« Pentax Updates Its Popular DSLRs | Main | Svelte Canon Rebel XSi »

Wednesday, 23 January 2008


guess that the built in shake reduction just isn't enough. Maybe if they had in-lens IS, it wouldn't have this problem ;)

A lot of the samples I've seen have appeared to have motion blur. Look at the ones from Dubai on photographyblog.com, for example. But Benji's ones are tack sharp, so that's put my mind at ease:


What's amazing is how bottomless these images seem - you look at them at 100% and then zoom in...and in...and in. Even if the details look a bit soft in cases, you also have to remember how small a part of the image you're looking at.

One intriguing thing: the LCD can be color calibrated, I wonder if it can be set to be in pure b/w? Combine that with live view and you've got real time b/w composition. Not sure if it's possible, but would be cool for those of us that work in bw.

Love the manual fine-tuning of focus (back/front focus compensation) per lens also. A good "cost saving" move for Pentax - a small cost built into the body saves a lot of service costs later.

In third sample image (the night townscape shot) one can clearly see two dust spots in the (blue) background - even if one scales the image down to screen size.

You'd think that somebody would have a close look at the sample pictures before they are posted, right?

Surely you jest on the "motion-blurred" comment! Can you see a directional blur, or are you seeing a general blur/fuzziness? If it is the latter, it might be a result of using a Bayer sensor, some glitch in the image pipeline, or maybe just a soft lens. Seriously, please share the attribute that leads you to suspect motion blurring in this image.

Is it just me or is the center of the b/w shot quite a bit sharper than the edges? That's a little odd, because even of this was shot at 100asa, the lens would have been stopped down quite a bit. Kit lens?

I'm not sure if that looks like motion blur or variable sharpening. You'd have thought they would publish images to satisfy the pixel peepers.

Mike, the crop looks better than the mother shot. Strange.

Just another example of, it not how many mega pixels. But all the stuff in front of it, anti shake,poor lenses,filter etc.
Then there is the photographer behind the camera.
I think this is a combination of, not the best lens in the world and maybe the anti shake does more harm then good, something like defraction at f/22.
14 mega pixels cameras need top of the line lenses and a good tripod.

Image parameters: 18mm f/5.6 1/160th ISO100, created in CS3 Mac

I doubt it is motion blur, more likely a combination of lens softness, focus, conservative sharpening, and B&W development. We don't know what mixing settings were used. If the red-channel was mainly used (going by the sky), then the resolution of a RGBG Bayer sensor is going to suffer at the per-pixel level.

Certainly looks a little motion blurred to me!

I'm not sure how shake reduction works, but if it's a servo mechanism there must be background movement which would show up at the limits of resolution. Which is probably why when using a tripod you're advised to turn it off.

I've gotta say, I continue to be baffled by these online samples, probably significantly downsampled, as a way of evaluating a 10-12 mp camera. Maybe it's just me but can you really tell anything about this camera by looking at this or the other sample images? In any event, who cares? Gimme more "thoughtful photographer" and less gear worship.

It looks a little blurred to me also. FWIW, the EXIF Viewer plug-in for Firefox shows this exposure information:

# Exposure Time (1 / Shutter Speed) = 1/160 second = 0.00625 second
# Lens F-Number/F-Stop = 56/10 = F5.6
# Exposure Program = aperture priority (3)
# ISO Speed Ratings = 100
# Exif Version = 0220
# Original Date/Time = 2007:10:10 18:47:27
# Digitization Date/Time = 2007:10:10 18:47:27
# Exposure Bias (EV) = -3/10 = -0.3
# Metering Mode = pattern / multi-segment (5)
# Flash = Flash did not fire, compulsory flash mode
# Focal Length = 1800/100 mm = 18 mm
# Colour Space = 65535
# Image Width = 3104 pixels
# Image Height = 4672 pixels
# Image Sensing Method = one-chip color area sensor (2)

(No guarantee that that's correct, but the camera model and other information it shows seems to be accurate.)

Looks a little motion blurred to me too ... or soft like a raw file that hasn't been sharpened.

Looks good to me.

Benjamin Kanarek (Pentax sponsored photographer) uploaded some sample shots here:


You can download the full size. The ISO 1600 (street sign photo) look good :-)

Isn't the blur an effect of the noise suppression? (Unsure why they would need that at ISO 100)

It's typical of camera companies to show example pictures that haven't been particularly carefully made. I've been noticing the tendency for almost thirty years now. I've seen some really egregious examples over the years--pictures in brochures that were out of focus, pictures supposedly showing lens resolution that were motion-blurred.... I've seen lots of examples over many years, from many different companies.

Mike J.

Dear Folks,

Leave pixel-peeping to the pros.

You're all fixating on some lens aberrations.

Download the full image from the Pentax web site. Take it up to 200% in Photoshop. Now look in the DEAD CENTER of the frame, not off-axis. See any smearing? I don't. Now start looking towards the edges of the field in any direction. What do you see? Increasing smearing as you move away from the center!

Solid evidence the lens used wasn't perfectly corrected for one or more of the off-axis aberrations. Not uncommon with extreme wide angles or wide-to-tele zooms used at the short end, which this could plausibly have been. Has nothing to do with the camera.

Repeat after me, "I will not obsess, I will not obsess, I will not obsess."

pax / Ctein

P.S. As an aside, when you get down to the resolution limits of a Bayer array camera (which this photo isn't), you shouldn't expect to see a radially-symmetric blur circle! The sensor pattern isn't radially symmetric, the processing algorithms aren't radially symmetric.

Is it just me, or do some of these fellas sound like they could find gristle in a piece of toast? The photo looks pretty good to me.

So the shot was taken at 18mm and F5.6 and it shows some off-axis aberrations. My guess is that this was taken with the kit lens zoomed to its shortest focal length(18mm). This example shows that the sensors are getting ahead of many lenses. Not surprising, really.


Good subject for a TOP posting - the worst photographs used by camera makers to advertise the quality of their camera :-)

My first impression of the full size version was "that looks a little soft." And then looking at the 100% crop I think it does need a bit more sharpening, whether or not it is motion blurred.

if the camera movement had the lens as the axis or "anchor point," then the further you moved from center the more blurred it would get. I've been having an occassional problem with the right side of my images blurring at slower shutter speeds; but only the right side. I finally figured out that it is most likely from camera movement when I depress the shutter. Since the left side of the camera is firmly anchored in my hand, that becomes the axis or "anchor point" and is sharp (or at least acceptably sharp), but as you get towards the right side you start to see some blurring. So if that posted shot was taken with the lens as the "anchor point," that could explain how it could be motion blur.

And for those who think that this is being petty; for me it certainly is not. For me it is not just about that posted photo. It's about understanding what is going on and why, as well as how things work, so that ultimately I can improve my own photos.

Dear Keith,

You can get a sharp center and blurred edges from camera motion, but only if you rotate the camera about the optical axis. It's not a common thing. So I assume horse, not zebra [smile]. Simply 'anchoring' the lens, as you put it, doesn't produce that effect.

The problem you're seeing is also commonly a side effect of the camera having a focal plane shutter. The film/sensor gets exposed in a stripe that moves from one side of the frame to the other at the synch speed of the camera. If the camera vibrates by different amounts during the exposure, different parts of the photo will have different amounts of blur. This is most likely to happen within a coupla stops of the synch shutter speed, either way.

Dunno if the right side of your photo is the leading or trailing edge of the exposure. If it were the leading, for example, your blurs could be caused by the small amount of jitter from when you press the shutter button, which dies out before the exposure is completed.

Messy stuff, photography! [grin]

pax / Ctein


Would the same sort of phenomenon apply to a vertical travel focal plane shutter? IF so, presumeably that would tend to show artifacts at the top or the bottom of the image depending on the direction of travel right?


When I said "anchor" I meant that the optical axis would be come the pivot point. In other words, if you were holding the lens solidly you might not be able to move the camera back or forth etc., but it might be easy to rotate it (on the optical axis) accidentally. That really isn't anything I have experience with, was just making a guess.

As far as the focal plane shutter goes, right... yeah, I wasn't thinking about that. Now it makes even more sense. Thanks!
I end up doing a good bit of shooting at about 1/20th of a second, but have even noticed it with some recent images that were shot at 1/45th sec., but again, only with some of the images.

Does anybody here know which way the shutter travels on a Canon 20D? I know there are ways of figuring that out, but I just can't think of any at the moment.

Dear Peter & Keith,

Yup, vertical shutter travel could produce horizontal bands of different blurriness.

Keith, you'll be amused at this-- my first reaction to your question was to recommend you open the back of the camera and observe the shutter curtain travel.

I think this would be ill-advised with a 20D!!

As for truly rotating the camera about the optical axis-- harder than you'd think! Set your camera to a slow shutter speed, like a half sec, and try to do it on purpose. it's hard to get the axis of rotation to line up will with the optical axis, so you don't get on-axis smearing.

pax / Ctein

I'm pretty certain it is a vertical travel shutter, but the softness is on the right side and from top to bottom... I think. That's the impression I have anyway, I will have to double check that it is top to bottom. If that is the case, then that puts me back at square one, which is that I'm having issues with jabbing at the release instead of squeezing it, which is pushing down on the right side of the camera.

Yeah... I'm probably not going to be opening the back up on my 20D. =)
I did take the lens off and look inside as I shot a few frames, I couldn't even see the shutter movement! I could try a couple of other things, but at this point it doesn't really matter since it is almost definitely a vertical travel shutter.

Motion blur? You mean, you think the cathedral moved while they took the photo?? :-)

Dear Steve,

Did it say *whose* motion????


pax / Ctein


If you happen to have a non-TTL strobe that you can trigger manually, try taking a picture at a shutter speed that slightly exceeds the top sync speed of your camera.

In the resultant image, there should be a band that's lit by flash and a band that isn't. The orientation of the band will tell you which direction your shutter is traveling.


I actually don't have any kind of strobe, and I've pretty much gone back to the idea that the blur is from me jabbing at the shutter release. But If I do come across a non TTL strobe I will give it a try, even just for the heck of it.

Thanks though, I do appreciate the suggestion!

I know we've beaten the shutter-curtain-travel thing to death, but here's a nice trick for them's what don't got a strobe but do have a CRT TV. (Never tried this with flat screens, so I don't know if that works).

Photograph your TV screen at 1/125th sec. The photograph will show a band of illuminated screen.

If the band is entirely horizontal, your shutter curtain travels vertically. If the band slopes, then the curtain moves horizontally. The leading edge of the curtain travel is on the upper side of the band.

pax / Ctein

Any flash can be used to check shutter curtain as long as the camera 'doesn't know' it is connected via the TTL circuits. PC cord, radio remote or tape all contacts on the hot shoe except the large one in the center and the ground on the side of the HS. Some cameras will not show a banding a bit faster than the synch speed, i.e. synch speed 1/250 band might not appear until 1/400 rather than 1/320. At a fast shutter two bands will appear, both curtains. 20D does have a verticle shutter. The banding started at the bottom on the camera I checked. Some cameras with a CCD and a fast synch such as 1/500 will not show any banding at any speed as the high speeds are: shutter open, sensor energized, sensor off shutter closes such as the Nikon D70 which uses this system at speeds faster than 1/90 i believe.

I agree that the picture isn't quite "tack sharp". From the EXIF info posted above, I noticed the F5.6. Do you know what the tack sharp apeture is for that particular lens? Was it the kit lens, did anyone figure out? Some lenses really benefit from that "sweet spot". You would think though, that Pentax would have gone out of their way to post the best possible pictures of this camera - as most dSLR shoppers are very critical photographers :-) Dpreview's sample shots with this camera are all quite sharp - but I noticed that most are taken at F4.0 - which I'm guessing is the sweet spot for the kit lens.

The comments to this entry are closed.



Blog powered by Typepad
Member since 06/2007