philber wrote:
I seem to recall that this question was asked from their head honcho, Maeda-san, in his DPR Photokina interview, and he came out very clearly saying that a good AA filter is required to avoid moire. His answer did not suggest that technology would lead to this filter becoming less and less of a requirement
That's a very interesting area. The Leica M8 doesn't have one from what I remember. It's why, despite the reletively low resolution, it's so sharp.
But, moire is a bit of a problem.
I've always suspected that as a weaker filter is needed the higher the resolution is, one may not be needed at all at some point.
There is a theoretical minimum size for a light bundle to be. At that point, diffraction prevents a smaller "dot". That's the maximum resolution possible. At that time, the AA filter won't be needed.
Regarding the two sample clips on DPR... I can't play them too well on my PC (which is a laptop), so I've converted them to play on my PlayStation3 on a 46" Sony XBR full HD panel. Seems the most appropriate way for me to play HD clips. (and besides, I hate QuickTime with a passion, every time I've installed QT on a PC the last 14 years, something's screwed up, and it's such a resource hog). I have noticed a dodgy frame in each of the two videos, which I have checked-out further frame-by-frame. I'm not sure if it's occurring in my transcoding to MP4 video (H264 Baseline v4.0) - I don't see these glitches in my transcoded Reverie - can anybody else confirm they are seeing these errors in the original .MOV file?
For the cyclist video, there's a shop with a sky-blue facia and a traffic light in front of it; the glitch occurs when the cyclist's position almost aligns with that distant traffic light in front of the blue shop.
For the market video, towards the end, there are two ladies walking towards the camera (dark blue top with pink shirt, and one with jeans and a caramel coloured top), and a guy with a suit and a black backpack walks past them away from the camera. When his position obscures the lady with jeans and caramel top, the glitch occurs. I noticed the problem in this clip first, there was a dark blip during playback that caught my eye.
The glitchy frames in both of the files is sort-of monochromatic, with some spurious cyan and magenta tinges. There are also subtle but noticeable cyan and magenta horizontal bands across the entire image. But every other frame on the videos looks great, and the colours are fantastic (especially reds!).
I noticed them too. Hopefully it's not a camera bug that makes the production release. That'd be suckage. Obviously Vincent Laforet didn't seem to have the issue, so who knows why it's happening.
Interesting! Thanks for replying. Perhaps a couple of dodgy bits on the memory card used by DPR, or some memory card incompatability with the 5D Mk II.
johnnydanger wrote:
I noticed them too. Hopefully it's not a camera bug that makes the production release. That'd be suckage. Obviously Vincent Laforet didn't seem to have the issue, so who knows why it's happening.
+1
It looks like it's on a 'Program AE' mode for video, and that it's exposing up a third and back in the bright frames. I'm guessing that Laforet shot in full manual.
It looks like it's on a 'Program AE' mode for video, and that it's exposing up a third and back in the bright frames. I'm guessing that Laforet shot in full manual.
I believe Laforet said he didn't know how to lock exposure at the time of shooting his video . (Which he later learned could be done with the "lock exposure" * button). There is no manual mode, but the exposure lock and exposure compensation is a good work around.
Thanks for your examples, Romy. I plan on using the 5D2 for fine art prints and web/digital presentation images and vids. Your sample vids/stills are very encouraging for my web plans. Now, as a 5D lover, I hope the 5D2's full res files live up to the image quality of the original 5D. We'll see...
dcmiller wrote:
I believe Laforet said he didn't know how to lock exposure at the time of shooting his video . (Which he later learned could be done with the "lock exposure" * button).
Laforet is a genius, but I can't help but laugh at your comment. :-)
Excerpt:
In addition to the basic settings, photographers can also select and adjust the following camera settings that are supported in Movie mode:
Picture Style
White Balance AE Lock
Exposure Compensation
Peripheral Illumination Correction
Auto Lighting Optimizer
Highlight Tone Priority
...the following settings are made automatically by the camera and cannot be adjusted by the user: ISO speed
Shutter speed (from 1/30 to 1/125)
Aperture (varies according to lens; no arbitrary restrictions)
Color space (sRGB)
Exposure metering pattern (Evaluative)
Mel Gross wrote:
That's a very interesting area. The Leica M8 doesn't have one from what I remember. It's why, despite the reletively low resolution, it's so sharp.
But, moire is a bit of a problem.
I've always suspected that as a weaker filter is needed the higher the resolution is, one may not be needed at all at some point.
There is a theoretical minimum size for a light bundle to be. At that point, diffraction prevents a smaller "dot". That's the maximum resolution possible. At that time, the AA filter won't be needed.
I think there's a significant difference between CCD and CMOS when it comes to the necessity of an AA filter to mitigate moire.
Along with the M8 (CCD), all MFDB (medium format digital backs) are CCD-based, and don't have AA filters. Moire seems to be managed in those MFDBs. I'm not sure how - possibly by the raw processing in post.
The M8 is vulnerable to moire patterns, but as others have pointed out, the lack of an AA filter means that you get exceptional pixel sharpness, giving the 10MP images a lot of headroom for interpolation to larger sizes.
I don't think it's so simple for Canon as to just remove or reduce the AA filter, since their working with CMOS technology, which is fundamentally very different from CCD.
Excerpt:
In addition to the basic settings, photographers can also select and adjust the following camera settings that are supported in Movie mode:
Picture Style
White Balance AE Lock
Exposure Compensation
Peripheral Illumination Correction
Auto Lighting Optimizer
Highlight Tone Priority
...the following settings are made automatically by the camera and cannot be adjusted by the user: ISO speed
Shutter speed (from 1/30 to 1/125)
Aperture (varies according to lens; no arbitrary restrictions)
Color space (sRGB)
Exposure metering pattern (Evaluative)
It does say aperture, but that is wrong. I think it's mixed up with not being able to shoot fully manual.
Romy can probably verify what can be set.
Added:
From Laforet:
......there is TOTAL control of Aperture - period. It’s called Aperture priority mode - you set it - the camera does the rest. Program mode, and shutter speed mode are different modes. Just wanted to lay this to rest. I set my cameras to either 2.0 or 2.8 for the almost the entire shoot - and the aperture never changed. We shot it in bright daylight - and that time I set it to 4.0 - and it stayed there.
.....So with Av priority in video mode, you can mount legacy alternative glass and control aperture on the lens. You would then change the ISO to control the shutter speeds within the 1/30 to 1/125 limits. Given that 1/125 is the fastest video shutter speed, what is the longest lens that people could recommend hand-holding? I realise that the best qulity will always be acheived on a tripod.
Would 200mm be possible using a monopod?
(I know nothing about the intricacies of video, but have a feeling I am about to learn a lot over the next couple of years).
I wouldn't go anything longer than 85mm without IS- at that point you may see some moderate vibration. Many movies are shot with a 35mm equiv of 50mm or less, and close-ups with up to 150mm (these are guaranteed to be on a tripod). It's not so much the shutter speed of the video that will matter (the shutter speed won't eliminate shake because frames are shot consecutively. Even though one frame could be shake-free with a high shutter speed, like in a photo, it won't be conjunct with the next frame. In other words, try handholding an 85mm lens on FF and see how steady you can hold it while looking through the viewfinder. Picture that as a video- then think about whether or not you would like to watch a video with that amount of positioning inconsistency. With the IS it'd be easier to handhold.
Stu Warner wrote:
..... then change the ISO to control the shutter speeds within the 1/30 to 1/125 limits ...
What video frame rate do you get with 1/30 shutter speed?
