Approve the Cookies
This website uses cookies to improve your user experience. By using this site, you agree to our use of cookies. Read More.
OK
Index  •   • New posts  •   • RTAT  •   • 'Best of'  •   • Gallery  •   • Gear  •   • Reviews
New posts  •   • RTAT  •   • 'Best of'  •   • Gallery  •   • Gear  •   • Reviews
Register to forums    Log in

 
FORUMS News & Rumors Photography Industry News
Thread started 12 Oct 2012 (Friday) 14:35
Prev/next
sponsored links
(this ad will go away when you log in as registered member)

DxOMark score for EOS 650D (T4i) Out: same score as the EOS 20D from 2004

 
Shadowblade
Cream of the Crop
Joined Dec 2008
Melbourne, Australia
Oct 14, 2012 18:11 |  #31

watt100 wrote in post #15122097external link
comparing 60D with Nikon D7000
Canon easily beats the Nikon 7000 in ISO performance
WTF ??????????????

Read the testing methodology.

DXO's ISO benchmark defines the ISO rating as the highest ISO at which a sensor can maintain 9 stops of DR, a 30:1 SNR and 18 bits of colour depth.

The Popphoto test measured only noise, after default noise reduction, and without (what does '2.2 noise performance' at ISO 1600 mean anyway?)

Also, DXO's measurements are done using RAW files with no noise reduction or sharpening applied. The Popphoto test you quoted did it test using RAW files with default settings applied - i.e. with software variation, not purely measuring hardware capability.




LOG IN TO REPLY
sponsored links
(this ad will go away when you log in as registered member)
gabebalazs
Bird Whisperer
gabebalazs's Avatar
Joined Nov 2008
Toledo, OH
Oct 14, 2012 18:52 |  #32

Shadowblade wrote in post #15122138external link
Read the testing methodology.

DXO's ISO benchmark defines the ISO rating as the highest ISO at which a sensor can maintain 9 stops of DR, a 30:1 SNR and 18 bits of colour depth.

The Popphoto test measured only noise, after default noise reduction, and without (what does '2.2 noise performance' at ISO 1600 mean anyway?)

Also, DXO's measurements are done using RAW files with no noise reduction or sharpening applied. The Popphoto test you quoted did it test using RAW files with default settings applied - i.e. with software variation, not purely measuring hardware capability.

Exactly.

I am a Popphoto subscriber overall I like the magazine but I've learned to take their noise tests with a large grain of salt, exactly for that reason.

Even when reading Popphoto's tests of other Canon bodies using (virtually) the same 18mp sensors, there are significant differences between noise perfromance - at least by Popphoto's standards.

Like mentioned above, they "measure" noise after opening the RAW file and using the MANUFACTURER'S own NR settings, which pretty much invalidates any kind of noise level comparison between brands and even between same brand models using virtually the same components.

Easy to see this by importing RAW files taken at the same ISOs from let's say the 7D, t2i, and t4i into DPP. If we look at the NR tab, we can see that the default noise reduction differs between these cameras. But Popphoto does not take this into consideration. So there goes level playing field.

This is exactly the reason why the heavily NR'd by default T1i had better noise level results than the then $8K Nikon D3X (and still "beats" most other Canon APS-C bodies too, even though I know from experience, and from other reputable RAW tests, that it's not the case.) :)

I'm still torn whether I'll keep the T4i that I ordered from Fry's. Right now, after staring at dozens of comparison images - RAW and Jpegs, I'm leaning towards sending it back (haven't received it yet).
Popphoto's T4i test found pretty good noise results, however, when looking at unprocessed RAW files (at DPreview for instance), we can clearly see a tiny but definite step back regarding high iso noise :( .
Of course Popphoto does not see this because the default NR for T4i RAW files opened in DPP are cranked up compared to older 18mp sensor Canon models (for example, the default luminance NR setting for ISO3200 images are 5 for 7D and T3i but 8 for T4i. The result is similar cleanness on the surface but obviously heavier NR and its detail smudging side effect.)

Oh, and this is exactly why the 60D (which I also used to own and liked it) beat the D7000 in Popphoto's noise tests, even though the unprocessed D7000 RAW files are a tiny bit less noisy (especially shadow noise) than the unprocessed 60D RAW files (see: DPReview)

What's curious though is that Digital Picture review found that the T4i was sharper than any of the other 18mp sensor Canon models... But that's another topic.


5DIV | 5DIII | 80D | A7RII | Canon 24-70 2.8L II | 70-200 2.8L IS II | 24-105 f/4L IS | 16-35 f/4L IS | 135 f/2L | 85 1.8 | 50 1.8 STM | 18-135 IS STM | SONY FE 28-70 OSS | Tamron 17-50 2.8 | Tamron 150-600 | Ʃ 35 1.4 ART | Rokinon 14 2.8 | Sigma 1.4x | Metabones IV | 2x Canon 600EX-RT
Gabe Balazs Photo (external link)
Nature Shots Portfolio (external link)

LOG IN TO REPLY
The ­ Fox
Goldmember
The Fox's Avatar
1,709 posts
Gallery: 1 photo
Joined Jul 2007
Huntington Beach, Ca
Oct 14, 2012 20:32 |  #33

996gt2 wrote in post #15121478external link
I think you should do a little reading on how DxO does their testing. The things in your post above are opinion, not the results of empirical testing, which is what DxO has done.

Their results are no joke. Whether you choose to accept them or not is your decision.

No, I could care less what camera I use to be honest anymore as long as the real world performance works. My clients are never going to pay me for having a nice camera, they pay me for results. I choose the tool that works for the job and for me 99% of the time is MF digital. MF digital will out resolve any D800, and the 16 bit color is going to give me the most information to work with. And going by the even how they test based on math, 16 bit has 65,523 tones per channel vs 16,383 on 14-bit. There is no way that you can not have better accuracy from the camera with almost 4 times the colors it can create per channel. The only way that they could is if the convert them both to 8 bit but I don't shoot in 8 bit so that is useless to me.

Oh and before I hear it in that I have never even used the D800. I am lucky enough to work for a company that has Nikon so I have all nikons bodies at the moment and both a D800/ D800E. Top nikkor glass and they are on their way out.

Nick


"I work from awkwardness. By that I mean I don't arrange things. If I stand in front of something, instead of arranging it, I arrange myself" -Diane Arbus
7D Gripped x2 | 50D Gripped | 17-50mm F2.8 | 35mm F1.4 | 50mm F1.4 | 85mm F1.8 | 70-200mm F4L IS |

LOG IN TO REPLY
The ­ Fox
Goldmember
The Fox's Avatar
1,709 posts
Gallery: 1 photo
Joined Jul 2007
Huntington Beach, Ca
Oct 14, 2012 20:38 |  #34

DXO Mark, making you feel good about your lens purchases since 2004


"I work from awkwardness. By that I mean I don't arrange things. If I stand in front of something, instead of arranging it, I arrange myself" -Diane Arbus
7D Gripped x2 | 50D Gripped | 17-50mm F2.8 | 35mm F1.4 | 50mm F1.4 | 85mm F1.8 | 70-200mm F4L IS |

LOG IN TO REPLY
996gt2
THREAD ­ STARTER
Goldmember
996gt2's Avatar
1,045 posts
Joined Jul 2008
Cincinnati
Oct 14, 2012 21:48 |  #35

tkbslc wrote in post #15121935external link
The sensor is still the same, regardless of the translucent mirror. You might get lower shutter speeds at the same ISO, but the mirror has the same impact as using a smaller aperture.

DxOMark is testing the RAW files from each camera. They are not removing the sensors from each camera and testing them alone, as that would be very difficult.

So what you're saying is not correct. It doesn't matter if the sensor is the same, the sensor performance (as tested from the RAW files) depends on everything in the light path, especially if there is a mirror that blocks 1/3 stop of incoming light. Let's say you take a D800 and put a piece of wax paper over the sensor...obviously that is going to have an effect on the RAW files from the modified D800. It has nothing to do with using a smaller aperture or anything like that, because all of the cameras are tested with consistent settings.

Comparing any of Sony's SLT cameras to a non-SLT with the same sensor and you will see exactly the same trend in performance from DxOMark. The SLT camera will perform worse every time even if the sensor is the same. This is one of the biggest downsides to Sony's fixed-mirror SLT design.


Buy/Sell Feedback
5Dc, 50D, Tokina 12-24 f/4, Tamron 28-75 f/2.8, Sigma 50mm f/1.4, 430EX

LOG IN TO REPLY
tkbslc
Cream of the Crop
24,566 posts
Joined Nov 2008
Utah, USA
Oct 15, 2012 00:33 |  #36

They aren't measuring the exposure, they are measuring the sensor data. If they expose the scene properly on the Sony (by adding 1/3 longer exposure time) the total light hitting the sensor will be the same and thus measurements will be directly comparable.

We can conclude that in the same scenario the SLT will get 1/3 a stop worse noise performance due to less light hitting the sensor, but when we have enough light for either SLT or SLR, then the translucent mirror has no impact.

The comparison you showed only had Sony trailing by 50. If it were a 1/3 a stop, it should be about 200.


Taylor
Galleries: Flickrexternal link
60D | ELPH 330 | iPhone 5s

LOG IN TO REPLY
tkbslc
Cream of the Crop
24,566 posts
Joined Nov 2008
Utah, USA
Oct 15, 2012 00:37 |  #37

Shadowblade wrote in post #15122138external link
Also, DXO's measurements are done using RAW files with no noise reduction or sharpening applied. The Popphoto test you quoted did it test using RAW files with default settings applied - i.e. with software variation, not purely measuring hardware capability.

How does one measure noise in a RAW file without applying some type of conversion algorithm to interpret the sensor data? Some settings are being applied that may or may not be the same formula used by Canon.


Taylor
Galleries: Flickrexternal link
60D | ELPH 330 | iPhone 5s

LOG IN TO REPLY
Shadowblade
Cream of the Crop
Joined Dec 2008
Melbourne, Australia
Oct 15, 2012 01:31 |  #38

tkbslc wrote in post #15123499external link
How does one measure noise in a RAW file without applying some type of conversion algorithm to interpret the sensor data? Some settings are being applied that may or may not be the same formula used by Canon.

By applying direct Bayer interpolation to the luminance values in the RAW file, without applying noise reduction of any kind. Also, the DXO tests rely on measuring dynamic range and colour depth, not on measuring noise per se.




LOG IN TO REPLY
tkbslc
Cream of the Crop
24,566 posts
Joined Nov 2008
Utah, USA
Oct 15, 2012 01:59 |  #39

I've use many RAW editors, including DxO and even with noise reduction set to zero, they all interpret Canon's output very differently. Noise is different, sharpness is different, color is different. So all we are doing is getting reviews based on DxO's interpretation of the RAW output.


Taylor
Galleries: Flickrexternal link
60D | ELPH 330 | iPhone 5s

LOG IN TO REPLY
Shadowblade
Cream of the Crop
Joined Dec 2008
Melbourne, Australia
Oct 15, 2012 02:07 |  #40

tkbslc wrote in post #15123685external link
I've use many RAW editors, including DxO and even with noise reduction set to zero, they all interpret Canon's output very differently. Noise is different, sharpness is different, color is different. So all we are doing is getting reviews based on DxO's interpretation of the RAW output.

At least it uses the same software for all cameras, thus equalising the software side of the equation and ensuring that it's just the hardware which is being measured.

Also, none of DXO's tests rely on measuring noise - they rely on measuring dynamic range, which doesn't vary between RAW converters. The ISO test isn't a test of noise - rather, it is a test of recoverable detail, defined as the highest ISO at which you still have nine stops of DR. This helps equalise for variations in noise reduction - an algorithm which applies NR (some do, even with NR set to 0) will score better in a noise test, but no better in a detail/DR test. After all, we have all seen photos taken at high ISO with minimal noise, but with minimal detail too, due to the conversion algorithm.




LOG IN TO REPLY
ZachOly
Goldmember
ZachOly's Avatar
1,043 posts
Joined Nov 2010
Canada
Oct 15, 2012 06:12 |  #41

Shadowblade wrote in post #15123701external link
At least it uses the same software for all cameras, thus equalising the software side of the equation and ensuring that it's just the hardware which is being measured.

I know this is crazy, but what if DxO's software interprets the readings from Nikon/Sony sensors more favorably?


5D3 / 5D2 / 16-35 4 IS / 24-70 2.8 II / 50 1.4 / 100L / 70-200 2.8 IS II | Flickr (external link) | www.zachwheeler.ca (external link) | Facebook (external link)

LOG IN TO REPLY
Shadowblade
Cream of the Crop
Joined Dec 2008
Melbourne, Australia
Oct 15, 2012 06:14 |  #42

ZachOly wrote in post #15124031external link
I know this is crazy, but what if DxO's software interprets the readings from Nikon/Sony sensors more favorably?

Possible, but, then, it wouldn't be just DxO which does this - others have replicated the tests and gotten the same results.

RAW is really a very simple format, and, if all you are doing is interpolating the raw luminance values for each pixel to form a colour image (rather than any of the other manipulation which can be done in converters) all of them should be basically the same.




LOG IN TO REPLY
pwm2
"Sorry for being a noob"
pwm2's Avatar
8,626 posts
Joined May 2007
Sweden
Oct 15, 2012 07:27 |  #43

Shadowblade wrote in post #15119630external link
Funny how other people who actually tests the sensors comes up with *exactly the same results* (with consistent variations depending on the chosen cut-offs of 'acceptable performance').

These guys haven't tested the 600D/650D yet, but, looking at other sensors and the discrepancy between Canon sensors and Nikon/Sony sensors (a few stops of DR at base ISO) you can see that DXOMark are not the only ones.

http://home.comcast.ne​t/~NikonD70/Charts/PDR​.htmexternal link

The people who blame DxO are normally the people who haven't read the documentation what they measure and how they present the information. So they think the figures doesn't make sense just because they don't understand the figures.

Especially how the figures are normalized. Yes - the figures aren't always so easy to use. But they are still relevant. And are backed by similar results from other sources.


5DMk2 + BG-E6 | 40D + BG-E2N | 350D + BG-E3 + RC-1 | Elan 7E | Minolta Dimage 7U | (Gear thread)
10-22 | 16-35/2.8 L II | 20-35 | 24-105 L IS | 28-135 IS | 40/2.8 | 50/1.8 II | 70-200/2.8 L IS | 100/2.8 L IS | 100-400 L IS | Sigma 18-200DC
Speedlite 420EZ | Speedlite 580EX | EF 1.4x II | EF 2x II

LOG IN TO REPLY
pwm2
"Sorry for being a noob"
pwm2's Avatar
8,626 posts
Joined May 2007
Sweden
Oct 15, 2012 07:31 |  #44

The Fox wrote in post #15120981external link
I always laugh at DxO scores. They score the D800 higher in dynamic range and in color accuracy higher then a 16 bit medium format camera. This is a joke, I am sorry but anyone that thinks that this is true needs to tell all the high level shooters that they need to trade up to a 35mm form DSLR. I shoot regularly with a MF camera for commercial clients and they deliver the image that is a step above that of any DSLR. Many of my clients would laugh and fire me for wasting their time with a DSLR when I need the color and dynamic range that MF can give me.

Nick

Your comment shows that you haven't looked closer at what DxO do say. They claim that 1mm2 of D800 is very good. They do not claim that the full-size sensor of a D800 matches a full-size sensor of a MF camera.


5DMk2 + BG-E6 | 40D + BG-E2N | 350D + BG-E3 + RC-1 | Elan 7E | Minolta Dimage 7U | (Gear thread)
10-22 | 16-35/2.8 L II | 20-35 | 24-105 L IS | 28-135 IS | 40/2.8 | 50/1.8 II | 70-200/2.8 L IS | 100/2.8 L IS | 100-400 L IS | Sigma 18-200DC
Speedlite 420EZ | Speedlite 580EX | EF 1.4x II | EF 2x II

LOG IN TO REPLY
pwm2
"Sorry for being a noob"
pwm2's Avatar
8,626 posts
Joined May 2007
Sweden
Oct 15, 2012 07:42 |  #45

The Fox wrote in post #15122668external link
MF digital will out resolve any D800, and the 16 bit color is going to give me the most information to work with. And going by the even how they test based on math, 16 bit has 65,523 tones per channel vs 16,383 on 14-bit. There is no way that you can not have better accuracy from the camera with almost 4 times the colors it can create per channel. The only way that they could is if the convert them both to 8 bit but I don't shoot in 8 bit so that is useless to me.

Question? How much statistics have you read at university level? You would be surprised if you just knew how wrong your above assumption is.

You can have more tones from a sensor with 1-bit sensor than you get from a 16-bit sensor. Because you get tones both from the number of discrete values each sample has, and from the averaged values of neighbor values.

Because of this, you can find 1-bit music players - just that the player don't run the DAC at 44100 Hz like you have on a CD but runs it at many times higher frequency. If you buy a light dimmer for your room, it's likely to have a 1-bit DAC - on or off - but driven with pulse-width control. So temporal resolution is used instead of the dynamic range of the 1-bit on/off regulation.

Spend some time and actually read the fine print at DxO. Make sure you understand. Come back and ask questions if something isn't clear. Just don't assume and post negative statements about them based on incorrect assumptions.


5DMk2 + BG-E6 | 40D + BG-E2N | 350D + BG-E3 + RC-1 | Elan 7E | Minolta Dimage 7U | (Gear thread)
10-22 | 16-35/2.8 L II | 20-35 | 24-105 L IS | 28-135 IS | 40/2.8 | 50/1.8 II | 70-200/2.8 L IS | 100/2.8 L IS | 100-400 L IS | Sigma 18-200DC
Speedlite 420EZ | Speedlite 580EX | EF 1.4x II | EF 2x II

LOG IN TO REPLY
sponsored links
(this ad will go away when you log in as registered member)

10,385 views & 0 likes for this thread
DxOMark score for EOS 650D (T4i) Out: same score as the EOS 20D from 2004
FORUMS News & Rumors Photography Industry News


Not a member yet? Click here to register to the forums.
Registered members get all the features: search, following threads, own gear list and ratings, likes, more forums, private messaging, thread follow, notifications, own gallery, settings, view hosted photos, own reviews and more...


AAA

Send feedback to staff    •   Jump to forum...    •   Rules    •   Index    •   New posts    •   RTAT    •   'Best of'    •   Gallery    •   Gear    •   Reviews    •   Polls

COOKIES DISCLAIMER: This website uses cookies to improve your user experience. By using this site, you agree to our use of cookies. Privacy policy and cookie usage info.

POWERED BY AMASS 1.4version 1.4
made in Finland
by Pekka Saarinen
for photography-on-the.net
Spent 0.00382 for 6 database queries.
PAGE COMPLETED IN 0.02s
Latest registered member is highlights
794 guests, 296 members online
Simultaneous users record so far is 6106, that happened on Jun 09, 2016