Why is everyone calling the mk3 sensor the same as mk2?
- Thread starter Cgdillan
- Start date
- Status
TTMartin said:
read noise doesn't matter that much once you are looking at tones as bright as what they test SNR at
there is read noise, late and early stage and random and fixed pattern
and then there is photon noise and some cameras are better at grabbing photons than others and some are more uniform photosite to photosite
and just look at say the fredmiranda comparison between 5D3 and D800 and you surely can see the same difference DxO and others report for low ISO DR
Upvote
0
TTMartin said:
no they don't
you don't even use a raw converter at all to do these tests, raw converters can cook the books
you simply look at the raw file itself with a program that looks straight at the full raw data, for dynamic range the test is easy enough, if you do it you'll get the same answers for ISO100 as DxO pretty much (some minor sample to sample variation)
Upvote
0
anyway back on topic 5D2 vs 5D3 sensor
they are a little bit different:
5d3 has a trace worse read noise at lower iso and slightly lower maximum dynamic range than the 5D2 although the differences are likely too small to matter in the real world and they are basically the same
5d3 has almost no horiz banding at low iso while 5D2 has a lot but since the 5d3 has a lot of vertical banding as low iso as does 5D2 in the end they both show banding and once it shows it doesnt really matter if it is one or both directions so it's kinda the same in the again
5d3 has at least 1/2 stop and maybe more like 2/3rds of a stop better SNR, not a huge difference, but it's a difference and considering how good the SNR was on the 5D2 it's not surprising there change was not much larger
5d3 has somewhat higher dynamic range, especially usable DR, at high iso than the 5D2, at times it's a minor difference, other times it make a real difference
5d3 has almost no banding at high iso while the 5d2 can get ugly in darker parts of high iso shots and when you get into super, super, super high iso this can start making a big difference in general and make the effective difference greater than the 1/2- 2/3rd stop SNR differece
5d3 is more color blind and can distinguish less colors than 5d2 (real world implications are very, very complex and i don't know anyone who has looked into the real world differences, all we know is the 5d3 measures the most color blind of any dslr made)
basically they are the same but 5D3 has modestly better SNR and far less uglies and banding and junk at high iso the higher you go the bigger the difference the quality of the high iso noise between the two
the don't rate iso the same, so if you compare both at same iso, then the 5D3 gets put at unfair disadvantage compared to the 5d2
they are a little bit different:
5d3 has a trace worse read noise at lower iso and slightly lower maximum dynamic range than the 5D2 although the differences are likely too small to matter in the real world and they are basically the same
5d3 has almost no horiz banding at low iso while 5D2 has a lot but since the 5d3 has a lot of vertical banding as low iso as does 5D2 in the end they both show banding and once it shows it doesnt really matter if it is one or both directions so it's kinda the same in the again
5d3 has at least 1/2 stop and maybe more like 2/3rds of a stop better SNR, not a huge difference, but it's a difference and considering how good the SNR was on the 5D2 it's not surprising there change was not much larger
5d3 has somewhat higher dynamic range, especially usable DR, at high iso than the 5D2, at times it's a minor difference, other times it make a real difference
5d3 has almost no banding at high iso while the 5d2 can get ugly in darker parts of high iso shots and when you get into super, super, super high iso this can start making a big difference in general and make the effective difference greater than the 1/2- 2/3rd stop SNR differece
5d3 is more color blind and can distinguish less colors than 5d2 (real world implications are very, very complex and i don't know anyone who has looked into the real world differences, all we know is the 5d3 measures the most color blind of any dslr made)
basically they are the same but 5D3 has modestly better SNR and far less uglies and banding and junk at high iso the higher you go the bigger the difference the quality of the high iso noise between the two
the don't rate iso the same, so if you compare both at same iso, then the 5D3 gets put at unfair disadvantage compared to the 5d2
Upvote
0
LetTheRightLensIn said:
If you are looking directly at the RAW data, and in fact there is additional data hidden in the RAW file how are you obtaining that data?
That's the whole point, you can't just look at the RAW data and see the data that Canon hid in it.
The RAW converter 'cooking' that you describe is Canon RAW converter accessing the hidden data.
Come on people, catch up, banding / pattern noise from every Canon camera, uniform 11 EV max from the unhidden RAW data across the board, and the CR2 file format can't have any hidden data?
Oh, and sensors that should have greater Dynamic Range like the 5D Mk III over the 5D Mk II have more read noise, and that read noise can't actually be data?
Upvote
0
TTMartin said:
Check sensorgen.info. The information on that site is derived from DXO raw data. It represents things in electron charge rather than levels or decibels, so its a little easier to perform direct comparisons.
As for DXO's SNR 18%, that is the signal ratio based on a fairly bright luminance level. At 18%, your well above the level where electronic noise is going to interfere with your image detail. You need to look at the "Full SNR (Logarithmic) chart to see why the D800 offers better DR when the signal is very low (i.e. a luminance level of 1% or less.) With the Full SNR (Logarithmic) chart, you can easily see where noise starts limiting 5D III SNR at the three highest ISO levels...they start to bunch up just under a 0.1% luminance level. The D800's signal exhibits a roughly linear pattern right down to ISO 100, allowing a minimum luminance level of around 0.01% to be recorded! That extra headroom down to 0.01% luminance is why Sony Exmor sensors offer unparalleled shadow lifting capability (i.e. low ISO DR).
TTMartin said:
I'm not sure what sensor you are referring to (I think the GX1 sensor is almost as large as an APS-C sensor, if that is what your are thinking about...quite a bit larger than 1/1.7). When it comes to smaller sensors, there are some things you can do to realize potentially huge gains. To name the most significant, moving to a backlit sensor design for tiny sensors, you move all the activate and read wiring to the back side of the sensor (relative to the light source), exposing the photodiode entirely to light without obstructions. Addition of a gapless microlens design, and your QE on a small sensor can get pretty high (for their size). These designs are rather costly, though. With large sensors, per-wafer yield is low, so R&D costs are very high and the benefits of backlit design isn't nearly as great as it is for small sensors. With small sensors, you can potentially get a hundred sensors or more from a single wafer, and you can often use smaller wafers to start with...so despite the more complex design of a backlit, microlensed design, its still more cost effective.
To improve efficiency of a larger sensor, Canon would have to design similar, but patently different, technology to Sony's Exmor sensor design. A backlit design won't do it...pixel pitch is large enough in an APS-C and FF sensor that flipping the chip won't gain you that much more area in terms of percentage. The cost would outweigh the gains. Canon would have to design additional levels of dark current noise removal beyond just CDS, move to some kind of onboard CP-ADC (column-parallel analog-to-digital conversion) to help mitigate FPN, find a way of addressing transistor differentials which also contributes to FPN, figure out how to avoid or mitigate signal interference which is often the cause of non-fixed banding or other pattern noise, etc. etc. And they would have to do it in a way that did not infringe upon Sony's patents that address the same issues.
Upvote
0
TTMartin said:
You seem to think that Canon's RAW file format, CR2, is some kind of super top secret, encrypted data format that contains the deepest, darkest secrets that would somehow unlock the full potential of Canon's theoretical Nikon-comparable DR...
That is not the case. The CR2 format is a public specification (Canon even offers free source code that demonstrates how to read, write, and process their CR2 format), which is essential to supporting third-party development of RAW conversion and processing tools...such as Lightroom, Aperture, etc. Additionally...its a "RAW" file...it doesn't contain anything particularly special or super-secret that would need to be "hidden" that didn't come strait off the hardware...it has the full sensor readout...both the unmasked light-sensing pixels as well as masked "calibration" or "baselining" pixels around the margins, camera settings from the firmware, etc.
This isn't much different from any other manufacturers' RAW format, only its in a structure and form that allows Canon to achieve the things they need to...such as a high write out rate that supports the frame rates of their sports bodies. Canon, like many sensor manufacturers, also makes use of what is called a "bias offset" allowing a dynamic black level to be used when processing the RAW files. Nikon does not use a bias offset...their black level is fixed at zero, however with their low read noise, thats not an issue. The only real question regarding DXO's handling of CR2 files is whether they are properly factoring in the bias offset. Working in Adobe Lightroom, I wouldn't say that the 5D III is as bad off as it seems relative to the D800 based on DXO data...but there is definitely a difference, and the D800 certainly has the edge.
Read noise is an electrical byproduct in the sensor HARDWARE. A sensor is a signal receptor....it receives a two dimensional spatial waveform that must later be read out to produce a digital image. The process of reading that image signal introduces its own signal into the final product. There are a variety of forms of electronic noise that can interfere with an image signal, including dark current noise, sampling noise, differential noise, conversion noise, thermal noise, and noise generated by external fields interfering with the signal of the sensor itself. In the grand scheme of things, electronic noise, even in the 5D III which has very high read noise, is a small fraction of the maximum image signal (in the case of the 5D III, read noise at ISO 100 is 33.1 electrons, while maximum signal potential, or maximum saturation, is 67531 electrons...a ratio of 2040:1 at max. sat.!)
The problem with read noise is not when your image signal for a given pixel is high. If you have a nearly white pixel, read noise is 1/2040th the amount of the image signal...inconsequential. You are producing an output sample that is 99.9995% accurate. The problem with read noise occurs when the image signal for a given pixel is low. If you have a nearly black pixel, the read noise becomes a far more significant fraction of the image signal, possibly even overpowering the image signal. Again, with the 5D III, if the SNR of the shadows is say 1/2, then your signal is around 66e-, while read noise is 33e-. You are keeping a mere 50% of the image signal in the output sample the sensor originally registered...you've lost 50% accuracy. If your SNR in the shadows is 1/1, then your signal and your read noise are both 33e-. Now, you have nearly a 0% level of accuracy producing an output sample for that pixel. If your signal level is lower than the read noise level, it just becomes more useless noise.
Now, contrast the same electron levels above, only with a D800. The D800 has a roughly constant read noise level at all ISO settings of 3e-. Its maximum saturation is 44972 at ISO 100, which means its signal to noise ratio is 14990:1 in terms of electron charge. If the signal charge in a pixel is 66e-, instead of having a 50% image signal loss you now only have a 4.5% image signal loss. If the signal charge is 33e-, you have a signal loss of 9%. You also have the ability to sample an image signal that is considerably lower than what the 5D III is even capable of. At 10e-, your still only losing 30% of your image signal to noise. Only by the time you reach an image signal level of 6e- do you reach that 50% level that the 5D III reached at a signal level of 66e-!
Read noise...probably better termed "electronic noise" as its due to the electric and thermal properties of sensor during readout...becomes a part of the output signal from the sensor. (Note: Read noise isn't independent...its not something that can be represented separately in a RAW file...its an intrinsic part of the signal!) That output signal is then passed through an ADC, or analog-to-digital converter, which produces a digital file that, as closely as possible, represents "the signal" the ADC processed. Since electronic noise is a part of "the signal", it effectively gets "baked in" to the RAW file. Once its baked, there ain't no unbaking it. You can scrape off the charred parts, smooth over other unsightly lines and edges, and frost the top...but your cake is still gonna smell burnt if its burnt. ;P
Upvote
0
That must explain why my D800 has so much more noise than my 5D MK III at high ISO's, 12800 and higher. It is great as long as I stay under ISO 800, and DR is supurb at ISO 100. But it really struggles as ISO rises. Of yourse, I could reduce the image to 8mp like DXO and reduce noise as well. however, the point of having 36mp is lost. You can have it one way or the other. Reduce it to 8mp and have low noise, or keep it at 36mp with high noise. Its those that measure noise at 8mp but then read the resolution at 36mp who are cooking the books.jrista said:
Upvote
0
Mt Spokane Photography said:
At high ISO, it boils down to physics. The higher noise levels of the D800 at 12800 are probably due to the greater degree of photon shot noise per unit area than you get with the 5D III. (Not sure what the ratio is at the moment, maybe on the order of 3-4x.) Same reason why the 7D also experiences more noise at higher ISO.
Indeed! DXO is double-cooking the books when they claim you can increase DR simply by downscaling. They are ignoring the fact that if parts of the signal on the sensor are blown when utilizing the full DR range of 13.2 stops (i.e. the "actual" real image signal requires a higher signal level than the sensor supports...greater than max saturation) then no amount of downscaling will correct that, nor will it allow DR to magically expand to 14.4 stops. Total fallacy.
Upvote
0
TTMartin said:
Because you can look through the whole file and not notice any areas with hidden data.
I also have to say why would canon magically hide away certain low order bits of information in some special block?? And why they then add fake noise at the low end and put fake data in the main section??? That would be so bizarre.
And why does Canon's own DPP not magically reveal more dynamic range than when using non-Canon converters? And why using say ACR for both a 5D2 and D800 do the files from D800 visibly show a lot more dynamic range if ACR is hooking into secret data for Canon?
They secretly overlay magic data in a random gaussian distribution and you are supposed to be able to read that? And again talk about bizarre.
Upvote
0
jrista said:
it's not on the top end it gains its on the bottom end and it's not double cooking and it's not anti-canon, when canon had high mp 5D2 and nikon had low MP D700 it went the other way
NOT comparing them that way would be cooling the books actually
Upvote
0
LetTheRightLensIn said:
Its the same problem with the 5D II and D700...I'm not brand biased in my assessment of DXO's Print DR...I don't believe it is possible to gain dynamic range by downscaling, despite the apparent reduction in noise. I've explained at great length in other answers, and you can search for them if you wish...so I won't go into it in detail again here.
Upvote
0
jrista said:
I did look at that stuff you posted before but it's not fair to compare images at different scales. You need to normalize to the same scale. It's not fair to compare at 100% view which is what you want to do no matter what. You are comparing noise at different powers as if they had the same power. Filter away the high frequency/power noise the second camera doesn't have and then compare fairly.
Upvote
0
I own a 5d Mkii, 2 x Mkiii and both the D800 & D800E... I prefer using the 5D mkiii's in most situations. I use the D800/E for product photography as well as some other studio work. I will sell the 5D mkii s as the mkiii are much more responsive, a bit better standard iSO image quality (and much better higher iso performance) and gives me a better success rates at capturing the correct focus, its a much faster responsive camera. The quality of the D800 is excellent especially if you use good technique and are very careful with lens/camera settings, the quality of the images from the 5D mkiii is also excellent but with little, if any care/setup required. The build quality/weather sealing is far better than the D800 and I prefer the default colour balance from the Canon's. The high ISO performance on the Canon is fantastic up to 12800 (above that there is too much noise for me).
Photography outside a studio is often about being quick and capturing the shot, you can't always stage it. So each camera has its use, yes the D800 has 36+ MP and great dynamic range at low ISO especially in shadows/black area but the sensor on the 5D is excellent as well and DXO's markings do not reflect the true "picture".
The programs used for conversion from raw has a dramatic impact on the overall quality of the image including noise.
LR, Aperture, Camera Raw 7.1 (I think it may use the same engine as LR), Phocus, DXO, ViewNX 2, Canon DPP. I would suggest comparing the difference and finding what suits you.
Photography outside a studio is often about being quick and capturing the shot, you can't always stage it. So each camera has its use, yes the D800 has 36+ MP and great dynamic range at low ISO especially in shadows/black area but the sensor on the 5D is excellent as well and DXO's markings do not reflect the true "picture".
The programs used for conversion from raw has a dramatic impact on the overall quality of the image including noise.
LR, Aperture, Camera Raw 7.1 (I think it may use the same engine as LR), Phocus, DXO, ViewNX 2, Canon DPP. I would suggest comparing the difference and finding what suits you.
Upvote
0
TTMartin said:
how come dpp doesn't reveal more DR than ACR then?
why does banding sometimes show in DPP too?
who would ever store data in that fashion? and for the random banding where is the hidden key in the file and why has nobody ever found one?
etc.
etc.
it's like saying you don't believe man ever landed on the moon, almost
Upvote
0
LetTheRightLensIn said:
DPP does reveal more DR.
I've never seen banding in DPP.
Who would store data in this fashion? Evidently Canon. Why would they do that? Well, a couple years prior to the CR2 format being released, Nikon updated the NEF format and encrypted part of the data. Why would Nikon do that? When Nikon did there was outrage, and of course the encryption was quickly broken. So when Canon updated their CR2 format, they hid the data in more or less plain sight. No up roar over the data being encrypted, nobody hacking it a couple weeks after it was introduced. Where the key is I don't know. Why has nobody found one? Probably because they haven't looked.
etc.
etc.
etc.
As for the moon, thing watch the movie 'Contact' if you don't get the concept.
Upvote
0
TTMartin said:
really?
show us
and it makes the 5D3 reveal more than the 5D2?
I have, although later versions control it a bit more.
And so then what about Nikon? SOme of their cams have banding to? SO have sensors from any digital camera ever made. All the banding was just secretly storing extra DR? There isn't even enough banding sections to cover the frame in a way to even propose what you propose. Total nonsense.
People have looked through CR2 through and through.
umm yeah
(and for the record that movie shows nothing whatsoever ever the moon landing conspiracy type stuff, at all)
Upvote
0
- Status
Similar threads
