Jman13 wrote:
Yes, it will crop in the viewfinder as well. Note on that...if shooting RAW with a non 4:3 aspect ratio will not affect the RAW file at all. The crop will be applied if using Olympus software, but it's ignored by things like Lightroom. Of course, JPEGs will be appropriately cropped.
Wow ... are you saying that the file size will be exactly the same whether your aspect ratio is 4:3 or 3:2 ... as long as you shoot RAW and import your images in LR? This is terrific news to me ... as I have yet to get used to and to like the 4:3 aspect ratio ... after 40 years of shooting in the 3:2 aspect ratio.
I'm saying the crop doesn't do anything to the RAW file. If you select 1:1 aspect or 16:9, the RAW will still be the full 4:3 image. If you develop in a RAW converter that reads the Olympus metadata for the crop, it'll show you the crop, but Lightroom doesn't...it just shows you the full RAW file. JPEGs are cropped, of course. The same thing happens with the digital TC...Oly just puts a 'crop box' around it in a metadata sense, which is ignored by Lightroom (so it's essentially only useful for framing if you only shoot RAW).
Edit: Ok.....apparently nevermind. Just did a shot at 1:1 and sure enough, it's only showing me the 1:1 crop in Lightroom. However, it still does record the whole image and the whole RAW file at 4:3...as playback on the camera shows a 1:1 box superimposed on the whole image. Wonder if it's a setting in Lightroom somewhere, as I'm almost positive I've seen it imported with the full captured frame.
In OlympusViewer, you can look at that RAW file and change the crop after the fact...but can't find it in Lightroom now. Maybe I just had it backwards in my head.
I no longer have my GH2 to experiement, but I wonder if the RAW files it captured actually could have shown the whole image area from the multi-aspect sensor ?
From what I remember the RAWs were displayed as just the chosen aspect ratio but it certainly would of been cool to get access to all the pixels, unless of course those pixels are somehow not even active depending on the aspect ratio
On a fixed sized sensor like the EM-5 I would have to think the real RAW data should be in 4:3 and LR etc just shows the crop as the camera was set to.
If it doesn't, and somehow deactivates those pixels, I certainly would shoot RAW in only 4:3 to make sure I'm getting all the pixels possible for more framing options in post.
Jman13 wrote:
I'm saying the crop doesn't do anything to the RAW file. If you select 1:1 aspect or 16:9, the RAW will still be the full 4:3 image. If you develop in a RAW converter that reads the Olympus metadata for the crop, it'll show you the crop, but Lightroom doesn't...it just shows you the full RAW file. JPEGs are cropped, of course. The same thing happens with the digital TC...Oly just puts a 'crop box' around it in a metadata sense, which is ignored by Lightroom (so it's essentially only useful for framing if you only shoot RAW).
Edit: Ok.....apparently nevermind. Just did a shot at 1:1 and sure enough, it's only showing me the 1:1 crop in Lightroom. However, it still does record the whole image and the whole RAW file at 4:3...as playback on the camera shows a 1:1 box superimposed on the whole image. Wonder if it's a setting in Lightroom somewhere, as I'm almost positive I've seen it imported with the full captured frame.
In OlympusViewer, you can look at that RAW file and change the crop after the fact...but can't find it in Lightroom now. Maybe I just had it backwards in my head....Show more →
So it crops the image to the aspect ratio that you want, and the file size shrinks. That's no good ... as that is the same as cropping in post.
Jman13 wrote:
If you shoot RAW, I've found that the OM-D has enough highlight headroom that I can easily shoot at ISO 200, f/1.4 and 1/4000s and almost always get all the detail back pulling it in RAW. There may be some really hot spots that stay blown, but those are specular highlights most of the time. Heck, I have even been able to get most of the detail back when shooting at f/0.95 on the Voigtlander 17.5mm in late afternoon sun.
Wouldn't it better to just dial-in some negative Exposure Compensation, so the captured image is not over-exposed?
bobbytan wrote:
So it crops the image to the aspect ratio that you want, and the file size shrinks. That's no good ... as that is the same as cropping in post.
But... what did you expect If they would add pixels at the top and bottom, the camera would grow. This is an option with a compact sensor, but not with sensors this large. I think it is awesome that the camera will let you compose in 1:1.
bobbytan wrote:
Wouldn't it better to just dial-in some negative Exposure Compensation, so the captured image is not over-exposed?
At f/1.4 in bright sunlight, you're at 1/4000s (plus)...EC won't do anything because it can't go faster than 1/4000s...EC would theoretically make the shutter speed faster, but it can't go any faster here. But, as I said, luckily there's enough highlight headroom with the E-M5 that most detail isn't blown in the RAW at 1/4000s and f/1.4 in bright sunlight. Now, shoot JPEG, and it'll be massively blown out.
carstenw wrote:
But... what did you expect If they would add pixels at the top and bottom, the camera would grow. This is an option with a compact sensor, but not with sensors this large. I think it is awesome that the camera will let you compose in 1:1.
How is that Panasonic can do it with their m43 cameras?
Panasonic can only do it with the GH1 and GH2, and it's because they use a larger sensor than strict m4/3 spec, that has room to meet the m4/3 diagonal at 4:3, 3:2 and 16:9 aspect ratios, so there are pixels on the edge and top that are unused in certain aspect ratios and then used in other aspect ratios.
There's obviously not an inherent limitation to this, it's just that only Panasonic has decided to implement a multi-aspect sensor in their cameras. I liked having it in my GH2, but I have found I really don't miss it in my E-M5. I'm not saying I wouldn't take it back, just that 16MP is high enough resolution that I don't really feel like I'm losing much if I have to crop to 3:2 or 16:9, and I can usually just back up a tiny bit if I want to maintain the width I need.
1:1 with a full MFT sensor in the width would also impact the size of the camera. At best, a compromise between 100% crop and some partial crop with a few extra pixels might fit, but honestly, for the extra cost, complexity and size, it would hardly be worth it. 99% of all owners probably don't use it or even know it is there. I would love it, but there are very few of us.
carstenw wrote:
1:1 with a full MFT sensor in the width would also impact the size of the camera. At best, a compromise between 100% crop and some partial crop with a few extra pixels might fit, but honestly, for the extra cost, complexity and size, it would hardly be worth it. 99% of all owners probably don't use it or even know it is there. I would love it, but there are very few of us.
actually the OM-D has to have an oversized sensor already in order IBIS to work - so moving the sensor around doesn't result in losing pixels.
No it doesn't. The way IBIS works is by basically making the sensor stay still while the lens and camera body rotate and shift around it. It's not capturing more pixels and then shifting things in software.
Jman13 wrote:
No it doesn't. The way IBIS works is by basically making the sensor stay still while the lens and camera body rotate and shift around it. It's not capturing more pixels and then shifting things in software.
if the sensor stays still while the camera and lens move around, it means that the sensor is moving in relation to the camera. or another way to look at it is that the image circle is moving around on the sensor. this means that the sensor must be large enough so that there are are always pixels under a 4/3 crop of the image circle.
sebboh wrote:
if the sensor stays still while the camera and lens move around, it means that the sensor is moving in relation to the camera. or another way to look at it is that the image circle is moving around on the sensor. this means that the sensor must be large enough so that there are are always pixels under a 4/3 crop of the image circle.
You're right that it moves around the image circle. Think about what you're saying here. How would having more pixels around the edge have anything to do with IS? How would that work exactly?
The sensor is not larger than normal, nor does it need to be. The limits are the image circle with IBIS, but there's enough slack to allow it. In fact, it wouldn't surprise me if the m43 spec slightly oversized the image circle requirements to allow for greater IBIS lattitude.
Jman13 wrote:
You're right that it moves around the image circle. Think about what you're saying here. How would having more pixels around the edge have anything to do with IS? How would that work exactly?
more room for the image circle to move. not sure what doesn't make sense to you about that. there are two possible implementations:
1) as you suggest the sensor is exactly the number of pixels used in an image and the image is taken from wherever IS has moved the sensor to. this depends on the lenses image circle being larger than necessary for a fixed sensor and implies the lenses must be slightly larger than they would need to be for a fixed sensor or that the image could often have asymmetric vignetting and other lens aberrations.
2) as i suggest the sensor is slightly larger than the number of pixels it takes to fill a 4/3 image and the image is taken as a 4/3 sized crop of the area of the sensor closest to the center of the image circle. in this case the camera needs to know the location of the sensor at time of image capture in order to choose the location of the 4/3 crop (it better for IBIS to work). this is how an olympus rep told me IBIS worked in 4/3 dslrs (circa E3). they may have decided that wasn't the best way to do things since then, or the rep may have been talking out his a**.
FlyPenFly wrote:
If that's the case then is every Pentax and Sony DSLR body have bigger sensors than Nikon?
as far as i know pretty much all sensors are larger than necessary, the question is what are the dimensions of the active pixels. perhaps i'm misremembering though...
sebboh wrote:
1) as you suggest the sensor is exactly the number of pixels used in an image and the image is taken from wherever IS has moved the sensor to. this depends on the lenses image circle being larger than necessary for a fixed sensor and implies the lenses must be slightly larger than they would need to be for a fixed sensor or that the image could often have asymmetric vignetting and other lens aberrations.
This is exactly what happens. That's why the IBIS systems are often called 'sensor shift' type IS. The sensor moves to compensate for camera movement. The E-M5's sensor is suspended in place by an electromagnet (take the lens off, shut the camera off, and you'll watch the sensor drop down). Lens image circles are always a little larger than the framed image anyway...these are small movement compensations...it can't correct for a 5mm shift or anything. And I have seen asymmetric vignetting on stabilized low shutter speed shots, though it's not bad enough to be distracting.
2) as i suggest the sensor is slightly larger than the number of pixels it takes to fill a 4/3 image and the image is taken as a 4/3 sized crop of the area of the sensor closest to the center of the image circle. in this case the camera needs to know the location of the sensor at time of image capture in order to choose the location of the 4/3 crop (it better for IBIS to work). this is how an olympus rep told me IBIS worked in 4/3 dslrs (circa E3). they may have decided that wasn't the best way to do things since then, or the rep may have been talking out his a**. ...Show more →
What you've described couldn't possibly work to stabilize an image. You are suggesting that the sensor records a constant stream of data, tracking where the movement has all been, then reassembles all that data on top of the correct pixels to gain the sharp image. This would be nigh impossible without a near infinite speed readout of the sensor.
To understand, think about this...if what your suggesting is what really happened, on say, a 0.4 second stabilized image (which I have done easily with the 12mm on the E-M5)...the first 0.1 seconds, you shift slightly up, so the sensor now fills these 'extra pixels' a bit...but the second 0.1 seconds, you shift a little left...now the exposure is filling ALL different light wells. Then you yaw a bit...and ALL different light wells are being exposed with the same detail again...then you shift again. Meanwhile, there are 90 pixels that have all experienced photons from the single point of light that should be exposing the sensor....how does the sensor 'reassemble' this image?
Also, remember that the OM-D's IS is visible in the viewfinder...it can't crop after the fact, since it displays the feed off the sensor live in the viewfinder...stabilized.
Heck, you don't need any more proof than this. The E-M5's sensor is 17.3 x 13mm. The GX1's sensor is 17.3 x 13mm. One has sensor shift stabilization, one doesn't. Same exact sensor size. The GH2's sensor is larger due to the multi-aspect nature, but the E-M5's is not.
Jman13 wrote:
What you've described couldn't possibly work to stabilize an image. You are suggesting that the sensor records a constant stream of data, tracking where the movement has all been, then reassembles all that data on top of the correct pixels to gain the sharp image. This would be nigh impossible without a near infinite speed readout of the sensor.
To understand, think about this...if what your suggesting is what really happened, on say, a 0.4 second stabilized image (which I have done easily with the 12mm on the E-M5)...the first 0.1 seconds, you shift slightly up, so the sensor now fills these 'extra pixels' a bit...but the second 0.1 seconds, you shift a little left...now the exposure is filling ALL different light wells. Then you yaw a bit...and ALL different light wells are being exposed with the same detail again...then you shift again. Meanwhile, there are 90 pixels that have all experienced photons from the single point of light that should be exposing the sensor....how does the sensor 'reassemble' this image?
Also, remember that the OM-D's IS is visible in the viewfinder...it can't crop after the fact, since it displays the feed off the sensor live in the viewfinder...stabilized. ...Show more →
you are making this much more complicated that necessary, the sensor doesn't need to shift mid capture it just selects the most centered pixels at the beginning of capture (or perhaps at the end though that seems wasteful) and records from them throughout the time the shutter is open. this should drastically reduce the incidence of asymmetric vignetting (which i never really observed in telephoto use) and allow lenses with an image circle smaller than aps-c.
Jman13 wrote:
Heck, you don't need any more proof than this. The E-M5's sensor is 17.3 x 13mm. The GX1's sensor is 17.3 x 13mm. One has sensor shift stabilization, one doesn't. Same exact sensor size. The GH2's sensor is larger due to the multi-aspect nature, but the E-M5's is not.
do they not report both sensor size and active pixel area anymore? the two never matched up.
sebboh wrote:
you are making this much more complicated that necessary, the sensor doesn't need to shift mid capture it just selects the most centered pixels at the beginning of capture (or perhaps at the end though that seems wasteful) and records from them throughout the time the shutter is open. this should drastically reduce the incidence of asymmetric vignetting (which i never really observed in telephoto use) and allow lenses with an image circle smaller than aps-c.
Um, that wouldn't stabilize an image. That would just avoid any initial shock. Trust me...the sensor moves...you can WATCH it move with the lens off.
If they would add pixels at the top and bottom, the camera would grow. This is an option with a compact sensor, but not with sensors this large. I think it is awesome that the camera will let you compose in 1:1.