time as it would take to boot up a R1. Conversely, we're usually on zooms and I enjoy that because at that point it's all about the prep and tracking them to bodies perfectly, and at that point I just sit back and wait for cables to fail or power to be pulled.

 

[Lucas Wilson]

It is also worth noting that purported "automatic" rig alignment systems do not currently work with prime lenses and do not always work faster or better than a skilled human.

Not quite. 3ality Digital's Intellecal system has been working with our beta customers on live productions for a little while now. Prime alignment is about 3 minutes. Zooms usually about 5 minutes.

And to my knowledge, 3ality Digital's rigs are the only ones currently in the industry that contain a robotic (closed loop feedback) height-pitch unit, thereby correcting vertical misalignment at acquisition... meaning it's the only system on the market that can accurately track through a zoom, and actually use a zoom as a variable prime without having to stop production for a realignment.

Best,

Lucas

 

[Jason Goodman]

We need to decide what this conversation is about. JD started the thread saying he was planning his first ever 3D test and asked what the "easiest" way to do it would be. Doug suggested he use primes. Fnding two matched prime lenses for rental is not "easy". In fact, unless you are mounting a major production with a mountain of cash behind it, doing this may not be possible. The best way to match primes is to purchase new and work with the manufacturer to ensure all possible steps are taken to match all optical characteristics as closely as they can.

Fact is, focal length changes are EASIER with zooms than primes. This is sort of like saying water is wet, an obvious truth. Focal length change on a zoom takes one AC one second. Changing primes takes one or two or three people 2 or 3 or 5 minutes, perhaps more. Change 5 - 10 times per day and you could be adding 30 minutes to an hour to your schedule instead of 12 seconds. Zooms are EASIER for 3D. Depending on what you need to do and how you need to do it, easier may or may not be better, but it is definitely easier.

Please explain how a wider IO would increase barrel distortion? I have not seen this and do not follow the logic. If you are at 3" and a steep angle of convergence you may be introducing depth plane curvature which is not the same as barrel distortion and has nothing to do with shooting zooms or primes. Ideas are being mixed up here.

I still recommend someone starting out in 3D should get two of the same cameras and two of the same lenses. To make your initial tests EASIER, I'd suggest zooms. I would not attempt to "tell" a DP what type of lens to use. I would merely inform him about aspects of 3D with which he may not be familiar so he can form his own opinion.

There is no doubt that great prime lenses look great. If a FOV difference exists between the two you're using for 3D, there is nothing you can do to address that properly during capture and preserve your full resolution. With zooms, it is easily handled. I brought up the image quality issue only because mismatched primes create a need for scaling or other types of correction in post and this may damage image quality, negating any inherent optical advantage the primes may have had. That is not necessarily obvious and potentially even counter intuitive.

I apologize if I misspoke with regard to the current capabilities of 3ality's systems. When I last saw it only a few weeks ago, primes were out for auto align. As for automatic alignment taking between 3 and 5 minutes, a skilled human on a capable rig can do it much faster. If this auto procedure must happen each time a lens is changed, that's 2 - 3 minutes for the change and another 3 to 5 for alignment. Zooms and humans are faster and easier.

If we want to debate the inherent quality of zooms vs primes, that's a different discussion.

 

[Lucas Wilson]

I apologize if I misspoke with regard to the current capabilities of 3ality's systems. When I last saw it only a few weeks ago, primes were out for auto align. As for automatic alignment taking between 3 and 5 minutes, a skilled human on a capable rig can do it much faster. If this auto procedure must happen each time a lens is changed, that's 2 - 3 minutes for the change and another 3 to 5 for alignment. Zooms and humans are faster and easier.

Hey Jason - I heard you're at the Sony class this week learning all about Stereoscopy and 3ality Digital gear... I expect you to be able to do it in 2 minute after you learn about our tech. ; )

Lucas