Moire, ah man.

D

DM_rider

Veteran
Be cautious, this is the first serious moire I've ran into with the camera. Figured it had to happen sooner or later. I wonder if it could be tamed within the detail settings?

Download the file to really see it
http://vimeo.com/30442889
password: test
 
Last edited:
Yep. That's one of the first things I found on it, it has the same kind of rainbow moire that the Canons do, but only on really fine detail. The Canons show it everywhere, but on the Sony it's only on the finest details, but if you have really fine patterns or bricks or something like the fine patterns you were showing here, it can happen...
 
Yeah I've encountered it w/ mine, particularly w/ screen doors and old vintage microphones. Not nearly as often however as when I shoot w/ the Canon 5D.
 
Barry_Green said:
Yep. That's one of the first things I found on it, it has the same kind of rainbow moire that the Canons do, but only on really fine detail. The Canons show it everywhere, but on the Sony it's only on the finest details, but if you have really fine patterns or bricks or something like the fine patterns you were showing here, it can happen...
Click to expand...

I had forgotten about the issue, because overall, it's fairly decent with detail and I hadn't run into any serious moire yet. Surprised me more or less.
 
Well I really wouldn't expect moiré from this low MP sensor. it surprised me too. I'm curious to know if the F3 would do the same
 
Moiré happens when two repeating patterns are overlaid, and are fairly close in spacing of their elements / angular alignments. Right now I'm looking at an office chair whose back is two layers of a mesh fabric wrapped around a tubular frame, and I'm seeing moiré. (With my eyes, not a camera... :-) )

The pixel grid in your camera's sensor is a pattern; so if you shoot another pattern with similar density, you'll get moiré. As far as I know, this is a fact of life with digital cameras; but because the pattern you're shooting has to be really fine for it to happen, it's not an issue most of the time.

The problem with the still cameras (5D etc) is that they downscale a high-res image by line skipping, which introduces a third, and coarser pattern (because as I understand they don't line-skip to exactly the right number of lines). So, they have MUCH worse moiré problems. That's where the FS100 scores in having a sensor designed for video. But moiré is always a possibility, and I guess we just need to work around it. The FS100 just makes that easier.

I gotta say, though, I was surprised how much there was in DM_rider's crop patterns.
 
AtticusLake said:
The pixel grid in your camera's sensor is a pattern; so if you shoot another pattern with similar density, you'll get moiré. As far as I know, this is a fact of life with digital cameras; but because the pattern you're shooting has to be really fine for it to happen, it's not an issue most of the time.
Click to expand...

I thought that digicams had a filter that should blur the image to a point where the resolution of the image is just lower than the resolution of the sensor

At that point the patterns in the image should never be close to or equal to the the pattern of the sensor

So it (moiree) should not happen with a properly designed device??

Obvioulsly still/motion cameras (eg 5d2) this woud not work becuase the operate at two different resolutions

But moire should not happen in a correctly specced single purpose camera??

Barry can expand no doubt

S
 
Well, the tuning of an OLPF (optical low pass filter) is a balance of compromises. Aliasing happens when the lens passes too much sharpness to the sensor. The OLPF is there inbetween to dumb down the sharpness to a point where the sensor can properly resolve it. But how aggressive do you make the OLPF? Tune it too far down and you'll eliminate all aliasing, but probably some legitimate sharpness too. Go too weak with it, and you run the risk of getting aliasing patterns on your footage. In Alan Roberts' evaluation of the F3, he questioned whether there was an OLPF in there at all. Panasonic tuned their OLPF to eliminate all moire issues, to the point where some folks say the GH2 looks like it has "sharper" video... which then gets back to the whole question of aliasing making things look "sharper" than they actually are.

Alister Chapman found moire patterns in bricks on his F3 early on, and I found the same thing with the FS100, and now DM_Rider found it in grain patterns. Traditionally Sony has struck me as a bit more aggressive than other companies, they balanced the EX1 to get the most sensitivity, and ran too far into allowing infrared and ended up with the IR contamination issue. And it seems to me that's what they've probably done here too, they were too aggressive with their OLPF (meaning, they erred on the side of chasing maximum sharpness).

What DM_Rider's footage didn't show is the rainbow pattern that happens. Shoot some bricks from a good distance and you can get the same purple/orange type of thing happening on the FS100 & F3 that happens on the Canon DSLRs (but to a much, much milder degree, obviously). I haven't had a chance to compare an F3 against an FS100 to see if the issue happens to the same degree or not, but comparing Alister's shot against mine, they do look comparable.

Anyway, once again, it's all a balancing act, just like in a car engine -- you can tune a car for maximum power, or maximum gas mileage, but usually not both. With a single-chip sensor you can tune it for maximum resolution, or maximum fidelity of avoiding aliasing, but not both. So each company makes their choices and lives with the consequences of those choices. In the case of the F3 and FS100, it is possible to get the same kind of aliasing and rainbow moire that happens on a DSLR, but to a much less prevalent degree.
 
  • Like
Reactions: SPZ
Barry, I remember on another thread you said the FS100 hasn't really got a low-MP sensor, from the info you gathered it really looked like a high-MP sensor (like in a DSLR).

Is that so or I'm getting it backwards?
 
alex-

did you spot the moire when you were shooting or only on later review?

can you recall the detail level set for that crop shot?

would lessening the detail level be the work-around?

again, thanks for making the effort and sharing your observations.

it might be worth it to repeat the obvious, "every camera is a compromise."

ymmv

be well

rob
smalltalk productions
 
Barry_Green said:
Well, the tuning of an OLPF (optical low pass filter) is a balance of compromises...
Click to expand...

Thanks for that very helpful post! I hadn't realised there were such differences in OLPF tuning between cameras.

Am I right in thinking that the OLPF on a stills camera basically does nothing for video, as it's passing everything below the camera's (much higher) stills resolution?
 
F3 will definitly do it.

Happens rarely, but it does happen. By product of being so sharp. Throw a HD soft filter to help calm the effect.
 
AtticusLake said:
Am I right in thinking that the OLPF on a stills camera basically does nothing for video, as it's passing everything below the camera's (much higher) stills resolution?
Click to expand...
Yes, that's exactly the case, and that's why a combo stills/video camera will have such aliasing problems as we see on the DSLRs. The only way to really resolve it would be to take the approach Red has taken, which is to make stills and video the exact same resolution, and that way the OLPF can be optimized for the sensor res. But when you have a case of 21 megapixel stills and 2 megapixel video... well, yeah, something's gotta give. And inherently it's got to be video performance that suffers, because what would be the point of having a 21-megapixel still camera that has a blur filter to lower the res down to 2-megapixel video?
 
Bummer! I was real set on an fs100 for deep, sharp focus. I knew it technically existed, but based on what I've seen and what everyone has told me, its virtually nonexistent. Maybe an EX1 is better for landscapes after all? Hmm...
 
You must log in or register to reply here.
Back
Top