Pages: 1 [2]   Go Down

Author Topic: 7D Hands on  (Read 14646 times)

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
7D Hands on
« Reply #20 on: September 05, 2009, 01:29:02 am »

Quote from: Wayne Fox
Isn't that more a statement about where we are in the evolvement of digital capture than anything else?  So far I see lots of ideas about the next big thing (ex. computational photography), but until that shows up we many not have much left but incremental changes.

I think it's more about marketing strategy. I'm essentially a Canon guy, although I also own a D700 and Nikkor 14-24 lens because the lens was so darned good I couldn't resist. But I'm not entirely happy about struggling with two different systems with different buttons to push.

In restrospect, I now feel I should have been less impetuous and opted for a 5D2 with the new TS-E 17mm which of course wasn't available at the time I bought the D700. Such is life!

The 14-24mm Nikkor is a different beast to the 17mm TS-E, but the fact remains, if a little birdie had whispered in my ear that Canon would soon announce a 17mm TS-E of exceptional quality, I would not have bought the D700 and Nikkor lens.

Like many of us, I'm a victim of the 'want now' syndrome.

Nevertheless, the D700 still retains some advantages over any Canon, such as, up to 9 frames exposure autobracketing, and full autobracketing of ISO in manual mode.
Logged

JohnKoerner

  • Guest
7D Hands on
« Reply #21 on: September 05, 2009, 09:09:19 am »

Quote from: Chris Pollock
The last APS-C camera that I used was the 20D, which as I mentioned was already limited by some lenses. What kind of results are 50D users getting? Are many lenses able to outresolve the sensor across the frame? I expect that the better primes could, but what about zooms?


Your understanding is exactly opposite to the reality. The reality is the 50D has been able to out-resolve some of the lower-end Canon lenses, showing their faults more clearly, while at the same time actually improving the results previously gained in the higher-end lenses.




.
« Last Edit: September 05, 2009, 09:11:59 am by JohnKoerner »
Logged

Slough

  • Guest
7D Hands on
« Reply #22 on: September 05, 2009, 01:01:51 pm »

Quote from: JohnKoerner
Your understanding is exactly opposite to the reality. The reality is the 50D has been able to out-resolve some of the lower-end Canon lenses, showing their faults more clearly, while at the same time actually improving the results previously gained in the higher-end lenses.

I think he actually said what you just said. I'm sure he'll confirm.
Logged

telyt

  • Guest
7D Hands on
« Reply #23 on: September 05, 2009, 03:21:54 pm »

Quote from: DarkPenguin
I think it was bob atkins that checked some pretty cheap lenses against a 50D.  There were gains to be had even with low end glass.

Did he actually test them, or is this another of his guesses?
Logged

DarkPenguin

  • Guest
7D Hands on
« Reply #24 on: September 05, 2009, 03:24:15 pm »

Quote from: telyt
Did he actually test them, or is this another of his guesses?

There were pictures and everything.
Logged

Wayne Fox

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4237
    • waynefox.com
7D Hands on
« Reply #25 on: September 05, 2009, 04:44:25 pm »

Quote from: Chris Pollock
All other things being equal, reducing the size of the pixels makes per-pixel noise worse, and also tends to reduce dynamic range. It's simple physics - a smaller pixel collects less light, and therefore provides a weaker signal, which tends to result in a lower signal to noise ratio.


You are right, but all other things are rarely equal. Improved noise performance is almost always associated with new sensors.  I think it's a little more complicated than simple physics ... yes a smaller sensor site allows less light in, but there are ways to reduce the electronic noise so that site can still accurately measure the light in relation to other sites.  I realize some noise has nothing to do with the electronics, but I think the electronic noise of the sensor is still the biggest challenge.

Also curious (my logic may be messed up here), but even if noise/pixel remains the same, if there are more pixels to record the detail wouldn't that mean you have less noise as it relates to detail?  Perhaps to realize that you would have to down rez the fie.  Can't quite get my head around this one ...
Logged

ejmartin

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 575
7D Hands on
« Reply #26 on: September 05, 2009, 05:09:50 pm »

Quote from: Wayne Fox
Quote
(Chris Pollock @ Sep 4 2009, 10:33 PM) *
All other things being equal, reducing the size of the pixels makes per-pixel noise worse, and also tends to reduce dynamic range. It's simple physics - a smaller pixel collects less light, and therefore provides a weaker signal, which tends to result in a lower signal to noise ratio.

You are right, but all other things are rarely equal. Improved noise performance is almost always associated with new sensors.  I think it's a little more complicated than simple physics ... yes a smaller sensor site allows less light in, but there are ways to reduce the electronic noise so that site can still accurately measure the light in relation to other sites.  I realize some noise has nothing to do with the electronics, but I think the electronic noise of the sensor is still the biggest challenge.

Also curious (my logic may be messed up here), but even if noise/pixel remains the same, if there are more pixels to record the detail wouldn't that mean you have less noise as it relates to detail?  Perhaps to realize that you would have to down rez the fie.  Can't quite get my head around this one ...

Actually, he is only right on the level of the individual pixel.  A smaller pixel collects less light, has lower S/N, and lower DR.  That does not necessarily mean a decrease in image quality however.  The point is that though the individual pixel is smaller and noisier, there are more of them, so each one of them doesn't need to do as much work in creating the image.  If we fix a given scale in the image (ie fix a particular spatial frequency, like one does for MTF graphs), then the noise at that spatial frequency is largely unchanged as pixel size varies.  What is happening at the pixel level is that noise is a rising function of spatial frequency, and smaller pixels sample higher spatial frequencies.  But noise at a fixed frequency is largely unchanged.

More precisely, there are two kinds of noise typically.  One is photon noise, and this is completely independent of pixel size since it is a property of the photons themselves:
http://forums.dpreview.com/forums/read.asp...essage=31922352
http://forums.dpreview.com/forums/read.asp...essage=31922793

The other kind of noise is electronic read noise.  There are many variables here.  IF one assumes that the read noise per pixel is fixed, then smaller pixels contribute more read noise since there are more pixels.  But read noise has been decreasing over time, and that allows a commensurate decrease in pixel size (though Canon has tended to decrease pixel size faster than they have read noise in proportion).

Electronic read noise in the 7D is pretty spectacular.  A very rough preliminary estimate gave about 1.8-1.9 electrons (compared to 2.5-2.6 for the 50D and 5D2).  What is also impressive is that Canon has finally beaten down the pattern noise that has been a problem for a long time.
Logged
emil

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
7D Hands on
« Reply #27 on: September 05, 2009, 05:38:54 pm »

Quote from: ejmartin
Electronic read noise in the 7D is pretty spectacular.  A very rough preliminary estimate gave about 1.8-1.9 electrons (compared to 2.5-2.6 for the 50D and 5D2).
Emil,

I am looking forward to details on the trend in noise levels vs ISO speed ... if possibly in actual measured units, not the fictional electrons which only really makes sense in reference to noise present before variable gain for ISO adjustment.

Can you think of a way that testing could help decide whether the 7D is applying ISO gain "ASAP": with a variable charge gain in the transfer of the signal from photosite to a capacitor at the bottom of the row? This approach is described in a Canon research paper on an experimental sensor of 50MP in "1D" format, and seems optimal for minimizing the effect of noise on high ISO images.

P. S. Keep up the good work on explaining "per pixel" vs "per image" noise levels!
Logged

Chris Pollock

  • Full Member
  • ***
  • Offline Offline
  • Posts: 206
7D Hands on
« Reply #28 on: September 05, 2009, 06:54:41 pm »

Quote from: JohnKoerner
Your understanding is exactly opposite to the reality. The reality is the 50D has been able to out-resolve some of the lower-end Canon lenses, showing their faults more clearly, while at the same time actually improving the results previously gained in the higher-end lenses.
How exactly is my understanding the opposite of reality? As I said, the 20D was already limited by some lenses, so I expected lens limitations to be more of a problem with the higher resolution 50D. I also stated that I expected some lenses (the better primes at least) to still be able to outresolve the sensor. Isn't that pretty much what you're saying above? Exactly what do you disagree with me about?
Logged

JohnKoerner

  • Guest
7D Hands on
« Reply #29 on: September 07, 2009, 07:35:45 am »

Quote from: Chris Pollock
How exactly is my understanding the opposite of reality? As I said, the 20D was already limited by some lenses, so I expected lens limitations to be more of a problem with the higher resolution 50D. I also stated that I expected some lenses (the better primes at least) to still be able to outresolve the sensor. Isn't that pretty much what you're saying above? Exactly what do you disagree with me about?


Upon re-reading your post, yes, that is what I was saying. On my first read, I only saw the part about the lenses outresolving the sensor ... when the problem was that the new sensor was outresolving many of the lenses. So my bad


.
Logged

Dan Wells

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1044
7D Hands on
« Reply #30 on: September 07, 2009, 10:22:33 pm »

The 7D looks like a "Canon D300" - Canon finally deciding to follow Nikon into the semipro APS market. If they got the image quality right (low noise, good DR), this will be a wildlife shooter's dream - ridiculously high pixel density, small sensor increases long lens crop. On the other hand, I'm worried about that pixel density - the 50D is not exactly the leader in per-pixel IQ, and this one's denser still. I wonder if Canon wouldn't have been better served by releasing the same camera with a really superb 12mp sensor? Something that was only 12mp, but had a D3x level of per-pixel IQ would have STILL made 16x24 inch prints (since the D3x goes 24x36!), and could have done that with enormous dynamic range and very low noise. I suspect Nikon had a very valid reason for NOT increasing pixel count on the D300s - they probably couldn't do it without other IQ compromises... Canon has historically been able to slightly better Nikon's pixel densities, but I'm not sure I believe more than 1.5x the density without sacrificing quality!

                -Dan
Logged

BernardLanguillier

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 13983
    • http://www.flickr.com/photos/bernardlanguillier/sets/
7D Hands on
« Reply #31 on: September 07, 2009, 11:27:34 pm »

Quote from: Dan Wells
I suspect Nikon had a very valid reason for NOT increasing pixel count on the D300s - they probably couldn't do it without other IQ compromises... Canon has historically been able to slightly better Nikon's pixel densities, but I'm not sure I believe more than 1.5x the density without sacrificing quality!

Focusing on enhancing the pixel quality at equal pixel count might have been good, but Canon might have been able to maintain a similar pixel quality while increasing the pixel count a bit. As far as Nikon goes, I am not sure whether this is the reason why Nikon kept the D300s at 12MP. I think that it has more to do with their minor/major design cycles and their basic intention to commonalize sensors accross bodies to reduce unit costs.

On the other hand I do indeed believe that Nikon marketing is now positioning Nikon as the company with the strongest focus on image qualities other than MP (DR, pixel sharpness,...), as the d3x marketing campains clearly showed. From that standpoint, I wouldn't be surprised if they tried to keep a pixel count below that of the corresponding Canon from now on...

Nonetheless, pixel still appear to be considered as the #1 metrics of a camera and a 7D at 18MP vs a G11 at 10MP is easier to understand for consumers than a 7D at 12MP vs a G11 at 16MP...

Regards,
Bernard

KevinA

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 979
    • Tree Without a Bird
7D Hands on
« Reply #32 on: September 08, 2009, 10:29:35 am »

Quote from: Wayne Fox
From everything I've read, increasing pixel count is a way to improve noise performance ... isn't that a major factor in improving dynamic range?  Even if you surpass the ability of the lens, so what?  shouldn't the lens itself be the limiting factor, not the sensor?  Increasing resolution doesn't make that worse ... once you hit the limit you hit the limit.  Until you hit that limit it seems sharpness and quality are perhaps just an artificial perception ... the sensor has placed an artificial limit.

I guess what I'm trying to rationalize in my own mind are two points ... passing the limit of the lens isn't necessarily a negative thing, and sensor density is about more than just resolution.

That's how I see it also. Plus even if the lens bottoms out before the sensor it's still better to make enlargements or crops from actual pixels rather than computer added ones.

Kevin.
Logged
Kevin.

KevinA

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 979
    • Tree Without a Bird
7D Hands on
« Reply #33 on: September 08, 2009, 10:33:06 am »

Quote from: KevinA
That's how I see it also. Plus even if the lens bottoms out before the sensor it's still better to make enlargements or crops from actual pixels rather than computer added ones.

Kevin.

Which also equates to me having a cropped to 320 mm f2:8 lens when I fit my 70 - 200m onto the 7D, which would be a big factor in my purchase of a 7D.

Kevin.
Logged
Kevin.

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
7D Hands on
« Reply #34 on: September 08, 2009, 11:11:56 am »

Quote from: DarkPenguin
I think it was bob atkins that checked some pretty cheap lenses against a 50D.  There were gains to be had even with low end glass.
Thanks for that report.

And surely, if one has bought some good lenses for the sake of getting sharp, detailed, high resolution images, it would be reasonable to seek a body with a sensor that get as much as possible out of the best of those lenses, not just one that hits the resolution limits of entry level zooms. So it makes sense for Canon to offer at least one EF-S body that can get as much as possible out of the best lenses likely to be used on it.

Ditto for EF mount bodies, so 30MP or even 38MP is probably not that far away. (Pixel sizes as in the 12MP and 15MP EF-S sensors.)
Logged

ejmartin

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 575
7D Hands on
« Reply #35 on: September 08, 2009, 11:39:08 am »

Quote from: BJL
Emil,

I am looking forward to details on the trend in noise levels vs ISO speed ... if possibly in actual measured units, not the fictional electrons which only really makes sense in reference to noise present before variable gain for ISO adjustment.

The numbers you are looking for were measured by Panopeeper:
http://www.panopeeper.com/Noise/ReadNoise.txt
Logged
emil

JohnKoerner

  • Guest
7D Hands on
« Reply #36 on: September 08, 2009, 02:19:45 pm »

Quote from: Dan Wells
If they got the image quality right (low noise, good DR), this will be a wildlife shooter's dream - ridiculously high pixel density, small sensor increases long lens crop.

This is what I am hoping will be the case with this camera ...




Quote from: Dan Wells
On the other hand, I'm worried about that pixel density - the 50D is not exactly the leader in per-pixel IQ, and this one's denser still.               -Dan

Apples & oranges though.

The 50D doesn't have the same technology (nor the same dual sensors) as the 7D ... nor does it have the same weather-sealing ... nor the same LCD ... etc., etc. ... so I don't think any of the limitations of the 50D can be applied to the 7D.

If anything, it seems like Canon really listened, and really took as many of the semi-pro limitations as it could seriously, and addressed them all on just about as many levels as is possible, and turned them into strengths in the 7D.

Soon we shall see,

Jack


.
« Last Edit: September 08, 2009, 02:20:57 pm by JohnKoerner »
Logged
Pages: 1 [2]   Go Up