Forgot Password
Pentax Camera Forums Home
 

Reply
Show Printable Version Search this Thread
07-12-2015, 10:44 PM   #31
Otis Memorial Pentaxian
Otis FanOtis FanOtis FanOtis FanOtis FanOtis Fan
Loyal Site Supporter
clackers's Avatar

Join Date: Jul 2013
Location: Melbourne
Photos: Albums
Posts: 16,394
QuoteOriginally posted by bossa Quote
The amount of light captured (brightness) depends on the size of the sensor relative to the the exit pupil of the lens. If you don't believe this then experiment with a magnifying glass, making large and small projections on a piece of paper, and you will find that the smaller the projection, the brighter the image.
That's a lovely tortoiseshell you own, Bossa, but your analogy is off.


The magnifying glass has a fixed aperture.


By moving it back and forth (eventually to a single bright point) what you're doing is focussing.


Last edited by clackers; 07-12-2015 at 10:50 PM.
07-12-2015, 11:44 PM   #32
Veteran Member




Join Date: Jan 2012
Photos: Gallery
Posts: 604
This equivalence thingy doesn't work with film. It's not photography.
07-12-2015, 11:53 PM   #33
Inactive Account




Join Date: Sep 2009
Location: Planet Earth, Sol system, Milky Way galaxy, Universe
Posts: 1,119
QuoteOriginally posted by clackers Quote
ISO performance is a function of the size of pixels ("water wells"), the electronics, and the number of pixels used to resample (software noise reduction), not sensor size.
Ok, but if you increase sensor size while keeping pixel size the same (thus adding pixels), wouldn't the noise - generally speaking - be more fine-grained? I mean, if I take a photo at ISO 3200 with my K-3 and look at the whole photo, it may be fine. But when I start cropping out some parts, I'm magnifying the rest, and I can see the noise more clearly.

And if you increase sensor size while keeping the number of pixels the same (thus increasing pixel size), wouldn't the wells be bigger in the first place, so less susceptible to noise? I mean, the Pentax K-5 and Nikon D4 are both 16MP. But noise at high ISO is significantly less with the D4.

I mean, there is a reason small-sensored cameras generally don't have the best ISO performance, right? I'm getting a Q10. Maybe I can use that at ISO 800. I hope. With a K-5, that's a no-brainer.
07-13-2015, 12:12 AM   #34
Veteran Member




Join Date: Jan 2012
Photos: Gallery
Posts: 604
Equivalent focal length with crop factor

Equivalence has this crazy assumption that you could just bump the ISO indefinitely. Truth is, equivalence doesn't scale upwards or downwards.

Compare the same m43 and FF sensors. If the sensor has a native ISO limit of 6400 it means that FF will have to push towards fake ISO 25,000. It doesn't scale up.

Now in the case of very low light, if the m43 sensel has captured 2 photons, the 2-stop underexposure of FF means it will not capture any photon. Bumping the ISO does not do anything because there is no signal to bump. It does not scale downwards.

Equivalence is wrong, plain and simple.

07-13-2015, 04:14 AM   #35
bxf
Veteran Member
bxf's Avatar

Join Date: Jul 2010
Location: Lisbon area
Posts: 1,660
QuoteOriginally posted by clackers Quote
Aperture is a function of the lens, not the sensor size...
I can see the logic of what you say, but, again I have to wonder why DXO high ISO scores are consistently significantly better for FF bodies.
07-13-2015, 04:24 AM   #36
Veteran Member
bossa's Avatar

Join Date: Jul 2011
Location: Adelaide
Posts: 4,546
QuoteOriginally posted by bxf Quote
I can see the logic of what you say, but, again I have to wonder why DXO high ISO scores are consistently significantly better for FF bodies.
Dx0 "normalizes" all images to 8MP (IIRC) and a lot of noise gets swallowed up by that process.
07-13-2015, 04:30 AM   #37
Veteran Member




Join Date: Jan 2012
Photos: Gallery
Posts: 604
QuoteOriginally posted by bxf Quote
I can see the logic of what you say, but, again I have to wonder why DXO high ISO scores are consistently significantly better for FF bodies.

DXO ranks sensors. It normalises to a 8x10 print size. Downsizing results in SNR improvement.

If you want to see the REAL noise performance look at the screen SNR graphs. Compare the K5, D7000 and D800 and you will see that they have exactly the same SNR because they use the same sensor.

---------- Post added 07-13-15 at 21:35 ----------

Normalised D7000 vs D800



REAL SNR D7000 vs D800



07-13-2015, 04:45 AM   #38
bxf
Veteran Member
bxf's Avatar

Join Date: Jul 2010
Location: Lisbon area
Posts: 1,660
QuoteOriginally posted by bossa Quote
Dx0 "normalizes" all images to 8MP (IIRC) and a lot of noise gets swallowed up by that process.
QuoteOriginally posted by dtmateojr Quote
DXO ranks sensors. It normalises to a 8x10 print size. Downsizing results in SNR improvement...
Ah, got it. Thanks guys.

I've been wondering about this for some time, as I saw the DXO scores as definitive evidence of FF's noise superiority, and hence the validity of using increased ISO in equivalence arguments.
07-13-2015, 05:28 AM - 1 Like   #39
Veteran Member




Join Date: Dec 2010
Location: Ontario
Photos: Gallery
Posts: 3,332
QuoteOriginally posted by dtmateojr Quote
If you want to see the REAL noise performance look at the screen SNR graphs. Compare the K5, D7000 and D800 and you will see that they have exactly the same SNR because they use the same sensor.
It's interesting that the one you've decide to label as "REAL" is the 100% view, pixel peeping version. What happens at the pixel level as technology has improved is interesting but for me it's way less important than a comparison of what the photos will look like as a whole when blown up to a given size - I can comfortably predict that if a digital FF comes my way I won't suddenly start printing 1.5 times the size I currently do.
07-13-2015, 07:59 AM   #40
Veteran Member




Join Date: Nov 2012
Photos: Albums
Posts: 963
QuoteOriginally posted by kh1234567890 Quote
I love these equivalence threads ...



Now, where's the popcorn smiley when you need one?
07-13-2015, 09:47 AM - 1 Like   #41
Veteran Member
audiobomber's Avatar

Join Date: May 2008
Location: Sudbury, Ontario
Photos: Albums
Posts: 6,806
QuoteOriginally posted by bxf Quote
I've been wondering about this for some time, as I saw the DXO scores as definitive evidence of FF's noise superiority, and hence the validity of using increased ISO in equivalence arguments.
All of the above still holds. DXOMark PRINT VIEW is the definitive determinant of sensor performance. FF has lower noise than smaller sensors, when viewing at equal resolution, or when printing to the same size. Viewing noise at pixel level, i.e. SCREEN VIEW, is virtually meaningless.
07-13-2015, 10:07 AM   #42
bxf
Veteran Member
bxf's Avatar

Join Date: Jul 2010
Location: Lisbon area
Posts: 1,660
QuoteOriginally posted by audiobomber Quote
All of the above still holds. DXOMark PRINT VIEW is the definitive determinant of sensor performance. FF has lower noise than smaller sensors, when viewing at equal resolution, or when printing to the same size. Viewing noise at pixel level, i.e. SCREEN VIEW, is virtually meaningless.
Right, understood. Thanks for the "reminder".
07-13-2015, 10:14 AM   #43
Inactive Account




Join Date: Sep 2009
Location: Planet Earth, Sol system, Milky Way galaxy, Universe
Posts: 1,119
QuoteOriginally posted by dtmateojr Quote
Equivalence has this crazy assumption that you could just bump the ISO indefinitely. Truth is, equivalence doesn't scale upwards or downwards.

Compare the same m43 and FF sensors. If the sensor has a native ISO limit of 6400 it means that FF will have to push towards fake ISO 25,000. It doesn't scale up.
Why not? The only difference between bumping ISO in the native range and bumping it further is where the bumping takes place: in the former case the analog signal gets amplified before conversion to digital. In the latter case, the digital signal gets "amplified" (technically, the signal is not amplified, because it's a series of ones and zeroes, but the digital signal is changed so that the outcoming ones and zeroes represent a bumped signal). So why does it matter where the bumping takes place?

Bumping outside of the native range does introduce a little more noise I think, because the D/A conversion itself will also introduce a bit of extra noise, which otherwise would not get amplified.

Anyway, we are dealing with a very specific corner case here, in which one system does not get bumped into non-native ISO and the other does. Surely you must see that.

QuoteOriginally posted by dtmateojr Quote
Now in the case of very low light, if the m43 sensel has captured 2 photons, the 2-stop underexposure of FF means it will not capture any photon. Bumping the ISO does not do anything because there is no signal to bump. It does not scale downwards.
So you are saying that, in some cases, you cannot bump the ISO on a fullframe sensor? Really?

QuoteOriginally posted by dtmateojr Quote
Equivalence is wrong, plain and simple.
Speak for yourself. If anything, I'm arriving at the conclusion that this is not plain and simple at all.

I'm also a little puzzled how it is possible that the D7000 and D800 use the same sensor, yet the one in the D800 is 2.25 times as big.



---------- Post added 07-13-2015 at 07:30 PM ----------

QuoteOriginally posted by audiobomber Quote
All of the above still holds. DXOMark PRINT VIEW is the definitive determinant of sensor performance. FF has lower noise than smaller sensors, when viewing at equal resolution, or when printing to the same size. Viewing noise at pixel level, i.e. SCREEN VIEW, is virtually meaningless.
What's also interesting to see is that the difference is about 4dB. 3dB means a signal twice as strong. The sensor area is 2.25 times as big on fullframe, so if it were equivalent, we'd expect a difference of about 3.5dB. Reality is not far off, at least in this case.

Last edited by starbase218; 07-13-2015 at 10:32 AM.
07-13-2015, 10:33 AM   #44
Pentaxian
normhead's Avatar

Join Date: Jun 2007
Location: Near Algonquin Park
Photos: Gallery | Albums
Posts: 40,442
QuoteQuote:
All of the above still holds. DXOMark PRINT VIEW is the definitive determinant of sensor performance.
I think you need to read the small print really closely to unsderstand what DxO determines.

QuoteQuote:
FF has lower noise than smaller sensors, when viewing at equal resolution, or when printing to the same size.
What about equalizing DoF? The thing the people who make these kinds of statements always leave out.

Let use the K-3 and D750 for examples.....



To get to the same DoF on the D750 you have on the K-3 you'd have to shoot stop slower aperture.

Pixel peeping you get this, the K-3 at 200 ISO, the D750 at 400...


You see the problem, there is no discernible noise in either picture. SO when does this theoretical noise difference kick in? DxO doesn't define the parameters.

QuoteQuote:
Viewing noise at pixel level, i.e. SCREEN VIEW, is virtually meaningless.
You mean it doesn't corroborate DxO's pointless propaganda that's main function is to sell their software? What does that make meaningless? Are you actually saying, don't look at images, listen to DxO and me. We'll tell you what's real?

I'm sorry, but I'm looking at these images taken under studio conditions, and what I'm seeing is no discernible difference in noise and the APS-c image looking sharper because of wider DoF. Now you are spouting some corporate DxO theoretical measurements, that don't mean anything in the real world, to the point you actually advise us not to pixel peep. Because reality shows how insignificant these differences are in the real world.

As for the nonsense that 24 mp of APS-c imagery is somehow different than 24 MP of FF imagery. Look at the images above. What is the difference? Blown up to the same size the APS-c still has the DoF advantage, so what are you talking about?

Where is this magical point where the APS-c image is going to suffer when being printed? All of the sudden those fine looking pixels are going to deteriorate, because they are APS-c pixels, not FF pixels? It's a fabrication. Best case scenario it's an exaggeration, worst case scenario, an outright lie. In both cases, it's not done in the name of science, it's done in the proud tradition of making up scientific mumbo jumbo and double speak, to sell product.

Last edited by normhead; 07-13-2015 at 11:01 AM.
07-13-2015, 10:52 AM   #45
Veteran Member
audiobomber's Avatar

Join Date: May 2008
Location: Sudbury, Ontario
Photos: Albums
Posts: 6,806
Norm, equivalence *always* begins with equalized DOF. Anything else is a non-starter.
Reply

Bookmarks
  • Submit Thread to Facebook Facebook
  • Submit Thread to Twitter Twitter
  • Submit Thread to Digg Digg
Tags - Make this thread easier to find by adding keywords to it!
body, crop, d7000, d800, distance, dof, dxo, equivalent, equivalent focal length, ff, fullframe, images, iso, k-mount, length with crop, lens, magnification, noise, pentax lens, scale, sensor, sensors, shot, slr lens, snr, view, vs
Thread Tools Search this Thread
Search this Thread:

Advanced Search


Similar Threads
Thread Thread Starter Forum Replies Last Post
Bird shooting: longer focal length or crop with an affordable prime? Sigmund Troubleshooting and Beginner Help 12 06-21-2015 06:59 PM
Focal length = min. shutter speed = ff crop? rzarector Troubleshooting and Beginner Help 14 03-21-2015 05:00 AM
Quick Method: What Is The Equivalent Focal Length DavidSKAF3 Pentax SLR Lens Discussion 21 12-15-2013 09:43 PM
Crop Factor, Focal Length and Field of View Ole Pentax Lens Articles 15 05-26-2013 12:41 PM
Setting up focal point for manual lenses. consider crop factor? D4rknezz Troubleshooting and Beginner Help 7 09-23-2009 01:00 PM



All times are GMT -7. The time now is 07:01 AM. | See also: NikonForums.com, CanonForums.com part of our network of photo forums!
  • Red (Default)
  • Green
  • Gray
  • Dark
  • Dark Yellow
  • Dark Blue
  • Old Red
  • Old Green
  • Old Gray
  • Dial-Up Style
Hello! It's great to see you back on the forum! Have you considered joining the community?
register
Creating a FREE ACCOUNT takes under a minute, removes ads, and lets you post! [Dismiss]
Top