Reading mode: Light Dark

Anynone care to comment on this hypothesis. bobn2 etc?

Go to last post
Kaj E said:
dtmateojr said:
Kaj E said:
You seem to be hung up on the "downsizing". Why do we not talk about upsizing or enlargement instead, because this is actually what we are doing going to a print from a DX or FX sensor. I doubt that you regularly print images the size of the sensor with these formats.

Let's assume an 8"x12" print from an image at ISO 6400 ,f/2.8, 1/100 sec. from a DX camera and an FX camera for instance (D7100 and D800 which have about the same pixel size and noise/pixel. The image from the D7100 needs to be enlarged more than the image from the D800. Therefore the noise will be more visible on the image from the D7100 than the D800.

The same applies to film. The print from an APS-C film camera would show more grain than that from an 135 format film camera with the same negative film emulsion, enlarged to prints of the same size.

--
Kind regards
Kaj
http://www.pbase.com/kaj_e
WSSA member #13
It's about time we started to take photography seriously and treat it as a hobby.- Elliott Erwitt
Click to expand...
No arguments there if you are comparing a D7100 with stupid 24mp against the mildly sane 36mp D800. However, a 16mp aps-c with the same SNR as a 7-year old D700 will match the output of the D800 even if you enlarge the 16mp to 36mp.
Click to expand...
Absolutely not true for instance in the 8"x12" print case above with a D7000 against a D800.
Click to expand...
I said D700, you were off by an order of magnitude.
Kaj E said:
You are stuck in your believes without a shred of fact. Facts are for instance the normalized DxO values. Normalization is not some kind of a magic trick, it is something that really happens when we make prints of the same size.
Click to expand...
I have covered this in the article COMPLETE WITH MATH.
Kaj E said:
--
Kind regards
Kaj
http://www.pbase.com/kaj_e
WSSA member #13
It's about time we started to take photography seriously and treat it as a hobby.- Elliott Erwitt
Click to expand...
 
dtmateojr said:
Iliah Borg said:
There should be some reason why Nikon are still making those "low" pixel count cameras?
Click to expand...
The Sony A7S is a 12Mp low light monster.
Click to expand...
"Monster" may be a little of a hyperbole, and let's see sales numbers. Right now I see it doing "only" 5 fps (2.5 fps with focusing), but high quality video.
 
dtmateojr said:
Kaj E said:
dtmateojr said:
Kaj E said:
You seem to be hung up on the "downsizing". Why do we not talk about upsizing or enlargement instead, because this is actually what we are doing going to a print from a DX or FX sensor. I doubt that you regularly print images the size of the sensor with these formats.

Let's assume an 8"x12" print from an image at ISO 6400 ,f/2.8, 1/100 sec. from a DX camera and an FX camera for instance (D7100 and D800 which have about the same pixel size and noise/pixel. The image from the D7100 needs to be enlarged more than the image from the D800. Therefore the noise will be more visible on the image from the D7100 than the D800.

The same applies to film. The print from an APS-C film camera would show more grain than that from an 135 format film camera with the same negative film emulsion, enlarged to prints of the same size.
Click to expand...
Click to expand...
Click to expand...
Click to expand...
 
And I'll bet you wish you hadn't linked to it now!

The linked article is rather hard to read. Not technically, I mean the style: it's rather preachy, and one has to grit one's teeth to read past the rant to get to any technical stuff.

Unfortunately, as others have said, his conclusions are quite wrong. Aside from cases where you stop down an FX camera to get equivalent DoF, there is more than 1 stop noise advantage of FX over DX in equivalent picture-taking (and picture-viewing) circumstances, and no amount of sophistry from dtmateojr alters that.
 
Lance B said:
57even said:
Lance B said:
http://dtmateojr.wordpress.com/2014/05/19/megapixel-hallucinations/

--
Lance B
http://www.pbase.com/lance_b
http://i.pbase.com/o4/21/489821/1/53232844.SydneyPanoVertSmall.jpg
Click to expand...
DxO normalises their results to show a meaningful comparison. The fact is I can print 2X the area with a D800 vs a D7000, or shoot at 1 stop higher ISO for the same print size (or take much cleaner long exposures).

If you don't need to do any of those things (and actually most people don't really print anything, least of all at 24" or larger) then APSC is pretty adequate for most purposes. But FX does have roughly 2X the information density of DX, whether you measure that in resolution or S/R ratio. A 16MP FX sensor will have far better colour fidelity at higher ISO than the DX equivalent, and the FX 36MP sensor will have far more resolution and less noise at the same ISO.

And that DOES translate into real world advantages IF you have a need to exploit them.

Of course, if Fuji's organic sensor improves S/R ratio by an order of magnitude, then the game changes until an FF sensor appears with the same technology. Comparisons are only valid for sensors of the same pixel architecture.
Click to expand...
Nice to hear from you and thanks for your input, Steve.

--
Lance B
http://www.pbase.com/lance_b
http://i.pbase.com/o4/21/489821/1/53232844.SydneyPanoVertSmall.jpg
Click to expand...
Hi Lance! I posted this comparison a while ago to show the areas of equivalence and difference between theoretical FX and DX sensors of similar architecture and pixel count, for instance a D7100 and D610.





The green area is where you could, should you need to, gain approximate equivalence between FX and DX in terms of exposure, noise and DOF.

For instance, a D610 at ISO 200 and F2.0 will produce more or less the same image as a D7100 at ISO 100 and F1.4 in terms of noise and DOF with the same shutter speed.

Resolution would of course be the same more or less (assuming the lens was any good).

Theoretically this could actually make a D610 cheaper because I could use F1.8 lenses instead of F1.4 lenses and still achieve marginally better performance than a D7100 at lower overall cost and weight (not something most people think about with FX).

But, if I am shooting at mid-apertures for sufficient DOF, I can always open up one stop more on DX and reduce the ISO by a stop for the same exposure. If I am shutter speed limited, a DX camera is seldom any handicap at all in such scenarios, for instance street photography or hand-held indoor spaces where I am on the shutter speed limit.

If on the other hand I am shooting in good light, or can shoot wide open, or use a tripod, then the FX camera will deliver a stop improvement in noise and subject isolation.

The question is, will you print big enough to notice, given that I am shooting at low ISO anyway and low noise levels at ISO200 can be easily dealt with?
 
Iliah Borg said:
Luke Kaven said:
Iliah Borg said:
There should be some reason why Nikon are still making those "low" pixel count cameras?
Click to expand...
High throughput with low noise through a limited number of channels.
Click to expand...
That would be very poor thinking from the side of sensor engs.
Click to expand...
You said "some reason". That is one reason for doing it that way, n'est ce pas? There are of course further reasons.

They obviously improved a few things along the way. High FWC at base ISO, around 120k e-. Read circuits shared and multiplexed between adjacent columns. Fast, on-chip readout with low thermal noise, limited to 24 channels, yields a high frame rate with a clean read.
 
joejack951 said:
First, equalizing DOF can be anything from using f/1.8 on full frame vs. f/0.95 on m4/3 to your more extreme example at the limits of a lens's f-stop range. In either scenario, when not light-limited, full frame, or any larger format, retains an advantage over a smaller format.
Click to expand...
Obviously when manufacturers don't make f/0.95 or whatever DX lenses equaling an FX counterpart there will be a difference. That may be decisive for your situation, or maybe not.

In my specific case consider the following: I shoot for a living, mainly portraits. I've never been able to sell a portrait taken @ f/1.4 (FX or DX) but need f/5.6 @ 70mm DX or f/8 @105mm FX for enough DOF (nose and eyes in focus, ears OOF).

Shooting @ f/5.6 means I have to spend half the money on flash equipment compared to FX while I spend more on flash equipment than on cameras. Stressing the equipment only half means less wear on the flash tubes and the recycle times are half as well. The latter means I can record the quickly fluctuating face expressions better.

Then the AF point layout of DX is way better for my purposes, because I can keep an AF point on the eye while retaing a full head-and-shoulders image in the viewfinder.

Next DX cameras are way cheaper while no client of mine has ever seen the difference between a shot taken with DX vs FX, prints usually up to A3 (11.69" × 16.54").

While you guys can 'reap' the benefits of FX all you like, for a lot of things I do DX is the better choice. YMWPV ('Your Mileage Will Probably Vary') but be sure not to get blinded by which is, in the end, a maximum of one stop difference. And this comes from a guy who used 4x5" film cameras almost exclusively for the first 15 years of his photographic working life.
 
Luke Kaven said:
Iliah Borg said:
Luke Kaven said:
Iliah Borg said:
There should be some reason why Nikon are still making those "low" pixel count cameras?
Click to expand...
High throughput with low noise through a limited number of channels.
Click to expand...
That would be very poor thinking from the side of sensor engs.
Click to expand...
You said "some reason". That is one reason for doing it that way, n'est ce pas?
Click to expand...
It is a complex reason, including that Nikon are still very proud of D2H sensor ;)
Luke Kaven said:
They obviously improved a few things along the way. High FWC at base ISO, around 120k e-. Read circuits shared and multiplexed between adjacent columns. Fast, on-chip readout with low thermal noise, limited to 24 channels, yields a high frame rate with a clean read.
Click to expand...
Having a stronger signal and averaging signal out before noise is added through digitizing.

> limited to 24 channels

Ummm....
 
Iliah Borg said:
dtmateojr said:
Luke Kaven said:
Kaj E said:
dtmateojr said:
And I quote again,

"There is an advantage to big pixels in low light (high ISO) applications, where read noise is an important detractor from image quality, and big pixels currently have lower read noise than aggregations of small pixels of equal area."
Click to expand...
Yes, present cameras with larger pixels have been optimized for low light. It is a question of technology not just sensor size. And these cameras also have large sensor because larger sensors gather more light.
Click to expand...
In this generation (e.g., the D4), it has nothing to do with the size of the pixel. The pixels on the D4 are large because the sensor architecture needs to push the entire frame through a 24 channel readout. Correlation with pixel size is an illusory correlation. The Exmor pixels are at least as good at high gain settings, except for a greater difficulty with thermal noise, and a slower readout.
Click to expand...
Don't you think that an Exmor sensor would benefit from a larger pixel size?
Click to expand...
The question is not about the sensor but about the sales. So much money is invested into making pixel count the single most important parameter that only few consumers look at any other parameters making their decisions between the cameras of similar size and weight.
Click to expand...
You know, you could have politely answered dtmateojr's question first as "yes" before suggesting your own.

Now, as to the question about the market for a low density Exmor sensor, there's definitely one to be tapped. Clearly there's a user base here who's NOT upgraded to the D800 because they are more concerned about image quality, file size, and fps than resolution.

If Nikon were to offer a modern 12 MP D750 Noct with a stronger CFA and equivalent handling specs to the D700 and market it as the company's specialized camera for ultra-high color fidelity & low light use, that thing would fly off the shelves. Certainly Sony with the A7s has realized there's an untapped market for a lower density sensor oriented towards image quality (and in their marketing twist, video quality).

fPrime
 
Last edited:
Luke Kaven said:
Iliah Borg said:
Luke Kaven said:
Iliah Borg said:
There should be some reason why Nikon are still making those "low" pixel count cameras?
Click to expand...
High throughput with low noise through a limited number of channels.
Click to expand...
That would be very poor thinking from the side of sensor engs.
Click to expand...
You said "some reason". That is one reason for doing it that way, n'est ce pas? There are of course further reasons.

They obviously improved a few things along the way. High FWC at base ISO, around 120k e-. Read circuits shared and multiplexed between adjacent columns. Fast, on-chip readout with low thermal noise, limited to 24 channels, yields a high frame rate with a clean read.
Click to expand...
Or another potential, using a more color selective (albeit more opaque) CFA over the sensor to boost color metamersim back to D700/D3 levels?

fPrime
 
fPrime said:
Iliah Borg said:
dtmateojr said:
Luke Kaven said:
Kaj E said:
dtmateojr said:
And I quote again,

"There is an advantage to big pixels in low light (high ISO) applications, where read noise is an important detractor from image quality, and big pixels currently have lower read noise than aggregations of small pixels of equal area."
Click to expand...
Yes, present cameras with larger pixels have been optimized for low light. It is a question of technology not just sensor size. And these cameras also have large sensor because larger sensors gather more light.
Click to expand...
In this generation (e.g., the D4), it has nothing to do with the size of the pixel. The pixels on the D4 are large because the sensor architecture needs to push the entire frame through a 24 channel readout. Correlation with pixel size is an illusory correlation. The Exmor pixels are at least as good at high gain settings, except for a greater difficulty with thermal noise, and a slower readout.
Click to expand...
Don't you think that an Exmor sensor would benefit from a larger pixel size?
Click to expand...
The question is not about the sensor but about the sales. So much money is invested into making pixel count the single most important parameter that only few consumers look at any other parameters making their decisions between the cameras of similar size and weight.
Click to expand...
You know,
Click to expand...
You have some questions to answer, you know :)
fPrime said:
you could have politely answered dtmateojr's question first as "yes"
Click to expand...
The answer meanwhile is "no".

--

About LibRaw | LibRaw

www.libraw.org www.libraw.org
 
fPrime said:
Luke Kaven said:
Iliah Borg said:
Luke Kaven said:
Iliah Borg said:
There should be some reason why Nikon are still making those "low" pixel count cameras?
Click to expand...
High throughput with low noise through a limited number of channels.
Click to expand...
That would be very poor thinking from the side of sensor engs.
Click to expand...
You said "some reason". That is one reason for doing it that way, n'est ce pas? There are of course further reasons.

They obviously improved a few things along the way. High FWC at base ISO, around 120k e-. Read circuits shared and multiplexed between adjacent columns. Fast, on-chip readout with low thermal noise, limited to 24 channels, yields a high frame rate with a clean read.
Click to expand...
Or another potential, using a more color selective (albeit more opaque) CFA over the sensor to boost color metamersim back to D700/D3 levels?
Click to expand...
Boosting colour metamerism is exactly the opposite direction. One wants to reduce it, or, rather, reduce metameric failure. Metamerism occurs when you see two samples match under one light source and not match under another. In digital photography metameric failure manifests itself mostly as a colour transform calculated for one illuminant fails in colour reproduction (with deltaE2000 > 6 for significant number of memory colours) if the shot is taken under a different illuminant. I suggest that you need to explain how is this connected to CFA being more opaque or more transparent.
 
Iliah Borg said:
Iliah Borg said:
In fact, as discussed thoroughly here even as of a few days ago, the D800 chroma noise is so strong that it actually shifts the color balance of the image.
Click to expand...
That's what you are doing. Why didn't you check the source of noise?

You are displaying processed images and try to make conclusions based not on the raw data, but on the capabilities of raw processing software (and the weakest link there is the accuracy of the colour transform, as I demonstrated on the images in the post you did not know how to answer).
Click to expand...
It didn't escape me that you finally posted an image to make your point! Is this a new Iliah? :D

Unfortunately it seems like any thread we engage on turns out to be hugely popular and before I could reply it had reached the 150 post max.
Iliah Borg said:
ba8adac91d454f0ba4355488b9cef8b0.jpg
Click to expand...
We know Dpreview is standardized on ACR's Adobe Standard processing. For better or worse, that at least ties down one variable for making cross camera comparisons. And as we saw, CNX2 duplicated the phenomena making it not likely to be RAW convertor related. Of course, that presumes we're not comparing the relative strength and proficiency of the convertor's default chroma noise reduction since the green shift is green channel noise related.

Because you provided converted TIFF's and didn't specify any of the conversion or processing steps behind them it's not really possible for anyone to comment on them other than to acknowledge that they look different from Dpreview's conversions. Please let us know how/what produced this crop below and perhaps we can comment intelligently.

fPrime
Iliah Borg said:
While D800 raw looks like this:

57cc548ad2494d9d973b96df1bddeae9.jpg.png

What noise you see on the above that throws off colour balance?

--

http://www.libraw.org/
Click to expand...
 
You must log in or register to reply here.

Keyboard shortcuts

f
Forum
m
My threads
Mobile site
About
Editorial content
Cameras & Lenses
Community
All content, design, and layout are Copyright © 1998–2025 Digital Photography Review All Rights Reserved.
Reproduction in whole or part in any form or medium without specific written permission is prohibited.
When you use DPReview links to buy products, the site may earn a commission.
©GPS Media - Guides, Products, Services.
Back
Top