The "What next Camera to get", Part 1: 3CCD vs CMOS (aka: rolling shutter)

[John Nantz]

Of the subject areas I want to make videos of has to do with sailing and boating in general. The videos would contain action - action that will, to a lesser or greater extent, result in the rolling shutter artifact. One of the major decisions in the next video camera acquisition is to determine how to weigh the importance of rolling shutter avoidance. Or as Shakespeare might of said, "To roll or not to roll".

Nearly all video cameras today have CMOS, or 3MOS in some cases, sensors. There are still some 3CCD cameras out there, both current models and a slew of older models.

The problem I have is how to weigh the trade-off value of 3CCD vs CMOS in trying to avoid rolling shutter artifacts.

The video examples tend to show things like trains going past, or the camera going past (or panning) buildings at right angles. These tend to be extreme cases, but in the real world of video one will experience something less dramatic. If a CMOS camera is being used then it would behove one to avoid the obvious bad situations and camera angles, but this isn't always possible, and you have to think about it (think? Uh-oh!). Think, as in pan slower or change the angle, for example.

Okay, so how bad might rolling shutter be with something like wave action? Or a video of a boat passing close by?

Are there applications (easy to use) for removing rolling shutter? There is an article about Google doing rolling shutter stabilization for Youtube uploads form 2012: Image Sensors World: Video Stabilization and Rolling Shutter Artifacts Removal on Youtube

One reason I'm asking this now is because I'm considering getting a 3CCD camera instead of one with CMOS.

Sample video with wave action taken with an iPhone. How would rolling shutter impact something like this?

20130528 Boat return from PT on Vimeo

This one (below), not really a good example, was taken with a 3CCD camera:
RVYC Juniors. Would CMOS have made anything worse?

0:30 boat going past at right angle - too fast for CMOS?
1:05 boats gong past - no panning- too fast for CMOS?
3:55 pan too fast for CMOS
5:10 ditto-- too fast for CMOS?
5:35 note lots of vertical lines (masts - typical of what might be in a boating video) - need to be careful
6:52 boats going past - too fast for CMOS?
7:31 VERY fast pan followed by fast moving background - too fast for CMOS?
8:00 more action - would this be too much - too fast for CMOS?

RVYC Jr Sailing Program (Beta) on Vimeo

Note: there is a jello part about half-way through that should not have had stabilization applied. Forgot to take it out.

I'm sorry I don't have any better examples but these are my best "typical" ones.

There are a number of other criteria that have to be weighed but the rolling shutter issue is a main concern.

[Gary Huff]

Originally Posted by John Nantz

Sample video with wave action taken with an iPhone. How would rolling shutter impact something like this?

Your iPhone has a CMOS sensor in it and has some of the worst rolling shutter issues. Any proper CMOS-based video camera will be far more resilient. Is there anything in that iPhone video that you do not feel is acceptable for your line of work?

[John Nantz]

Gary - I see what you're saying, and I guess the answer to the "... Is there anything in that iPhone video that you do not feel is acceptable ..." is probably "no". And the reason may be because everything is moving around so much one can't tell if there's an artifact. It's different than the train/car/building going past because the waves have no straight edges to them.

I've been doing a LOT of internet reading since I posted the questions and I'm gradually coming to the conclusion that CMOS, for the application with boats, may not be all that bad, especially if one is careful with how the camera is used. In the earlier years - '07 to around 2012 say, there were a lot of questions being asked by others about 3CCD vs CMOS.

Smart phones have really helped with improving CMOS technology by funding R&D so the consumers of videocameras have greatly benefited.

CCD vs CMOS - Difference and Comparison | Diffen
Has a table of comparisons

What will help is to have applications that will remove rolling shutter artifacts.

[Dave Blackhurst]

Other than that bit at 7:31, I don't think you'll have a problem with CMOS - keep your pans slow and deliberate and it's really not as huge an issue as you might think if you rely on the internet...

Just because someone can swing a camera wildly about and cause nasty distortions doesn't make them a "camera operator", nor a "proper" camera reviewer. Yes, there is a time delay when reading out a CMOS sensor. That does not make it "defective" or offset the advantages...

This isn't to say you shouldn't be aware of Rolling Shutter and the potential for skew, it just means use reasonable camera technique, and clip out the parts where you don't...

[Brian Drysdale]

How important rolling shutter is depends on what you're shooting with it, on static shots it won't affect things, while on an action film with fast camera moves in an urban environment it can be annoying. The level of rolling shutter effect varies from camera to camera, so you should test to see or compare the levels in tests done by the more technical reviewers.

[John Nantz]

Gary and Dave: I really like your answers! Frankly, that is what I was hoping for. Not looking for perfection, just good enough. People watching this will realize the video wasn't from a Hollywood set.

Brian: What I'm finding is that in the past few years is that the manufacturers have been trying to come up with solutions to the problem and while not necessarily a 100% solution, say, there has been progress. Finding technical reviews and tests on recent cameras in this subject area has been difficult, though not impossible. Usually the reviews are for the new model release and then the manufacturer comes out with version updates that don't get reviewed.

There has been progress on the shooters end, too, learning what or how to shoot in a certain situation.


My thoughts now:
Rather than take the shotgun approach with the question of "what camera to get next", the plan was to break the analysis, or question, down into parts. The 3CCD vs CMOS, or vs "3MOS", seemed like a good place to start. Unfortunately, the deeper I get into this the more questions and problems crop up and this is really bogging me down. Budget-wise, CMOS with a global shutter is probably not an option.

One of the problems is that within any kind of budget range, in today's market, and even considering used, the CCD option severely limits one's choices - and not just by a little bit.

Smear, Skew, Wobble, and Partial Exposure:
The thought now is whether going CMOS is a deal breaker and while all the other camcorder selection options haven't been weighed yet, it doesn't seem like it will be. Workarounds could be to use faster scanning to reduce skews and then there is just the plain "watching what gets captured in the scene."

For boating-related shoots with moving targets (boats and waves) and exposure extremes (lights from sparkling water to darks), a lot of what gets captured is going to fall into the realm of "what you see is what you get." I'll just have to try and be in the best location possible at the time and shoot on the fly. There will never be a "Take 2" option like on a Hollywood set. The really annoying stuff will just have to be edited out.

As for other motion stability, there may be some camera improvements that will help. For example, looking into this further I learned that the Canon XF200 has a newer 5-axis stabilizer that makes for very smooth operation but while it is not something that reduces or has an affect on "rolling shutter" it could help in my situation. Other cameras also have built-in IS too, both optical and software. Having built-in IS will help the operator to think about other aspects of the shoot, including things that will cause artifacts.

While I have a GlideCam which was the hopeful solution for this purpose, operating it with a camcorder from the moving and yawing deck of a boat isn't as easy as one might think. As an aside, the IS on the 40x Fuji Techno stab binoculars is really impressive.

I'm putting a lot of effort into the camera selection because I like multi-cam and the idea is to get cameras that will work together image-wise. Plus, I'm not fond of spending a few thousand on something only to discover it does't really do what I need.

Open for more input but I'm thinking about what the next important criteria should be to narrow down the field.

[Dave Blackhurst]

Not sure a steadicam or similar is the best answer... take a look at the Sonys from 2-3 model years back with the BOSS "magic eyeball" stabilization - compact, easy to shoot with, consistent results with very little effort. Probably the best "solution" to stable shooting which seems to be the primary concern (understandable when shooting from a moving object at moving objects).

Model #'s to look at are CX760, PJ710 (no VF), PJ760, PJ790, and the "pro" version of the 760, the NX30. Single 1/3" (roughly) CMOS, some manual control, good auto, and should be available under/around the $1K price point IF you can find them, or a little over that for the 790, which is closed out but still available "new". The NX30 is still current, and around $2K


I shot a bit last week with the AX100, and found with some tweaking of the shutter speed and careful camera work, skew was not a huge problem, even though the camera has been criticized for RS/skew - it CAN be bad, but manual settings helped a surprising amount (somewhat specific to this model and 4K).

Good camera work ALWAYS is helpful, but that "magic eyeball" is like an internal steadicam... sorta hard to beat!

[John Nantz]

Dave: Thanks for coming back again with some more good info!

You know, a couple months ago (I'm doing this research in spurts) I read about the "magic eyeball" but didn't dig any deeper into it's capabilities. Went back and read some more about it just now and for the most part there are good reviews (Amazon) but peppered with a couple that said it didn't work well for them. For me on the boat, though, something that can help with motion and isn't overly large is a plus.

What is difficult to describe is how hard it is to shoot while trying to keep ones balance in a cockpit or on a deck that is sloped 30º, maybe wet, where the anti-skid is a misnomer and wishful thinking, and what you're holding onto with one hand (always one hand for the boat!) is leaching and yawing while pounding into waves and swells of unequal size. All this and then trying to look at a small screen with high ambient light, keep the horizon level, and roll with "the punches" (waves and swells). Using a viewfinder borders on the impossible because it's easy to loose your own balance.

Trying to do this using something like a GlideCam with the thumb or finger on the base, pointing the camera at what you want to shoot, is difficult, and I've already tried that. The pendulum can easily be hitting or snagging on things. This is not easy.

In the case of the iPhone video (Post #1), my camcorder was at home, the weather forecast was benign, my crew (wife) took the car and drove home while I went solo. The sea conditions were nothing like forecast and I wanted to show her what I had to go through (yes, seeking sympathy!). Shortly after the above shot the sea state got so bad the Autopilot couldn't steer the boat and I had to take the wheel. Heavy spray was coming over the dodger and it was like someone on the bow throwing buckets of cold salt water in your face. Wish I could have got a shot of that but didn't want to ruin the iPhone to do it. For those who haven't done this, pictures don't begin to show what it's like.

Anyway, I'm saying all this to explain what will be some of my "like to have" criteria down the road: A camcorder that is kinda mid-size. Built-in optical ND filters, three-rings, lots of button controls (vice touch-screen), 60p, …… Haven't mentioned these yet.

The ones you suggested would fill the bill size-wise. Something like the CX760, if the OIS (Optical IS) works well enough might do it, just for shooting from the deck along with some steading effort on my part. OIS combined with the stabilization option in FCPX could work, hopefully with a minimum of Jello artifacts. Use screw-on ND filters.

Oddly, the Sony HXR-NX30 would work, is a nicer cam, and their used prices seem to be better than the 760. Why would that be? I like your AX100 (but wouldn't use it for this work) because it would be good with FCPX. Don't know if the same is true with the 760 or the NX30.

The real bummer is that this year's sailing season is already half over. Thanks for the suggestions.

[Dave Blackhurst]

The NX30 adds an audio block, but otherwise is more or less the "pro" version of the PJ760. They seem to go at a bigger discount off the list price, probably due to them more expensive up front, but the PJ760 being the same cam... so the PJ holds value.

Just to toss another angle into your equation... Sony has "sports packs" that can be used to make these small cams waterproof up to around 15 ft... so you could put the cam in one, get some sort of clamp mount, and go right on in bad weather! They've had various incarnations of the shells, but if you modify the sleds, most any Sony Handycam can be fitted. They are expensive new, but can be had used for reasonable prices. I've got a couple here that I just realized I won't have any cameras to fit, as I've been cutting out cameras now I'm using the AX100 and RX10!

You don't get super manual control with the Handycams, but they shoot fairly well in auto and you can do a little easy tweaking to get what you're after.

I'm always a little skeptical of online "reviews" - if you look around DVi, you'll find a number of users who swear by the BOSS stabilization, it really is pretty effective. I'm going back and forth on whether I should keep one camera with this feature for handheld use.... I'm going to have to decide soon, as I'm down to one 710 and one 760... slowly selling down the multicam set!