--
Rikke
-
Welcome to the new forums! Please read this first. For known issues we are working to resolve, click here.
Confused by DxO Mark in DR E510 vs. E5
- Thread starter feinschmecker
- Start date
Rikke
That it isn't clear to you, doesn't mean that it isn't clear.
--
Rikke
rovingtim
Veteran Member
sorry posted in the wrong spot
rovingtim
Veteran Member
A friend of mine and I have been RAW testing numerous cameras against each other and pushing the files to see what's what. This includes m4/3rds, 4/3rds, Nikon cropped cameras (old and the new Sony sensor) Canon 1Ds MIII and Nikon full frame cameras.
We've spent a lot of time examining files taken at the same time and the same conditions and you know what? If you push their RAW files to the limit, you find DXO to be bang on.
For example, they say the the Nikon 5100 (new Sony sensor) is cleaner at base then the Nikon D3s. Guess what? When we tested them the D5100 was better ... significantly. When we compare the E3 to all these cameras we fine it performs exactly where DXO says it should be.
For example, DXO says the E3 has an ever so slightly better signal noise ratio than the E5 and yet people swear blind that the E5 is cleaner. Two things here: 1) if you don't sharpen the E3 files they look very clean indeed. 2) the E5 is indicating ISO's almost a stop under what the E3 is indicating. For example, when the E5 indicates 800 ISO the measured ISO was 486 ISO whereas the E3 indicating 800 ISO is measured at 804 ISO.
Funnily enough, the E5 looks cleaner when indicating 800 ISO compared to the E3.
If you are shooting out of camera, or you are applying only one kind of processing (that will inevitably favour one camera over another) then you will not fully understand what DXO is saying.
It's all there for those who want to see.
We've spent a lot of time examining files taken at the same time and the same conditions and you know what? If you push their RAW files to the limit, you find DXO to be bang on.
For example, they say the the Nikon 5100 (new Sony sensor) is cleaner at base then the Nikon D3s. Guess what? When we tested them the D5100 was better ... significantly. When we compare the E3 to all these cameras we fine it performs exactly where DXO says it should be.
For example, DXO says the E3 has an ever so slightly better signal noise ratio than the E5 and yet people swear blind that the E5 is cleaner. Two things here: 1) if you don't sharpen the E3 files they look very clean indeed. 2) the E5 is indicating ISO's almost a stop under what the E3 is indicating. For example, when the E5 indicates 800 ISO the measured ISO was 486 ISO whereas the E3 indicating 800 ISO is measured at 804 ISO.
Funnily enough, the E5 looks cleaner when indicating 800 ISO compared to the E3.
If you are shooting out of camera, or you are applying only one kind of processing (that will inevitably favour one camera over another) then you will not fully understand what DXO is saying.
It's all there for those who want to see.
Rriley
Forum Pro
- Messages
- 21,846
- Solutions
- 1
- Reaction score
- 474
in an argument between noise and detail, how does DxO, which doesnt measure resolution, determine the victor in noise. For inevitably as ISO climbs higher and noise is more apparent, the real measure is what detail survives...
of course, i dont expect you to answer as people do in conversation, just hide another eggshell statement as though you replied without being contested
sounds like your view is rather like religion
--
Riley
any similarity to persons living or dead is coincidental and unintended
feinschmecker
Forum Enthusiast
OK, one more question: People here have talked about that the DxO marks making only sense when processing from RAW, not looking at JPEGs.
I would like to ask a question about that: I am not a big PP person, but what I do do is to shoot RAW and then to adjust contrast, WB and also the exposure. So I guess via that processing: simply adjusting the exposure, I should see what DxOmark has measured in terms of DR, right?
I am asking because some people in this thread have stated that this DR score would come out when adjusting the tone curve, but I am not really sure what's meant with this - is that in principle the same as adjusting the exposure?
Cheers,
Pablo
I would like to ask a question about that: I am not a big PP person, but what I do do is to shoot RAW and then to adjust contrast, WB and also the exposure. So I guess via that processing: simply adjusting the exposure, I should see what DxOmark has measured in terms of DR, right?
I am asking because some people in this thread have stated that this DR score would come out when adjusting the tone curve, but I am not really sure what's meant with this - is that in principle the same as adjusting the exposure?
Cheers,
Pablo
feinschmecker
Forum Enthusiast
So you are saying that the E5 and E3 have the same noise, but that the E5 simply measures ISO differently? Or did I misunderstand that?
Cheers,
Pablo
Cheers,
Pablo
Robert Schroeder
Senior Member
DxOMark's ISO numbers are wrong – they are not ISO numbers at all. They claim to adhere to "the" ISO 12232 specification, but there is much doubt to that, since the one of the ISO standard's several options to define ISO speed which comes closest to what they seem to be doing exists only for the processed JPEG, not for RAW data, and it is for a reason that it only exists for the processed JPEG.
And because they're determining ISO the way they do, the E-30 get's a 1 EV penalty for ISO speeds of 200 and above.
In http://www.iemp.net/rs/fototechnik/e30-dr/index.html and http://www.iemp.net/rs/fototechnik/e30-noise/index.html I tried to visualize how the E-30's compares to the E-510 regarding highlight dynamic range and noise (i.e. shadow dynamic range), back when the E-30 was still new, using RAW and developing with Silkypix. I'd expect the E-5 to behave very similarly to the E-30 in this regard.
Cheers,
Robert
And because they're determining ISO the way they do, the E-30 get's a 1 EV penalty for ISO speeds of 200 and above.
In http://www.iemp.net/rs/fototechnik/e30-dr/index.html and http://www.iemp.net/rs/fototechnik/e30-noise/index.html I tried to visualize how the E-30's compares to the E-510 regarding highlight dynamic range and noise (i.e. shadow dynamic range), back when the E-30 was still new, using RAW and developing with Silkypix. I'd expect the E-5 to behave very similarly to the E-30 in this regard.
Cheers,
Robert
Nemianiu Skqergl
Well-known member
the exposure is the amount of light in the photo
it is set when you take it
processing just changes how it looks
it cant make there be more light
rovingtim
Veteran Member
Yes. Noise to signal ratio is a measurable fact. When measured, the E3 and E5 are practically the same.
What you see in the final files can be influenced by several factors. One good thing about the E5 is that because of the weaker AA filter, the files don't need as much sharpening. As such, noise isn't sharpened as much and therefore the final file looks cleaner. Also, the E5 doesn't band as much so even though it is slightly noisier than the E3 at high ISO's, the E5 creates more useable files in many cases.
However, the core technology of the E3 and E5 sensor is pretty much the same. As such, the noise is pretty much the same, as are all the other measurable qualities of the sensor.
Overall (if you don't mind false detail and more moire ... which many people of this forum gladly accept in exchange for increased sharpness) the E5 is a better camera than the E3 but more because the lighter AA creates less of a need for sharpening (and therefore less apparent noise) and has more advanced noise suppression. Most importantly, it is not as prone to banding as the E3 is.
Caveat -- if you look into this forum's history you will see that many insist the E3 doesn't band at all. Personally, I found banding. Of course, we were also informed by knowledgeable forum members that the E3 that it was razor sharp ...
Edit -- re your question on DR
jpgs are 8 bit and have default curves placed on the files. RAW is 12 bit and has all the information possible available. You always throw out information to make a jpg. If you simply look at the jpgs, you aren't looking at the full potential of the camera.
Olympus are very good at making jpgs meaning partly because they put much more emphasis on high quality jpgs than other manufacturers. So while working in RAW can perform apparent miracles for other cameras, working in RAW is frequently unnecessary with Oly because they have already done a really good job for many types of images.
What you see in the final files can be influenced by several factors. One good thing about the E5 is that because of the weaker AA filter, the files don't need as much sharpening. As such, noise isn't sharpened as much and therefore the final file looks cleaner. Also, the E5 doesn't band as much so even though it is slightly noisier than the E3 at high ISO's, the E5 creates more useable files in many cases.
However, the core technology of the E3 and E5 sensor is pretty much the same. As such, the noise is pretty much the same, as are all the other measurable qualities of the sensor.
Overall (if you don't mind false detail and more moire ... which many people of this forum gladly accept in exchange for increased sharpness) the E5 is a better camera than the E3 but more because the lighter AA creates less of a need for sharpening (and therefore less apparent noise) and has more advanced noise suppression. Most importantly, it is not as prone to banding as the E3 is.
Caveat -- if you look into this forum's history you will see that many insist the E3 doesn't band at all. Personally, I found banding. Of course, we were also informed by knowledgeable forum members that the E3 that it was razor sharp ...
Edit -- re your question on DR
jpgs are 8 bit and have default curves placed on the files. RAW is 12 bit and has all the information possible available. You always throw out information to make a jpg. If you simply look at the jpgs, you aren't looking at the full potential of the camera.
Olympus are very good at making jpgs meaning partly because they put much more emphasis on high quality jpgs than other manufacturers. So while working in RAW can perform apparent miracles for other cameras, working in RAW is frequently unnecessary with Oly because they have already done a really good job for many types of images.
The sensors have the same size so the photon shot noise part of the total noise is of course the same for both.
As have others, myself included. The same is true of the 5DII, by the way. In that world there are also many who will insist it isn't so.
But of course and an E-510 with a kit lens beats everything else.
--
Rikke
jim stirling
Veteran Member
Perhaps instead of being confused by DXO or those praising it and denying it, simply download some of the hundreds of freely available RAW files taking in reasonably controlled scenarios and directly compare the NR off shots to your heart’s content. Not surprisingly the most discontent with DXO comes from those whose much loved camera choice does relatively poorly. For what it’s worth I have found that the differences predicted by the DXO graphs typically reflect real life results obtained from the respective cameras especially when compared to each other.
Jim
Jim
rovingtim
Veteran Member
I'm not going to go into all the problems here, but just address one.
On the second link he says: 'Method: ISO-Series shot with both cameras (50-200 at 50mm, f/6.3, tungsten lighting), RAW files chosen with similar brightness, developed with Silkypix 3 (and with identical settings: my personal defaults for E-510 ISO 100), took comparable crops (E-30: 806x604, E-510: 730x547), scaled up with PMView Pro to the same size (1024x768), then cropped them again (300x400). '
So here's the problem -- he says 'RAW files chosen with similar brightness'. So, in the E5 and E3 situation he would set the ISO to 800, for example, and then do different exposures and then picked two with 'similar brightness' ... so if the E5 is exposing more for a given ISO, then that would disappear when he picked file of 'similar brightness'.
To do this right, you have to pick the same exposure in the same lighting and then see if the files are the same brightness. That tells you how the sensor is exposing.
Then the two images taken at the same exposure need to be adjusted to the same brightness in order to check for noise.
On the second link he says: 'Method: ISO-Series shot with both cameras (50-200 at 50mm, f/6.3, tungsten lighting), RAW files chosen with similar brightness, developed with Silkypix 3 (and with identical settings: my personal defaults for E-510 ISO 100), took comparable crops (E-30: 806x604, E-510: 730x547), scaled up with PMView Pro to the same size (1024x768), then cropped them again (300x400). '
So here's the problem -- he says 'RAW files chosen with similar brightness'. So, in the E5 and E3 situation he would set the ISO to 800, for example, and then do different exposures and then picked two with 'similar brightness' ... so if the E5 is exposing more for a given ISO, then that would disappear when he picked file of 'similar brightness'.
To do this right, you have to pick the same exposure in the same lighting and then see if the files are the same brightness. That tells you how the sensor is exposing.
Then the two images taken at the same exposure need to be adjusted to the same brightness in order to check for noise.
rovingtim
Veteran Member
... the Panasonic L10 was the first 4/3rds camera to 'lift the veil'. The E510 was sharp, but not that sharp. I'm waiting for someone with an L10 and an E5 to post in order to see which has the light AA.
The L10 is a moire beast.
The L10 is a moire beast.
Can you explain this further please? I don't know what 'photon shot noise' is.
Rriley
Forum Pro
- Messages
- 21,846
- Solutions
- 1
- Reaction score
- 474
well exactly
along with those that just plain have doubts
that includes beta testers oriented by DxO methodology
i havent found DxO more influential than the reality stated previously
is that a draw?
--
Riley
any similarity to persons living or dead is coincidental and unintended
I was making fun of an old JK comment.
The problem of light or absent AA.
The photon shot noise is a fundamental physical phenomenon ultimately arising from Heisenberg's uncertainty principle. In layman's terms there is a fluctuation in the number of photons that arrive at the sensor and it turns out that when there are sufficiently many photons (there normally are
) the S/N ratio will be the square root of the number of photons detected by the sensor. How many photons that will reach the sensor for a given exposure setting is of course proportional to the size of the sensor (an FF sensor gets about four times as many as an FT sensor). In our case the sensor size is the same for the cameras we are discussing in this thread, so the noise contributed by nature as part of the incident light will be the same for any Olympus DSLR camera from the E-1 to the E-5. What can be improved is the so-called quantum efficiency of the sensor (think of it as how good the sensor is to detect that a photon has hit it) and amplifier noise and so on. And let's of course not forget the ability to avoid banding that you brought up (which can totally ruin a shot). Getting rid of that would be a big improvement even if nothing else changed. --
Rikke
rovingtim
Veteran Member
Thanks for this.
While I'm not a trained expert, I do know a bit about quantum physics. I'm most interested when quantum characteristics appear in the macro world ... such as 'rogue waves' (solitons, if I'm not mistaken).
While I'm not a trained expert, I do know a bit about quantum physics. I'm most interested when quantum characteristics appear in the macro world ... such as 'rogue waves' (solitons, if I'm not mistaken).
Rriley
Forum Pro
- Messages
- 21,846
- Solutions
- 1
- Reaction score
- 474
where did he actually say that "an E-510 with a kit lens beats everything else"
or are you busy discrediting people without any real factual credence, as usual
--
Riley
any similarity to persons living or dead is coincidental and unintended
Rriley
Forum Pro
- Messages
- 21,846
- Solutions
- 1
- Reaction score
- 474
shot noise requires nothing of 'quantum physics'
and has been negotiated here many many times before
--
Riley
any similarity to persons living or dead is coincidental and unintended
You just discredited yourself even further.
By mental giants like you no doubt.
--
Rikke
Keyboard shortcuts
- f
- Forum