G7 killer

[Ray]

That cuts both ways:
- to get the newest lens-based stabilization, replace each lens that you want stabilized, use existing bodies,
- to get the newest sensor-base stabilization, replace each body (one for most of us) and use all existing lenses with stabilization.

Right! It does cut both ways. In fact, in the absence of any inherent superiority of one system over the other, sensor based stabilisation is more efficient since it applies to all lenses that fit the camera. Since camera bodies are developing and improving at a faster rate than lenses, there's more potential perhaps to lever a greater all-round improvement simply by making a body with a better anti-shake sensor, especially in situations where the total value of all one's lenses is far greater than the cost of the latest body.

I guess one can't have everything   . I recall when I upgraded from my D60 to a 20D, I felt I had acquired at least a 2 stop image stabilisation advantage for all my lenses, since performance at ISO 1600 with the 20D is at least as good as ISO 400 on the D60. In fact, it was better than a mere 2 stop advantage because higher ISO also allows the opportunity to prevent motion blur in the subject.

If Canon had previously adopted a sensor-based image stabilisation system and had been able to provide an additional stop of IS in the 20D body as well as much improved performance at high ISO, then that would have been just super   .

Which brings me back to Howard's suggestion that Canon's reason for choosing lens shift IS as opposed to sensor-shift IS is perhaps a means of ensuring backwards compaitibility to older bodies. How can this be? It's another of 'Howie opinions' that doesn't make sense to me. The Canon lens-based IS system was born in the days of film, long before Canon released it's first DSLR, the D30. I had actually switched from Minolta to Canon primarily because of Canon's IS system before I had an inkling that Canon was soon to release a 3mp DSLR. Howard as an engineer should appreciate that an anti-shake film based system would not have been an option.

The only good reason I can now see for Canon to switch to a sensor-based IS system (for its DSLRs) is if it were possible to provide an improvement over the lens-shift system. If Canon were to come out with a camera body that offered such improvement, there could hardly be any complaint from people who already own IS lenses. They would simply switch off the IS function in their IS lenses and enjoy the improved results from the camera body's internal IS.

[Jonathan Wienke]

I don't see how a sensor-shift IS system would work well with really long lenses (400mm on up). The sensor would have to move a pretty large fraction of its size quite fast to be effective. I think jiggling a lens element a few mm in the lens somewhere is a more advantageous approach, as it can work on a lens of any focal length.

[BJL]

I don't see how a sensor-shift IS system would work well with really long lenses (400mm on up). The sensor would have to move a pretty large fraction of its size quite fast to be effective.
[a href=\"index.php?act=findpost&pid=100832\"][{POST_SNAPBACK}][/a]

You may be right, but I am not sure, as I hinted in my earlier posts. (Remember the camera that started this is a digicam with maximum focal length of about 80mm.)

A summary of my earlier calculations.

1) If sensor stabilization is active during composition as well as during exposure, it might need up to 1cm or more of movement with 400mm, which might be difficult. However, if I remember right, Konica-Minolta said that its DSLR sensor stabilization system had over 1cm of movement. (I would not worry much about the sensor moving out of the image circle: super-telephoto lenses typically project an image circle distinctly larger than needed, and certainly can be designed that way at no additional cost if needed for optimal stabilization performance.)

2) If stabilization is activated only as shutter release starts (as in one mode of Panasonic OIS), long lenses are probably not a problem, as sensor movement only has to keep up with camera movement during the exposure time, plus a bit of "getting up to speed" time. To get about four stops of improvement over un-stabilized performance (16 times the shutter speed) the sensor movement needed during the actual exposure is no more than sixteen times the tolerable amount of image movement across the sensor in un-stabilized operation. Estimating the latter to be at most 1/1000th of sensor width, the sensor needs to move no more than 16/1000ths of sensor width during the exposure to get four stops of stabilization, regardless of focal length.

How much "up to speed" time is needed, and how much more movement does this require? It seems likely that a stabilization system need to handle changes in direction of camera movement within the exposure time, so it must have a "reaction time" no greater than exposure time. If so, the up to speed time would not need to be more than the exposure time. So double the time that the sensor needs to be in movement, and double the needed movement to about 32/1000ths of sensor width; only about 1mm for four stops in 35mm format.

[BJL]

Ray and Jonathan in particular,

what about the following dual stabilization approach for DSLR's?

- Some stabilized lenses, particularly at longer focal lengths where this technology might be the best option.

- Sensor stabilization working with all other lenses, including ones like wide angles where stabilization can occasionally be useful, but stabilized lenses are unlikely to be offered.

- Only one system used at as time, unless clever coordination can be worked out.


Pentax and Sony in effect have this dual option to a small extent, due to Sigma's stabilized OS lenses [correction: Sigma's only OS lens so far is not yet available in Pentax or Sony mount], and I am guessing that the FourThirds system will also have this soon, via Panasonic OIS lenses and new Olympus bodies.

[Ray]

Ray and Jonathan in particular,

what about the following dual stabilization approach for DSLR's?

- Some stabilized lenses, particularly at longer focal lengths where this technology might be the best option.

- Sensor stabilization working with all other lenses, including ones like wide angles where stabilization can occasionally be useful, but stabilized lenses are unlikely to be offered.


[a href=\"index.php?act=findpost&pid=100893\"][{POST_SNAPBACK}][/a]

I'm all for that approach. One of my most used lenses is the Sigma 15-30 with no IS. Since I try to use exposures that are at least 1/FL with IS switched on, (which is often possible with good lighting even at f11 and ISO 100, and if not, an increase in ISO fixes that), I should by the same standards be using 1/4FL with my Sigma 15-30. I'm afraid for some reason I don't keep to those same standards.

However, whilst this idea of a dual image stabilisation approach would be very attractive for the consumer, I think Canon would be reluctant to implement it because it might give a competitive edge to other lens manufacturers, such as Sigma. There would be many owners of a Canon DSLR system who would have considered the purchase of a cheaper third-party lens, having read perhaps it is optically on a par with the more expensive Canon equivalent, but have finally been dissuaded from such a purchase due to the lens not having image stabilisation. In situations where the more expensive Canon lens has IS and the cheaper 3rd party equivalent doesn't although it's optically as good, it's almost a no-brainer as to which lens to get, unless you are one of those who always uses a tripod.

[Goodlistener]

I didn't know what was the appropriate forum for this, but since Michael reviewed the G7 and there was a long discussion about the missing RAW on this one, I post it here...

The Askey forum has a news note on a new Olympus P&S, the SP550, which has an 18x (!!!) zoom lens -- 35mm equivalent of 28-504 -- with image stabilization, which it probably desperately needs at the long end. Anyway, the length of the zoom made me laugh. Other stuff is pretty interesting: 7.1 mp chip, which meant that they stayed clear of the MP race, and some of the 7 mp chips have been really good. Also, it has RAW. If the high ISO capability is there, and it could be (has settings for ISO 50-5000) with the slightly larger pixels of the 7mp chip, and if the glass is up to Olympus' higher standards...it could be a contender.

On the downside: electronic view finder.

JC
[a href=\"index.php?act=findpost&pid=97518\"][{POST_SNAPBACK}][/a]

There are a lot of High Zoom Electronic Viewfinder cameras that shoot RAW. Most of them look like a DSLR but are just a bit smaller and less expensive. They do have their advantages, but they are NOT pocket sized. Panasonic Lumix and Leica seem to be the only pocket sized cameras I know of that shoot RAW. However, "Word on the street" is that these models have noisy sensors.

If anyone knows of a widely available pocket sized camera with up to date features and which shoots RAW, I would love to know about it, but I have looked a lot and not found one, other than Panasonic and Leica.  (Which I won't buy).

In the interim I plan to get a Fuji S30d for about $300 or maybe $350 and wait for the ideal camera to come along.  There is one other option I have Found: Fuji E-900.  It shoots RAW at 9 MP and has a 2" LCD on the back.  menus on the screen are similar to Canon pull down style on the digital rebel series - which is not as appealing as the Nikon P&S use.  In th eideal world, Fuji would add a 3" screen + a viewfinder and clean up the user interface, but its probably a good camera.

Anybody who has experince with these models would be really good to hear from, but the best advice from members is always welcome.