p.13 #1 · Announced: RX1R II with 42MP sensor and EVF
sebboh wrote:
i think the main point of the variable AA filter is so it will function well for video which requires a strong AA filter...
There's a widespread misconception about this because people look at the aliasing resulting from resampling down to video resolutions and confuse that with the sampling going on at the sensor itself.
Due to chroma subsampling, there are only 540 lines of colour information in the typical 1080p video. If you really did have an antialiasing filter suited to that, image quality for stills would be completely destroyed. Of course, it is not done. It is not even done for 4:4:4 with 1080 pixels of colour information.
The correct way to make 1080p from a sensor with more pixels than that is to to fit a curve between the points and interpolate, in much the way we are used to using bicubic splines to resize a still image in Photoshop. This is called "full resolution resampling", and Sony do it in the RX10, which is why I bought that as my video camera. If you do this, the antialiasing filter on the sensor can be and is tuned to the full sensor (stills) resolution.
If you do not do this, as in (I believe) the original RX1 and most or all Canon DSLRs, you can get aliasing as a result of the binning or "line skipping". But this is different from aliasing introduced by the sensel grid on the sensor, and is present despite most of those cameras having antialiasing filters suitable for stills.
(I'm sure it's true that postprocessing 2 hours of video to remove moiré originated at the sensor level is even more painful than fixing it in 100 stills.)
p.13 #3 · Announced: RX1R II with 42MP sensor and EVF
I have been shooting video professionally for 30+ years; and it's possible that all pro video cameras do have AA filters—it was just news to me, is all. And even 1080p is so lo-rez compared to sensor pixel resolution, I had just assumed that AA filters were not relevant. I do understand about the artifacts created by downsampling and the like. I didn't want to derail this thread; just curious. The Panasonic G series makes lovely video, which is why we have four of them. And (@melcat) I iwll look into the RX10; cheers.
p.13 #5 · Announced: RX1R II with 42MP sensor and EVF
@justruss: the ND filter on a button (along with completely silent shutter) are THE reasons I can't sell the X-100s. I honestly don't think a variable ND is needed (though would be useful occasionally) but having a 3-stop ND filter on a button is something that gets used almost daily.
p.13 #6 · Announced: RX1R II with 42MP sensor and EVF
Kit Laughlin wrote:
I have been shooting video professionally for 30+ years; and it's possible that all pro video cameras do have AA filters—it was just news to me, is all.
I'd be very surprised if dedicated video cameras (that read the entire sensor) didn't have AA filters tuned to the pixel pitch of the sensor. For exactly the same reasons that 2 and 8 megapixel DSLRs all had AA filters.
It's possible 8K cameras could get away without an AA filter, but I doubt they'd risk it.
p.13 #7 · Announced: RX1R II with 42MP sensor and EVF
Kit Laughlin wrote:
@justruss: the ND filter on a button (along with completely silent shutter) are THE reasons I can't sell the X-100s. I honestly don't think a variable ND is needed (though would be useful occasionally) but having a 3-stop ND filter on a button is something that gets used almost daily.
Yeah, even a single ND (moderate strength) would be fine. Totally agree. An ND, or WATE-style ND (3 settings) would just be a bonus!
p.13 #8 · Announced: RX1R II with 42MP sensor and EVF
WOW, I certainly fall under the "amnesia" comment - I forget that I've had the RX1 for that amount of time. I did not realize that the pop up EVF was two years after the RX1... still it's a neat innovation. I have looked at the RX100mkIV as a potential RX1 replacement. Then I started reading reviews of the Leica Q - bad move on my part, but this new RX1 is intriguing to say the least. Perhaps my only complaint is the amount of memory that the 42MP sensor will require.
p.13 #9 · Announced: RX1R II with 42MP sensor and EVF
Kit Laughlin wrote:
sebboh arote:
I was not aware of this—can you elaborate? The X-100s does not have one, AFAIK, and while its video is difficult to use, if exposed correctly is perfectly acceptable. I use Panny G6 bodies for all my video, and they do have AA filters, but I was not aware of the requirement for them re. video. TIA.
i don't know about the x-100 specifically, but moire can be a huge problem with video (videographers tell me) due to the low resolution. anecdotally, anytime i take a video of one of the kids walking past certain furniture or a window screen with the rx1 or a7 i see a typical rainbow pattern (i also see all kinds of weird moire patterns on any type of grating when i pan, but that is probably sampling related). i see lots of jaggies in most non dedicated movie cameras, but i assume those are downscaling artifacts.
p.13 #10 · Announced: RX1R II with 42MP sensor and EVF
justruss wrote:
I don't really get this statement. How could Sony have offered this in the beginning?
Thank you for that point. And this doesn't just relate to Sony.
A common statement in photo forums is that "Manufacturer X should have offered this from the beginning." The should part of this is mostly inaccurate — in almost every case the decision to not offer some particular feature or capability was likely more complex that people imagine today, and it probably had an effect on cost to customers (a concern — witness the comments about the cost of the camera under discussion in this thread) and on other factors that would have been affected by the inclusion of the feature.
What posters really mean is that they wish that the more contemporary and advanced features had been available in earlier cameras. Don't we all!? ;-)
justruss wrote:
I've never found any of the Fuji video very good-- it's all jaggies all the time.
But I think Fuji XT has more issues going on than just no AA filter when it comes to video, so it's probably not a great comparison to make since the XT CFA and its de-mosaic process is a huge variable that we can't really untangle.
Fujifilm X-trans cameras are not known for their video, and anyone looking for the best video in a small mirrorless camera will likely want something else. (I write this as a still photographer who loves the Fujifilm system.) It is not an single issue — it is more that generally Fujifilm apparently chose to focus on the quality of still images and to not focus on video, as reflected in certain video format issues, the way video is incorporated into the cameras' interfaces, the software included with the cameras.
Zony_user wrote:
To be honest, I would much rather have a variable ND filter than a variable low-pass filter.
Amen!
- - -
Finally, regarding moire... Having or not having moire/aliasing is not exactly a binary choice. There are a number of situations in which you'll see it from any digital camera, including those with AA filters. With my older cameras, which had AA filters, I encountered it with certain subject including corrugated metal buildings with perspective convergence, certain patterns on things like rooftops, etc. It was a rare event, but it most certainly did happen.
With another of my cameras, a 16 MP AA-filter-free body, so far I have not had to deal with aliasing in images even though this camera is used to shoot urban subjects that are full of potential causes of visible aliasing. Perhaps I've been lucky, perhaps this cameras unusual sensor design reduces the potential for aliasing, perhaps aperture choice has played a role. My other non-AA-filtering camera is a very high MP camera, but so far I have yet to find aliasing in a photograph made with it.
I suppose that if you frequently photograph subjects that you know will generate moire you might want to make sure to have a camera that does AA-filtering, and perhaps also avoid optimal apertures, preferring to generate just a bit of softness to creating visible image aliasing. If you are a typical photographer, you may never or almost never see aliasing, particularly if you don't go searching for it in your images. But when you do, you can most often compensate in post in one of several ways.
p.13 #11 · Announced: RX1R II with 42MP sensor and EVF
@sebboh: thanks; that all makes sense.
In my commercial videos, I usually shoot with primes and close to wide open—so there is enough blurring of any background to mean that moiré has never been a problem for me. As a stills photographer though, recently retired, I have seen it in radiator grills and rooftops many times (I shot for John Deere in Australia for years); expecting this, I shoot at whatever distance I am working at and review, then move slightly; this usually fixes the artifacts.
p.13 #12 · Announced: RX1R II with 42MP sensor and EVF
At nearly 3x the price, No IBIS, 4k video or weather sealing, for me it would have to drop kick the low light and AF performance of the x100T through the goal post of life to be a viable option. Like others I'm not really wanting 40+Gigapixels in a pocket camera. While I'm sure we all understand we can crank up the SS to get an appropriate exposure that is still a compromise Vs a lower MP option or having IBIS. While every camera has some compromise this one also comes with a hefty price tag. I very much look forward to some one here getting hands on and doing a shoot out Vs the Fuji.
p.13 #13 · Announced: RX1R II with 42MP sensor and EVF
Why do people keep referring to the exclusion of IBIS as a reason for not wanting a 42MP camera? At pixel level, both a 24MP and 42MP image will have the same amount of blur with camera shake. Resize the 42MP image to 24MP and both will look identical.
p.13 #14 · Announced: RX1R II with 42MP sensor and EVF
DLP wrote:
At nearly 3x the price, No IBIS, 4k video or weather sealing, for me it would have to drop kick the low light and AF performance of the x100T through the goal post of life to be a viable option. Like others I'm not really wanting 40+Gigapixels in a pocket camera. While I'm sure we all understand we can crank up the SS to get an appropriate exposure that is still a compromise Vs a lower MP option or having IBIS. While every camera has some compromise this one also comes with a hefty price tag. I very much look forward to some one here getting hands on and doing a shoot out Vs the Fuji....Show more →
seems like a waste, the 3 year old rx1 already kicks the x100t's ass in terms of every image quality metric. there are lots of reasons to choose the x100t over the old or new rx1, but iq isn't one of them.
p.13 #15 · Announced: RX1R II with 42MP sensor and EVF
Zony_user wrote:
Why do people keep referring to the exclusion of IBIS as a reason for not wanting a 42MP camera? At pixel level, both a 24MP and 42MP image will have the same amount of blur with camera shake. Resize the 42MP image to 24MP and both will look identical.
Not sure you really need IBIS in the first place on a small , lightweight cam like this. My bet with a A7rII and the 35 1.4 IBIS on you could match it with this little puppy with shutter speed handheld. I would have no problem shooting this at 1/8th of a second handheld. I did it with Leica M all the time. Need to remember the ISO reach is far higher than the previous cam as well.
p.13 #19 · Announced: RX1R II with 42MP sensor and EVF
Zony_user wrote:
Why do people keep referring to the exclusion of IBIS as a reason for not wanting a 42MP camera? At pixel level, both a 24MP and 42MP image will have the same amount of blur with camera shake. Resize the 42MP image to 24MP and both will look identical.
The worry about hand-holding the 42-MP sensor is somewhat amusing. How often does one take a hand-held shot and decides to make a 60x40 print out of it requiring the 42MP. The type of shots which are targeted for large output sizes like that generally take a lot more planning in terms of lighting, composition etc. At that point you are working of a tripod anyway. Maybe I am missing something but I just don't see a use case where my IQ needs are stringent enough to require 42MP of resolution and yet I am constrained to work handheld. Thats even true for 36 or 24MP sensors as well.
The other use of higher resolution is of course cropping. In that case if you say use the 1.5x crop mode, then the simple solution is to treat it as an 18MP APS-C camera, and consider the effective FL as 1.5 time the real FL and up the shitter speed as you would on an APS-C camera.
