Pages: [1]   Go Down

Author Topic: Canon: Motion blur depends on pixel size. Me: Really?  (Read 6261 times)

Walter Schulz

  • Full Member
  • ***
  • Offline Offline
  • Posts: 105
Canon: Motion blur depends on pixel size. Me: Really?
« on: February 05, 2012, 04:50:47 am »

I stumbled over a report done about a Canon 1D X presentation.
http://www.nikonuser.info/fotoforum/viewtopic.php?f=35&t=1481&sid=fc1b93b823dc0fcd687f64f4f2003688
and tried to make sense about this slide shown during this show:
http://www.myphotogallery.name/NewProd/20120203-CanonD1Xdemo-6.JPG
For those not able to read german I try my best to translate it into germish. Please feel free to correct me. Brackets=My Comments

===
Motion blur

*Thanks to larger pixel size motion blur will be reduced compared to EOS-1Ds Mark III.
*What is motion blur?
* Shorter shutter time are (sic!) needed to freeze moving objects. Pixel size has an influence on that.
* Smaller Pixels are requiring a shorter shutter time for sharp pictures without motion blur.
- A graphic showing the point (sorry for the pun)

(I think "shutter time" should be translated to "exposure time". And the first word of the last sentence is spelled wrong. There is no such word as "nothinger" or "noer" known to me and if: It doesn't make sense here.)

===
I thought "Hä?" which can be translated to "WTF?". What does this graphic show? I think they refer to the gapless lens introduced with 1D X. What is not shown is pixel pitch and without this the slide is meaningless. The concept transported in the slide is misleading.

Am I missing something, is there a correlation between pixel size and motion blur?
Thought experiment: I paint a grid (with width of the lines = x) onto the screen and watch a movie of a car race. Now I reduce the width of the lines. I cannot see any improvement concerning motion blur, I suppose.

Maybe someone is able to make sense out of it and explain it to me. I will try to understand, I promise!

Ciao, Walter
« Last Edit: February 05, 2012, 05:29:05 am by Walter Schulz »
Logged

hjulenissen

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2051
Re: Canon: Motion blur depends on pixel size. Me: Really?
« Reply #1 on: February 05, 2012, 05:40:12 am »

If you buy primarily an "18 megapixel" camera, and not a "24x36mm" camera and primarily look at images at 1:1 pixel on your screen, it might make sense to think about per-pixel motion blur.

To me it makes little sense, I hang my pictures on the wall or look at them on my PC-monitor/tv. If the camera was moved a fraction of a degree during capture, this blur would be a fixed length "tail" on my prints no matter the pixel size.

-h
Logged

Walter Schulz

  • Full Member
  • ***
  • Offline Offline
  • Posts: 105
Re: Canon: Motion blur depends on pixel size. Me: Really?
« Reply #2 on: February 05, 2012, 06:08:00 am »

Thanks for the answer.
That's my credo too: It's all about output format. Hope I'm not misinterpreting you ...

The thought about downsized pixel count occured to me, too. But I'm really wondering. Is Canon in fact that desperate to sell this well known concept to a pro audience in such a cheap way?

Ciao, Walter
Logged

clkirksey

  • Newbie
  • *
  • Offline Offline
  • Posts: 19
Re: Canon: Motion blur depends on pixel size. Me: Really?
« Reply #3 on: February 05, 2012, 09:53:14 am »

Yes pixel size does impact sharpness due to motion blur. Image that the combination of camera and/or subject motion results in a motion of the image at the image plane of X mm/sec. If the shutter duration is T seconds then the image will move TxX mm in T seconds. Now the if the pixel size is Y mm then ithe fraction of one pixel size that the image moves is will be t will take Y/(TxX). Typically this motion should be less than 10-30% of a pixel size before any noticeable degardation occurs. So a smaller pixel for the same shutter duration will result in more blur than a larger pixel. Simple, no?
Logged
A bird in the viewfinder...

Walter Schulz

  • Full Member
  • ***
  • Offline Offline
  • Posts: 105
Re: Canon: Motion blur depends on pixel size. Me: Really?
« Reply #4 on: February 05, 2012, 10:34:37 am »

As said above: This is a well known issue and I can cannot imagine why Canon tries riding a dead horse in front of a pro audience.

Ciao, Walter
Logged

hjulenissen

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2051
Re: Canon: Motion blur depends on pixel size. Me: Really?
« Reply #5 on: February 05, 2012, 10:53:12 am »

So a smaller pixel for the same shutter duration will result in more blur than a larger pixel. Simple, no?
But only because the larger pixels are more blurry. Trying to make this an advantage of big-pixel sensors is like saying that unsharp lenses is a good thing since you wont have a benefit from a recent sensor with many megapixels.

-h
Logged

clkirksey

  • Newbie
  • *
  • Offline Offline
  • Posts: 19
Re: Canon: Motion blur depends on pixel size. Me: Really?
« Reply #6 on: February 06, 2012, 09:14:40 am »

But only because the larger pixels are more blurry. Trying to make this an advantage of big-pixel sensors is like saying that unsharp lenses is a good thing since you wont have a benefit from a recent sensor with many megapixels.

-h
But motion blur that isn't perceptible for a larger pixel could be perceptible for a smaller one. Isn't this the issue? Trading a shorter shutter duration with higher ISO and smaller pixels versus larger pixel with longer shutter duration and lower ISO. It is not black and white but gray. There will exist a certain amount of motion which makes the resolution gained by using a smaller pixel insignificant, not zero but insignificant.
Logged
A bird in the viewfinder...

hjulenissen

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2051
Re: Canon: Motion blur depends on pixel size. Me: Really?
« Reply #7 on: February 06, 2012, 09:40:51 am »

But motion blur that isn't perceptible for a larger pixel could be perceptible for a smaller one. Isn't this the issue? Trading a shorter shutter duration with higher ISO and smaller pixels versus larger pixel with longer shutter duration and lower ISO. It is not black and white but gray. There will exist a certain amount of motion which makes the resolution gained by using a smaller pixel insignificant, not zero but insignificant.
I believe that the described advantage of larger pixels in a 18 megapixel FF camera can be pretty accurately simulated by taking a 36 megapixel FF image and downscaling it by the appropriate function. The "trail" will still be x mm in a fixed-size print, but the resolution of the image changes. "Larger pixels" will blur the image and serve to hide motion-blur. Not a really big advantage in my book.

Now, larger sensels may or may not have other advantages, but I believe that was not what was argued by Canon here.

-h
« Last Edit: February 06, 2012, 09:42:55 am by hjulenissen »
Logged

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
Me: Motion blur depends on pixel count through degree of enlargement
« Reply #8 on: February 06, 2012, 05:20:07 pm »

What Canon is getting at is that increasing the pixel count, using the same shutter speed, printing/displaying at equal PPI (so, enlarging more) and viewing the result from the same distance (greater apparent size) will make motion blur more visible: not because the "speed lines" are any larger relative to the image but because the greater enlargement makes them more visible.

If instead, one displays and views images at the same size with Canon's 18MP and its un-named 22MP, 24MP or 36MP rivals, motion blur will be equal ... except to the extent that the lower resolution of the 18MP camera obscures it. Yet more "pretend that higher resolution is a disadvantage by comparing images at different degrees of enlargement" nonsense, but this time it has escaped from forum trash talk and contaminated a camera company's marketing department.

It must make fine camera designers wince when their employer's marketing department blows smoke in customers' eyes like this.
Logged

KevinA

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 979
    • Tree Without a Bird
Re: Canon: Motion blur depends on pixel size. Me: Really?
« Reply #9 on: February 09, 2012, 11:43:48 am »

What amazes me about my Canon 1DsmkIII is that if I shoot quicker than 1/2000th I stand a good chance of getting what looks like camera shake. I know other photographers that report the same.
Most noticeable on a batch I shot on a wide angle, I was not paying attention and fired a sequence off at quicker than 1/3000ths, the double image effect looked more like 1/20th. Fortunatly I switched my brain back on and did the rest nearer 1/1200th and these are pin sharp. Canon should sort that out before worrying us about less pixels make a sharper picture.

Kevin.
Logged
Kevin.

Pingang

  • Full Member
  • ***
  • Offline Offline
  • Posts: 117
Re: Canon: Motion blur depends on pixel size. Me: Really?
« Reply #10 on: February 11, 2012, 07:03:08 am »

I think it is at a very limited situation. Typical of Japanese or many non-English country, that could have been wrote better.

Pingang
Logged

clkirksey

  • Newbie
  • *
  • Offline Offline
  • Posts: 19
Re: Me: Motion blur depends on pixel count through degree of enlargement
« Reply #11 on: February 12, 2012, 10:05:52 am »

What Canon is getting at is that increasing the pixel count, using the same shutter speed, printing/displaying at equal PPI (so, enlarging more) and viewing the result from the same distance (greater apparent size) will make motion blur more visible: not because the "speed lines" are any larger relative to the image but because the greater enlargement makes them more visible.

If instead, one displays and views images at the same size with Canon's 18MP and its un-named 22MP, 24MP or 36MP rivals, motion blur will be equal ... except to the extent that the lower resolution of the 18MP camera obscures it. Yet more "pretend that higher resolution is a disadvantage by comparing images at different degrees of enlargement" nonsense, but this time it has escaped from forum trash talk and contaminated a camera company's marketing department.

It must make fine camera designers wince when their employer's marketing department blows smoke in customers' eyes like this.

Motion blur is not something that you would normal want in an image. It degrades resolution. I think that is all Canon is stating. Maybe not in the clearest way based on the translation I do not know. Viewed from the point view of resolution degardation a larger pixel would be better than a smaller pixel. It should be very clear that camera and or subject motion will degrade the resolution of a camera. The smaller the pixel the greater the degradation everything else being equal.
Logged
A bird in the viewfinder...

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
Re: Me: Motion blur depends on pixel count through degree of enlargement
« Reply #12 on: February 12, 2012, 10:23:43 am »

Motion blur is not something that you would normal want in an image. It degrades resolution. I think that is all Canon is stating.
That might be all that Canon is explicitly stating, but in its comments about pixel count, it is clearly implying more: that having a lower pixel count rather than a higher pixel count will give an image quality advantage due to less motion blur. That is utterly false in the only reasonable comparison, which is comparing the rival images at the same size/viewing distance.

People view pictures, not pixels, so this nonsense of comparing images with different amounts of resolution (through different pixel counts) by judging per pixel performance or cropping to equal pixel counts and thus comparing different portions of the image (as in 100% on screen comparisons) is utterly misleading.

A prediction: Canon will soon release a new 36x24mm format camera with a sensor of significantly more than 18MP, and nowhere in the promotional materials will it talk about the alleged disadvantages of smaller pixels.
Logged

AJSJones

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 357
Re: Me: Motion blur depends on pixel count through degree of enlargement
« Reply #13 on: February 12, 2012, 03:15:56 pm »

That might be all that Canon is explicitly stating, but in its comments about pixel count, it is clearly implying more: that having a lower pixel count rather than a higher pixel count will give an image quality advantage due to less motion blur. That is utterly false in the only reasonable comparison, which is comparing the rival images at the same size/viewing distance.

People view pictures, not pixels, so this nonsense of comparing images with different amounts of resolution (through different pixel counts) by judging per pixel performance or cropping to equal pixel counts and thus comparing different portions of the image (as in 100% on screen comparisons) is utterly misleading.

A prediction: Canon will soon release a new 36x24mm format camera with a sensor of significantly more than 18MP, and nowhere in the promotional materials will it talk about the alleged disadvantages of smaller pixels.


I totally agree about the comparisons at same size/viewing distance.  However, there are folks who think more MP = bigger prints and look at those bigger prints more closely - these are the folks for whom the 7D, for example, represents a greater technical challenge due to the greater geometrical enlargement that results from that approach (whether they realize it or not) - here DoF and misfocus and all sorts of blurs will be enlarged more and they say "My 7D is soft" - you've seen the discussions I'm sure.  However, to turn it around and say "We have bigger  pixels therefore motion blur is less of an issue" is some of the purest marketing smoke and mirrors (BS grade) I've seen.

Nonetheless, the recent spate of "superduper" lenses, after the $$$$$uperduper new $upertele$ really does bode well for those with expectations for a higher MP model from Canon.  The smoke will have dissipated by then :D
« Last Edit: February 12, 2012, 07:18:11 pm by AJSJones »
Logged

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
Re: Me: Motion blur depends on pixel count through degree of enlargement
« Reply #14 on: February 12, 2012, 07:05:20 pm »

I totally agree about the comparisons at same size/viewing distance.  However, there are folks who think more MP = bigger prints and look at those bigger prints more closely ...
Indeed, so the correct statement is that to get an advantage from a higher resolution camera, you often have to work harder, including using higher shutter speeds to freeze motion. But if you fail to make sufficient extra effort, the results will not be any worse; they will just fail to be better by the degree hoped for. It constitutes no argument in favor of a lower resolution camera.
Logged
Pages: [1]   Go Up