[simons-photography]

right thats settles that

[reeftool]

The shake reduction works well for me for hand held shots so far. I did have some of my eclipse shots come out bad because I left it on while on the tripod. Live and learn i guess. My daughters college rowing season starts in another week ( if the darn ice ever melts) and I will be shooting my first shots with the K10D and Sigma 70-300. I have a monopod and will be panning. I will probably experiment with SR on and off. Panning with a tripod or monopod isn't the same as shooting night scenes. Also what about tripod shots from an unstable platform like a boat? SR on or off?

[blwnhr]

Below is an example I took on the weekend. I honestly didn't think it'd work, thought it would just be too slow. However, SR came through with the goods!

As for SR while panning...this get's complicated. Basically I would say it doesn't work. Simple as that. Panning for a person in a slow jog at 100mm you get triple images or ghosting in the image. My only thoughts are that the sensor is trying to stay still and moving left right, but can't keep up.

However, if I'm shooting soccer with the Bigma then SR works if I'm over 300mm. Probably because the movement of the camera is slower then the SR can keep up. As soon as you get below that 300mm mark there is blur and ghosting coming into the images.

Forget about while shooting cars at any focal length, it is just too fast for the SR.

I wish I had examples of what panning looks like with SR on but I don't, delete them straight away.

Anyhow, shot this, K10D with the Sigma 18-125mm, excuse the subject matter, but the image is what's important. The was hand held on a chilly night and I'd been shooting for 14 hours. I think the only detail lost in there is because of the high ISO. Image is totally unprocessed.

18mm, 1/2sec., f/3.5, ISO-800.

[rfortson]

I'll let you decide.

This was taken with the K100D and the kit lens at 18mm, f/6.7, ISO 800, 1.5 sec (that's one and one half second), handheld.



This one was at 23mm, f/5.6, ISO 800, 1 second, handheld.



This was taken with the Tamron 18-250 at 55mm, f/4.5, ISO800, 2/5 second, handheld.



I'm not saying these are great photos, but I don't think I would've been able to capture anything close to this without the shake reduction on the Pentax.

[rfortson]

As for SR while panning...this get's complicated. Basically I would say it doesn't work. Simple as that. Panning for a person in a slow jog at 100mm you get triple images or ghosting in the image.

I've never noticed any problem with SR and panning. On my K20 with the 50-135, I got some great shots of runners and I had the SR on the entire time (indeed, I rarely turn it off).

This was taken when the runner was perpendicular to me, meaning I was moving the camera at maximum speed (granted, she wasn't running that fast). I don't see any ghosting or blurriness. You can check the ginormous full size shot on my flickr stream to see it up close. It's as sharp as any image I have.

[Gooshin]

SR works, period.

[simons-photography]

nice pictures everyone, but on the upside if the SR is not dependand on an analisis of the image but a motion sensor then a moving picture is not a problem say a street with moving cars or a windy day I was under the impression that a non static subject would not work

[Canada_Rockies]

why exspensive ? the image sensor is already there it would just mean a different interfacing for the image sensor as oposed to a motion sensor

Expensive, as in differentiating between subject motion and camera motion. You are opening a whole mess with this for the firmware. Basically, we would need firmware updates that included a large sample of typical motion that would require SR, and a whole other set that would require no SR. The processing power to work on something like this would be more than the whole camera uses now to determine scenes for the matrix metering.

[falconeye]

Hi simons-photography,

the most intuitive evidence how well SR works with Pentax (a K20D in this case) I've found so far is this:

The Pentax K20D: a RAW review, page 4 .

It isn't scientific. But all of a sudden, you will see why you can't live without (BTW, don't take the high ISO images on that site as typical. They are not, for some unknown reason).


The Pentax SR detects body movements by gyro sensors which measure the coriolis force. So they do measure speed (in two dimensions), not acceleration. This is more accurate but also a reason why pan doesn't work which would work if acceleration were measured in the first place.

The SR system must do a frequency analysis of movements and phase-shift a little to compensate because otherwise, any compensational force would always be too late. This is why an image-based method would fail (if it sees a shift it would already be too late!).

Don't forget that a reduction by 4 f-stops means nothing less than a compensation of 94% of the movements (so with a 500mm lens, shakes by 0.1mm happening within, e.g., 1/50s are reduced to only 5 micron or 1 pixel)... Don't try this at home or with image processing

[simons-photography]

Expensive, as in differentiating between subject motion and camera motion. You are opening a whole mess with this for the firmware. Basically, we would need firmware updates that included a large sample of typical motion that would require SR, and a whole other set that would require no SR. The processing power to work on something like this would be more than the whole camera uses now to determine scenes for the matrix metering.

yes well I assumed it would assume the subject is a still one so if it did move it would track it but I suppose most of us just want general shake reduction and not for panning so what is the provided SR is best