Do you want image stabilisation in camera?

[Roger Hicks]

But turn an S2 on its side and you have a poor format. I've always thought 3:2 doesn't make a good portrait format. So you can crop but the 3:2 format doesn't help with composition and framing when in portrait orientation. YMMV.

MMDV (My Mileage Does Vary). I regard 3:4 as an unnecessarily squat format in either orientation. I doubt I'm entirely alone.

Personally I prefer to crop the long dimension whenever possible, rather than the short dimension. Both 3:4 and 2:3 have to be cropped to fit on a standard A4 page.

Cheers,

R.

 

[jaapv]

If only it were a Nikon or Canon with larger sensor it would have UK service support, reliable Japanese electronics and it would be very considerably less expensive. Agreed Leica can not compete in the electronics fast lane but their optics are pearless (at least the 35mm format and I assume the S ones are too). Hasselblad solved this via Fuji. I can only think that Leica would benefit hugely by a collaborative approach. Not only would this bring huge R and D funding but there would also be some one else to carry the can if something flopped! Obviously Leica would loose some control but I suspect we would benefit as users with a better product in the greater scheme of things.

Out of interest what would folks prefer in their M10, German electronics as is, or a reworking from one of the two Japanese camera giants?

Best wishes

Richard

Now I wonder what is wrong with Fujitsu electronics, which are inside the S2, as is a Kodak sensor.
Software is possibly by Jenoptik, who are a preferred supplier in the high-end miltary and commercial field. Leica has made some smart choices there, I think.

 

[rxmd]

MMDV (My Mileage Does Vary). I regard 3:4 as an unnecessarily squat format in either orientation. I doubt I'm entirely alone.

Well, it does have a fair bit of tradition behind it, though; 9x12 cameras were rather common, over here at least. Also you can choose any old format out there, between square and the golden ratio, and I'm sure you'll find someone who'll find it either squat or narrow.

Cropping is normal. The "standard" bit of the "standard A4 page" you mentioned is fairly arbitrary, too, based only on the interesting mathematical quirk that you can double the area of a 1:√2 rectangle easily while maintaining the ratio of the sides. It became standardized only in the 1920s in Germany, roughly at the same time when 3:4 ratio plate cameras were popular. :angel:

 

[Roger Hicks]

Well, it does have a fair bit of tradition behind it, though; 9x12 cameras were rather common, over here at least. Also you can choose any old format out there, between square and the golden ratio, and I'm sure you'll find someone who'll find it either squat or narrow.

Cropping is normal. The "standard" bit of the "standard A4 page" you mentioned is fairly arbitrary, too, based only on the interesting mathematical quirk that you can double the area of a 1:√2 rectangle easily while maintaining the ratio of the sides. It became standardized only in the 1920s in Germany, roughly at the same time when 3:4 ratio plate cameras were popular. :angel:

Highlighted portion: absolutely. That's why I ca't see the argument that 2:3 is a 'poor format' when used in portrait orientation, or that 3:4 is inherently superior in any way whatsoever.

But A-series ISN'T arbitrary, and it's not a 'quirk'. It's a rational choice, based on minimal paper wastage and a constant aspect ratio across all sizes. I knew it was German, but I hadn't realized it went back to the 1920s: when I was at school in the 60s, we still used quarto and foolscap, and of course the Americans still use their own weird sizes.

The aesthetic qualities of the A-series (or any other format) are a completely different matter. My own favourite is 5x7, surprisingly close to A-series (1:1.4 instead of 1:1.414) but I also like whole-plate (1:1.31) which is actually more squat that 4:3 (1:1.333333).

I firmly believe, though, that (a) one can often compose to fit just about any format, from square to panoramic, and that (b) there is absolutely nothing wrong with cropping any format to any shape, including circles and ovals, if that's what the picture needs.

Cheers,

R.

 

[jaapv]

I wonder what all the hoohah is about, IS has been available for the M9 and M8 all the time; at least, if we are to take this post seriously:

It works like a gyroscope. If you remove a bike wheel from the bike and spin it and hold the axel ends with your hands and try to move the wheel while it's spinning, you will see a bit of how it works. Research: Gyro-stabilization.

 

[tlitody]

I wonder what all the hoohah is about, IS has been available for the M9 and M8 all the time; at least, if we are to take this post seriously:

Now that's what I call a strapadictome 😱

 

[japro]

I believe they use an accelerometer which tells the camera which direction and, funnily enough, the acceleration and that operates the motor the sensor is attached to, to nullify the camera movement within certain parameters. i.e. it won't fix panning effects. It can give you 4 or so stops of extra speed.

Actually the exact opposite is true. At least the stabilisation systems that I found more details about use angular velocity sensors and only compensate panning errors. Compensating the linear displacement of the camera would be very hard. Even if you have the data from the accelerometer you don't know the velocity in regards to the object you want to take a picture of (you don't know which inertial frame you are in...). Also the errors produced by the linear displacement are a lot smaller than the ones that are caused by panning. Focusing to infinity would pretty much deactivate the stabilisation.

My experience with image stabilisation is, that it's quite useful with longer focal lengths but gets pretty much useless below say 35mm. The frequencies of the "vibrations" that cause 1/10s photos to be blurry are usually outside the operation range of the stabilisation mechanism.

One possible disadvantage of sensor based stabilisation that hasn't been mentioned is the cooling of the sensor. Since the sensor is pretty much detached from the rest of the camera heat dissipates slower which can lead to higher thermal noise when doing long exposures.

 

[tlitody]

Actually the exact opposite is true. At least the stabilisation systems that I found more details about use angular velocity sensors and only compensate panning errors. Compensating the linear displacement of the camera would be very hard. Even if you have the data from the accelerometer you don't know the velocity in regards to the object you want to take a picture of (you don't know which inertial frame you are in...). Also the errors produced by the linear displacement are a lot smaller than the ones that are caused by panning. Focusing to infinity would pretty much deactivate the stabilisation.

My experience with image stabilisation is, that it's quite useful with longer focal lengths but gets more or less useless below say 35mm. The frequencies of the "vibrations" that cause 1/10s photos to be blurry are usually outside the operation range of the stabilisation mechanism.

I think you need to be careful here because all image sensor stabilisation is trying to do is to correct any change in camera velocity at the point you trip the shutter. Change in velocity = acceleration (or deceleration) so accelerometer is perfectly valid for this. If the camera is moving at 100mph or zero makes no difference because that is not what in camera sensor stabilisation is aiming to correct for. It is trying to correct for shake at the point of tripping the shutter.
So all it needs to do is move the sensor in the same direction and speed as the camera shake at the point of tripping the shutter. It does this within limits. It doesn't care if you are panning. It only cares if you change panning speed at the point you trip the shutter. Its not perfect!

Don't confuse with video camera stabilisation. Completely different because that is about smoothing camera movements and not about correcting for camera position movement.

 

[japro]

I think you need to be careful here because all image sensor stabilisation is trying to do is to correct any change in camera velocity at the point you trip the shutter. Change in velocity = acceleration (or deceleration) so accelerometer is perfectly valid for this. If the camera is moving at 100mph or zero makes no difference because that is not what in camera sensor stabilisation is aiming to correct for. It is trying to correct for shake at the point of tripping the shutter.

Well, for one thing, the information about the sensors actually measuring angular velocity stems from a rather technical document about the stabilisation system KonicaMinolta/Sony use in their dSLRs, Your argument about only the change of velocity being important only works if you assume that the camera is at rest in regards to the subject just before tripping the shutter. Also as already mentioned, correction linear displacement would be irrelevant to a camera focused to infinity...

 

[Paul Luscher]

Not really. Sounds to me like what I call "feature creep." I saw one post somebody did on all the things he wanted in the M9, and by the time he was done, I thought "he's asking for another DSLR!"

K.I.S.S. principle should apply in the case of the M9. Otherwise, you may end up with another gadget-laden monster sporting way too many buttons, and we have plenty of those already. I.S. isn't really needed on the M9--with a steady hand--given the lack of mirror slap.

After all, the Big Boys (and Girls) of photography took all those great pictures in the days of yore without a lot of fancy features....

 

[ederek]

To the OP - No.

For static subjects it would help [a little], but can use a tripod for those.

It's not going to add value when the subject is moving. A rangefinder can be handheld reliably to 1/12 of a second, even with longer focal lengths.

A sensor with better low light performance / dynamic range is MUCH more important than stabilization. ..More important than more pixels.

Here are two examples from last week. Was shooting a 5D (1st version) and the M9 side by side. ISO 1250 wasn't quite enough, had to go to 2000 on the M9 (1600 on 5D w/ the 85/1.2 and 50/1.2).

BTW - for an event like this with limited available light, the 5D (not even MkII) outperformed the M9 and produced much better files. I would have needed to keep the ISO below 800 on the M9 to get an equivalent file quality. The 5D was easier/faster to focus (manual focus with upgraded focusing screen, also these are 1.2 lenses which helps). If I were doing weddings, I'd have to sell a Leica lens to get a MkII body. Maybe use a flash with the M9. 😱 Regardless, I'm happy with the M9 and carry it every day. 🙂

M9, 35mm biogon, between F2 and 2.8; ISO 2000; 1/12 sec

M9; 75mm lux; between F1.4 and 2; ISO 2000; 1/25 sec

Image stabilization relative to other priorities = waste of money and limited Leica engineering resources.

 

[tlitody]

Not really. Sounds to me like what I call "feature creep." I saw one post somebody did on all the things he wanted in the M9, and by the time he was done, I thought "he's asking for another DSLR!"

K.I.S.S. principle should apply in the case of the M9. Otherwise, you may end up with another gadget-laden monster sporting way too many buttons, and we have plenty of those already. I.S. isn't really needed on the M9--with a steady hand--given the lack of mirror slap.

After all, the Big Boys (and Girls) of photography took all those great pictures in the days of yore without a lot of fancy features....

True but they did it with the clunking great camera firmly fixed on tripod and using a leaf shutter or a top hat 🙂

 

[Roger Hicks]

True but they did it with the clunking great camera firmly fixed on tripod and using a leaf shutter or a top hat 🙂

Not really. The equivalent of ASA 200 film, f/1.5 lenses, 1/5 second exposures and elbows braced on the table worked pretty well for the great picture magazines.

Cheers,

R.

 

[tlitody]

Well, for one thing, the information about the sensors actually measuring angular velocity stems from a rather technical document about the stabilisation system KonicaMinolta/Sony use in their dSLRs, Your argument about only the change of velocity being important only works if you assume that the camera is at rest in regards to the subject just before tripping the shutter. Also as already mentioned, correction linear displacement would be irrelevant to a camera focused to infinity...

the link I gave for the accelerometer shows acceleration in 3 directions. From that you can calculate the actual acceleration in the cumaltive direction which means you can fix camera shake.

I think the angular velocity detector and accelerometer need to be carefully assessed because the accelerometer is measuring rate of change whereas measuring anglular velocity is measuring speed and direction. I don't think the starting speed is relevant if at all unless you are designing a perfect system. Camera shake from hand holding works within known parameters so an accelerometer can know that if the rate of change in any direction is x then it must move the sensor at a speed and direction of rate of change applied to known factor. Angular velocity is irrelavant unless its a video camera. Do those cameras have video in them? An M9 doesn't.

That's not to say some of these cameras don't measure angular velocity but just that I think its irrelevant for fixing camera shake in stills cameras but then I'm not designing a perfect system but merely one that will give 2 to 4 stops hand holding speed advantage.

 

[tlitody]

Not really. The equivalent of ASA 200 film, f/1.5 lenses, 1/5 second exposures and elbows braced on the table worked pretty well for the great picture magazines.

Cheers,

R.

Were they as sharp as they could have been with IS? 😉

 

[Roger Hicks]

Were they as sharp as they could have been with IS? 😉

Would more sharpness have made them better?

Cheers,

R.

 

[Paul_C]

Would more sharpness have made them better?

Would less?

If "more sharpness" was so superfluous, why did they bother bracing their elbows on the table?

 

[Roger Hicks]

Would less?

If "more sharpness" was so superfluous, why did they bother bracing their elbows on the table?

Remember the old hippy dictum: "Any is sufficient. Enough is too much."

Cheers,

R.

 

[japro]

the link I gave for the accelerometer shows acceleration in 3 directions. From that you can calculate the actual acceleration in the cumaltive direction which means you can fix camera shake.

Whats a "cumulative direction"? You can integrate the acceleration of the camera once to get the velocity up to a unknown constant velocity and then again to get the location up to an again unknown "offset". Those two unknowns depend on the inertial frame you want to use, which is usually the one the subject is in. But there is no way the camera can know that. The camera would have to guess (possibly by averaging its location) the desired inertial frame...

...Angular velocity is irrelavant unless its a video camera. Do those cameras have video in them? An M9 doesn't.

So you say there is no tilting of the camera going on during exposure and the main source of blur is the linear displacement?

 

[Ben Z]

Weighing in late on this thread, but IMO I'd just as soon Leica didn't put IS in the next M10. IS would be one more new trick for Leica to learn (slowly, as they tend to) and one more thing to boost the cost of an already hideously expensive body. It works well in my Canons, but it's only there with lenses that need it (physically long, and somewhat slow f-stop). And it's essential in my DLux3, which because it doesn't have a normal viewfinder requires being held away from the body. I've never had a problem bracing an M camera steady, particularly since available focal lengths are at a conservative limit.