Re: DxOMark rates the Olympus Pen E-P1 RAW performance
Quote:
Originally Posted by Ian
This is what prompted me to post the thread earlier in the week about whether or not high ISO E-P1 improvements were RAW or JPEG-only.
DxOMark also pours cold water over the suggestion that ISO 200 is better than ISO 100.
Ian
Ian,
As far as I can see, there is no place in their assessment (DxO) where optics comes in. They seem to be measuring the sensor response to coarse structures only, things like:
how much light is needed to make the A/D converter clip (their ISO definition)
if we illuminate it at this level, how much noise is in the raw file
etc.
The sensors in the E-P1 and the E-30/620 are about the same apart from the optics in front (e.g. AA filter). We know from the "horses mouth" (Olympus) that the E-P1 has a weaker optical AA filter than the SLRs and the TruePic V is needed to deal with the problems that causes (Moiree, aliasing etc.).
So what do we have here? Once the light reaches the actual photo-sites of the sensor, there has been less optical blurring and the image is crisper. The sensor now converts that into an electric signal, which after a few amplifiers etc. gets converted into digital (A/D converter). At least some of the increased crispness (due to the weaker AA) carries through until behind the converter. Now comes the image processing. Since the image is less blurry to begin with, we need less sharpening to get to a similar level of fine detail. Less sharpening also means less noise. Isn't it all magic
If they indeed (I am not 100% certain) do not take the required sharpening levels into account the results are in no way surprising. The electrical properties of sensor in the E-30/620 and the E-P1 are the same. That's what they are seeing with their measurement.
People using the cameras to take photos, apply less sharpening (either they themselves or the person who tuned the processing engine (either in camera or external RAW converter)) to the E-P1 images than to the E-30/620 images and get less noise. Intop of this, Olympus might have used a better/different NR algorithm in the E-P1 than the earlier E-30/620.
Re: DxOMark rates the Olympus Pen E-P1 RAW performance
Quote:
Originally Posted by joachim
Ian,
As far as I can see, there is no place in their assessment (DxO) where optics comes in. They seem to be measuring the sensor response to coarse structures only, things like:
how much light is needed to make the A/D converter clip (their ISO definition)
if we illuminate it at this level, how much noise is in the raw file
etc.
The sensors in the E-P1 and the E-30/620 are about the same apart from the optics in front (e.g. AA filter). We know from the "horses mouth" (Olympus) that the E-P1 has a weaker optical AA filter than the SLRs and the TruePic V is needed to deal with the problems that causes (Moiree, aliasing etc.).
So what do we have here? Once the light reaches the actual photo-sites of the sensor, there has been less optical blurring and the image is crisper. The sensor now converts that into an electric signal, which after a few amplifiers etc. gets converted into digital (A/D converter). At least some of the increased crispness (due to the weaker AA) carries through until behind the converter. Now comes the image processing. Since the image is less blurry to begin with, we need less sharpening to get to a similar level of fine detail. Less sharpening also means less noise. Isn't it all magic
If they indeed (I am not 100% certain) do not take the required sharpening levels into account the results are in no way surprising. The electrical properties of sensor in the E-30/620 and the E-P1 are the same. That's what they are seeing with their measurement.
People using the cameras to take photos, apply less sharpening (either they themselves or the person who tuned the processing engine (either in camera or external RAW converter)) to the E-P1 images than to the E-30/620 images and get less noise. Intop of this, Olympus might have used a better/different NR algorithm in the E-P1 than the earlier E-30/620.
Re: DxOMark rates the Olympus Pen E-P1 RAW performance
Quote:
Originally Posted by Ian
This is what prompted me to post the thread earlier in the week about whether or not high ISO E-P1 improvements were RAW or JPEG-only.
DxOMark also pours cold water over the suggestion that ISO 200 is better than ISO 100.
Ian
Another weird one. I cannot fathom how DxO manage to get their numbers. I'm wondering if the sensor size is somehow being factored in. Based on the inspection of the RAW files, there didn't seem much doubt the E-P1's RAW performance at 3200 visibly bested the E-30 or E-3 - I don't think we were all kidding ourselves. When you start looking at the DR vs the competition, DxO are claiming cameras like the 40D have at least an extra stop of DR than the E-30 or E-3 - which, again, just doesn't seem to be correct. I've said it before, but I will stick to DPReview's tests...
Re: DxOMark rates the Olympus Pen E-P1 RAW performance
Quote:
Originally Posted by Ian
Joachim - that makes a lot of sense.
Ian
+1 - this does make a lot of sense. I wonder if somehow there are comparing based on a linear tone curve translation of the RAW data rather than the in-camera/ACR/Oly Studio one. This would explain the lower apparent dynamic range as the numbers are effectively compressed in the final stop of DR until clipping occurs. I don't wish to be critical of DxO - but their lack of explanation and lack of translation to real-world performance is frustrating.
Another obvious point is that a less blurry image is also going to look less blurry after noise reduction - so the result is going to look a lot better and more detailed.
DxOMark rates the Olympus Pen E-P1 RAW performance
Linear Mode
