Forgot Password
Pentax Camera Forums Home
 

Reply
Show Printable Version Search this Thread
04-18-2014, 08:26 PM   #226
Veteran Member
jsherman999's Avatar

Join Date: Dec 2007
Photos: Gallery
Posts: 8,228
QuoteOriginally posted by dtmateojr Quote
Well that's why I chose a 12Mp full frame D700 over D800. For most photographers, 12Mp is gigantic! I have printed 1m wide with 12Mp and it looks fantastic. The largest print I've seen for 12Mp is FIVE STOREYS high!!!
Something 5 stories high is meant to be seen from blocks away, and/or is upsampled by some professional software like Genuine Fractals. In Mpls here we reportedly had an iphone shot on the side of a building - looked just fine from most of downtown.

If you want to be picky and maintain 300dpi and above, 12MP kinda limits you to 8x10, although you can still get very good results at 11x14'' (I have with my D700.) 36MP gets you up to very good 13x19 or even 24x36'' while maintaining high-DPI.

QuoteQuote:
We chose pixel-level quality over nonsense megapixies. I can bet you that my "inferior" K5 is every bit as good as whatever full frame you have. Nobody here will recognize between your prints and my prints when they are put side by side.
Little bit of a mixed message there - "pixel-level quality" and "print side by side." PIxel-level quality would be the same, side-side prints - not. Doesn't mean K5 wouldn't be 'good enough', no-one is saying that.


QuoteQuote:
[/COLOR]I think I know why there is disagreement on the perceived noise between different sensor sizes. My explanations and my experiment deals with noise measured at the sensor (pixel) level while others treat noise at print level. A larger image would have less apparent noise when printed at the same size as a crop sensor. That's pretty basic. However, that does not imply that the cleaner output is due to the larger sensor gathering more light. It doesn't.
The only time that would happen is if the larger sensor has larger sensels assuming other factors are held constant (e.g. efficiency, read noise, etc...). A k5 will have the same noise profile as a D800 at the sensor level. A D800 will look cleaner at the same print size.
So how could your 'side-side' output look the same, then?

Take the D800 vs. K5 out of it, because that introduces an annoying brand-warfare subtext - just say Pentax came out with a FF K1 tomorrow that had the exact same Sony EXMOR sensor the D800 uses. You'd claim that at your typical FOV's and F-stops, that you wouldn;t see any noise advantage with the FF Pentax K1 over the K5? I'm not talking about 100% magnifications of both, I'm talking about output, your results.

By the way, here's what you're moving towards, and I'm pretty sure I tried to talk to this before in one of the other threads months ago but you blew it off or discounted it in some way:

K5 vs. D700 vs. D800, 100% magnification view:


K5 vs. D700 vs. D800, same display size (labeled 'print', but normalized and displayed on screen or print.)

.


That last graph shows what you would see on your screen or in print if you displayed the images from all three cameras, same size, side-side. It describes your output. And it's why DxOMark chooses to score based on it:
.



04-18-2014, 08:50 PM   #227
Pentaxian




Join Date: Jan 2012
Photos: Gallery
Posts: 588
QuoteOriginally posted by jsherman999 Quote
Something 5 stories high is meant to be seen from blocks away, and/or is upsampled by some professional software like Genuine Fractals. In Mpls here we reportedly had an iphone shot on the side of a building - looked just fine from most of downtown.

If you want to be picky and maintain 300dpi and above, 12MP kinda limits you to 8x10, although you can still get very good results at 11x14'' (I have with my D700.) 36MP gets you up to very good 13x19 or even 24x36'' while maintaining high-DPI.



Little bit of a mixed message there - "pixel-level quality" and "print side by side." PIxel-level quality would be the same, side-side prints - not. Doesn't mean K5 wouldn't be 'good enough', no-one is saying that.




So how could your 'side-side' output look the same, then?

Take the D800 vs. K5 out of it, because that introduces an annoying brand-warfare subtext - just say Pentax came out with a FF K1 tomorrow that had the exact same Sony EXMOR sensor the D800 uses. You'd claim that at your typical FOV's and F-stops, that you wouldn;t see any noise advantage with the FF Pentax K1 over the K5? I'm not talking about 100% magnifications of both, I'm talking about output, your results.

By the way, here's what you're moving towards, and I'm pretty sure I tried to talk to this before in one of the other threads months ago but you blew it off or discounted it in some way:

K5 vs. D700 vs. D800, 100% magnification view:


K5 vs. D700 vs. D800, same display size (labeled 'print', but normalized and displayed on screen or print.)

.


That last graph shows what you would see on your screen or in print if you displayed the images from all three cameras, same size, side-side. It describes your output. And it's why DxOMark chooses to score based on it:
.

Those graphs are in agreement with my statements. The first graph shows the noise profile of K5 and D800 being the same and the D700 besting both of them. The second graph is normalized so at the same print size, a bigger file will catch up. And this is what I have been saying all along. Equivalency vs noise is pure nonsense.

Now let's focus on print size. A 12Mp can print 12x18 and even 24x36 just fine if you do not put your nose on the wall. Trust me. If you want anything larger than that then blow it up with Genuine Fractals or even Photoshop. You wont see any more details but you can view the print farther away. Blowing up for printing makes sense.

Downsampling a 36mp image for the sake of noise comparison does not make sense. In fact it's cheating. DXOmark has to normalize the results because they are in the business of ranking cameras. Ranking requires normalization. That's fine and nothing wrong with that...well except when people use this data to spread nonsense like equivalency vs noise.
04-18-2014, 09:14 PM   #228
Veteran Member
jsherman999's Avatar

Join Date: Dec 2007
Photos: Gallery
Posts: 8,228
QuoteOriginally posted by dtmateojr Quote
The first graph shows the noise profile of K5 and D800 being the same and the D700 besting both of them. The second graph is normalized so at the same print size, a bigger file will catch up.
Yes, I know, and when I tried to say the same thing to you before it was discounted, and you tried to maintain that your D700 wil give better output results because it had larger pixels.

QuoteQuote:
And this is what I have been saying all along. Equivalency vs noise is pure nonsense.
How is it even 'Equivalency vs. noise"? What does that even mean?!? What specifically do you mean by 'equivalency vs. noise'?

QuoteQuote:
Now let's focus on print size. A 12Mp can print 12x18 and even 24x36 just fine if you do not put your nose on the wall. Trust me. If you want anything larger than that then blow it up with Genuine Fractals or even Photoshop. You wont see any more details but you can view the print farther away. Blowing up for printing makes sense.
Sure - and more MP gives you even more data to work with, more sizes available at 'excellent' or 'good' DPI resolution. And what if you wanted to crop your 12MP image 10, 30, 50%?

QuoteQuote:
Downsampling a 36mp image for the sake of noise comparison does not make sense. In fact it's cheating.
It in fact is not 'cheating', it provides a way to compare the images without having one be larger than the other, which would be an unfair comparison and wouldn't describe what photographers do with their output.

You could do other things - you could upsample the D700 to the D800 native output - do you somehow think the D700 will fare better doing that in comparison? I had both a D800 and D700 for a year together, shot both side-side quite a bit in all sorts of situations. There was no scenario at any ISO or display dimension where the D700 was 'better'**. it was often the same but it was often obviously worse, in resolution, noise, and noise profile/pattern - the noise that was there in the D800 was finer-grained, looked more 'natural' and held up to NR better while retaining detail. Trust me. Or don't, trust the many others who have owned both and said the same thing.

** Edit: In IQ. D700 is 'better' in FPS and speed in PP (throughput & processing.)
.

Last edited by jsherman999; 04-18-2014 at 09:20 PM.
04-18-2014, 09:22 PM   #229
Pentaxian




Join Date: Jan 2012
Photos: Gallery
Posts: 588
Well probably because you were being cheeky and you only showed me the 2nd graph. I'm not familiar with dxomark and I honestly don't care about their rankings.

And this equivalency vs noise is the OP.

04-18-2014, 09:35 PM   #230
Veteran Member
jsherman999's Avatar

Join Date: Dec 2007
Photos: Gallery
Posts: 8,228
QuoteOriginally posted by dtmateojr Quote
Well probably because you were being cheeky and you only showed me the 2nd graph. I'm not familiar with dxomark and I honestly don't care about their rankings.

And this equivalency vs noise is the OP.
Fine, I'll re-read the OP, although it would be helpful if you elaborated because 'equivalence vs. noise' does not appear in the OP. But I spoke of the difference between 100% views and same-display-size views several times before, and how the former shouldn't be used to judge output performance and as I recall you fought that concept tooth & nail.

Now I should also say that this MP battle doesn't mean the D700 isn't, in my view, still one of the best cameras you can buy in terms of output quality and especially AF performance - might be better than the D800 in AF consistency - and my FF preference doesn't mean I don't think the K5 isn't the best aps-c IQ/price ratio there is. I'm not trying to badmouth anyone's cameras, I'm just talking about equivalence, MP advantages and what those things can mean to shooting realities.
04-18-2014, 10:28 PM   #231
Pentaxian




Join Date: Jan 2012
Photos: Gallery
Posts: 588
The equivalence argument says that if everything else is held constant then sensor size will affect noise because of light gathering capacity. FF having a larger area will have lesser noise vs aps-c. The graphs you showed and the samples I got from dpreview says otherwise.

My stand has always been if everything else is held constant then noise is affected by light intensity as per f-stop. A f-stop is a f-stop no matter which format and therefore noise is the same for different sensor sizes at the same f-stop. Not aperture. The aperture-only argument totally ignores focal length and I have disproven this with my own experiments. This aperture-only argument is what makes the equivalency argument (35/5.6 vs 50/8.0) a pure nonsense when plotted against noise. Equivalency makes sense when you talk about FoV and DoF but NEVER about noise performance. In the case of 35/5.6 aps-c vs 50/8.0 FF the aps-c will have better SNR by a stop. If their findings show that they are equal then that's not because ff is better sensor-wise but because they normalized instead of viewing at 100%. And that's where those two dxomark graphs come in which proves my point.
04-18-2014, 10:39 PM   #232
Veteran Member




Join Date: Feb 2011
Posts: 4,893
Original Poster
Uh... if you're saying that noise per pixel is the same but noise per picture is less for full frame, then everyone has always agreed with you.

It's tough to tell, though, because you mentioned 'everything else held constant' which is tough. Is f stop held constant or is dof? Is focal length held constant or is aov? Is pixel size held constant or is number of pixels?
04-18-2014, 11:06 PM   #233
Pentaxian




Join Date: Jan 2012
Photos: Gallery
Posts: 588
QuoteOriginally posted by ElJamoquio Quote
Uh... if you're saying that noise per pixel is the same but noise per picture is less for full frame, then everyone has always agreed with you.

It's tough to tell, though, because you mentioned 'everything else held constant' which is tough. Is f stop held constant or is dof? Is focal length held constant or is aov? Is pixel size held constant or is number of pixels?

My examples have always been K5 or D7000 vs D800 coz they have the same sensor. So I dont care about focal length or number of pixels (sensor size). As long as f-stop is the same then noise is the same. Change the AoV if you want but as long as the exposure reading remains the same (e.g. a uniformly lit huge subject) noise will be the same. So 50/5.6 on aps-c or ff or any format with the same sensor will have the same noise. The equivalency-fu does not agree with this. Even that guy Joseph James who is constantly being referred to here says that sensor area is the single biggest factor that affects noise performance. That is so very wrong.

04-19-2014, 03:03 AM   #234
Site Supporter




Join Date: Mar 2009
Location: Gladys, Virginia
Photos: Gallery
Posts: 16,258
QuoteOriginally posted by dtmateojr Quote
My examples have always been K5 or D7000 vs D800 coz they have the same sensor. So I dont care about focal length or number of pixels (sensor size). As long as f-stop is the same then noise is the same. Change the AoV if you want but as long as the exposure reading remains the same (e.g. a uniformly lit huge subject) noise will be the same. So 50/5.6 on aps-c or ff or any format with the same sensor will have the same noise. The equivalency-fu does not agree with this. Even that guy Joseph James who is constantly being referred to here says that sensor area is the single biggest factor that affects noise performance. That is so very wrong.
Per pixel noise doesn't matter. Sure, it is the same on the D7000, K5, and D800 and less on the D700, but the question is at a specific viewing size which will have the least noise. The answer is the D800. This will not be particularly apparent at small print sizes and low iso, but as you move up in print size and go to higher isos, the D800 will out perform the other cameras.
04-19-2014, 03:07 AM   #235
Pentaxian




Join Date: Jan 2012
Photos: Gallery
Posts: 588
QuoteOriginally posted by Rondec Quote
Per pixel noise doesn't matter. Sure, it is the same on the D7000, K5, and D800 and less on the D700, but the question is at a specific viewing size which will have the least noise. The answer is the D800. This will not be particularly apparent at small print sizes and low iso, but as you move up in print size and go to higher isos, the D800 will out perform the other cameras.

By your logic the 41Mp Nokia phone will outperform the D800. Let's all buy cellphones instead. A D800 can't even access facebook. Should I laugh now?
04-19-2014, 04:24 AM   #236
Inactive Account




Join Date: Mar 2014
Posts: 71
QuoteOriginally posted by dtmateojr Quote
My examples have always been K5 or D7000 vs D800 coz they have the same sensor. As long as f-stop is the same then noise is the same.
QuoteOriginally posted by Rondec Quote
This will not be particularly apparent at small print sizes and low iso, but as you move up in print size and go to higher isos, the D800 will out perform the other cameras.
Imprecise wording is the killer. Picking single aspects and hiding the consequences is another.
  1. The noise "on the sensor" of a K-5 and a D800 is the same. Same noise on same sensor area of say 1cm2.
    Simply because at same aperture + shutter + ISO the same amount of light hits a same size sensor area.
  2. If (important word) you look at same size pictures from same distance then your perception of the noise changes: the little noise spots are magnified x1,5 and you will perceive them as more distracting. So you will see more noise. More noise perceived.
  3. If you print a picture from both cameras with different sizes (relationship just the same as the sensor sizes) and watch them from the same distance: not even noise perception difference.
  4. If you print both pictures the same size and move closer to the D800 picture "because it has so much more detail resolution": not even noise perception difference.
Four different types of looking at the same result. Twice A is better than B, twice A is just the same as B. All are right. All have their application. None is better or more worthy than the other.
04-19-2014, 04:42 AM   #237
Pentaxian
Fogel70's Avatar

Join Date: Sep 2009
Location: Stockholm, Sweden
Photos: Albums
Posts: 1,748
QuoteOriginally posted by dtmateojr Quote
The equivalence argument says that if everything else is held constant then sensor size will affect noise because of light gathering capacity. FF having a larger area will have lesser noise vs aps-c. The graphs you showed and the samples I got from dpreview says otherwise.
What equivalence really says is (very simplified): - With equivalent settings all systems will collect the same amount of light (total amount of light). I don't think there is many websites that benchmark photography equipment with equivalent settings.

After all, all sensors use the same laws of physics. To create the same pictures (as close as it's possible) across all formats the same amount of light is needed. Equivalence is usually only a simplified way of explaining the laws of physics.

The advantage of larger sensors is that they can have equivalent settings not available on smaller formats (FI a image created with 50/1.4 on FF would need a 33/0.9 lens on APS-C). Or a 35/2.4 image on APS-C would need a 12,5/0.9 lens on Q7.

But smaller formats might have advantage of cheaper lightweight equipment not available on larger formats.
04-19-2014, 07:15 AM   #238
Veteran Member
jsherman999's Avatar

Join Date: Dec 2007
Photos: Gallery
Posts: 8,228
QuoteOriginally posted by dtmateojr Quote
The equivalence argument says that if everything else is held constant then sensor size will affect noise because of light gathering capacity. FF having a larger area will have lesser noise vs aps-c.
No - equivalence says (very quickly) that if everything is held constant, the total light gathered will be the same and noise, DOF etc will be the same. It's provided as an equation, a relationship. Equivalence very clearly states that this is how the larger sensor affects it : that the larger sensor mandates a larger entrance pupil at the same AOV and F-stop, and this larger entrance pupil (mandates also same distance and AOV) is what's bringing in more total light (which is different than 'exposure') - and resulting in less noise. It's also resulting in less DOF in that scenario. It very clearly states that any smaller sensor of equal efficiency can get the same noise performance as any larger sensor - if there was a lens for the smaller sensor that allowed a large enough entrance pupil.

Often I suspect people don't follow links or really read them. Below excerpt is no substitute for the whole article, but hopefully it helps some people with one or two things. (I highlighted in yellow something Rondec & I were just talking about.)





QuoteQuote:
The graphs you showed and the samples I got from dpreview says otherwise.
Actually equivalence does describe exactly in part how their 'print' graph shows the larger sensors' showing better noise performance than the K5, even though the K5 has the same per-pixel quality as the D800 - because they don't keep linear aperture the same when they compare, they only keep F-stop and AOV the same - which allows more total light to fall on the sensor, because of the larger entrance pupil.





.
04-19-2014, 07:58 AM   #239
Site Supporter




Join Date: Mar 2009
Location: Gladys, Virginia
Photos: Gallery
Posts: 16,258
QuoteOriginally posted by dtmateojr Quote
By your logic the 41Mp Nokia phone will outperform the D800. Let's all buy cellphones instead. A D800 can't even access facebook. Should I laugh now?
The D7000 and D800 have the same size pixels. The D800 is a newer generation of tech versus the D700. This all makes a difference. The Nokia obviously has significantly smaller pixels than any of the cameras discussed here. But, the idea that having more pixels makes noise worse is definitely not true. At worst, noise will be equivalent, but even in the case of the Nokia, it performs better than most other camera phones with regard to high iso performance.
04-19-2014, 09:32 AM   #240
Inactive Account




Join Date: Mar 2014
Posts: 71
QuoteOriginally posted by Rondec Quote
But, the idea that having more pixels makes noise worse is definitely not true.
As matter of fact that statement is not true. More pixels = more noise. But in many cases such as 36x24 vs 24x16 sensors this more noise is not significant.

The whole reason for the existence of back side illuminated sensors is driven by the underlying reason: more pixel = more "wiring". "Wiring" does not gather light, so area is lost for light, so more noise. To avoid this they put the wiring on the other side. Since it's only significant for extreme pixel densities this is the only area where manufacturers do it.
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!
24x36mm, aperture, aps-c, camera, check, da, dof, equivalence, f/8, ff, ff resolution, full-frame, lenses, mf, pentax, photo, photography, post, practice, reasons, resolution, sensor, shot, shutter, theory, time, wrong
Thread Tools Search this Thread
Search this Thread:

Advanced Search


Similar Threads
Thread Thread Starter Forum Replies Last Post
How to shoot a wedding with APS-C and FF nick52 Photographic Technique 8 08-05-2013 05:14 PM
K-5IIs reviews at Amazon - Better than FF Canon? Docrwm Pentax K-5 4 02-19-2013 12:12 PM
Quick question regarding field of view - FF vs APS-C glass? Julie Pentax SLR Lens Discussion 5 12-23-2012 05:33 PM
APS-C does not increase focal length over FF, it decreases field of view. TomTextura Photographic Technique 135 06-09-2012 04:58 PM
What is your best example for FF over APS-C? leeak Pentax DSLR Discussion 21 12-11-2008 12:25 PM



All times are GMT -7. The time now is 08:30 PM. | 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