DxOMark has added a lens/body combo rating

Take a look at the lens/body combo ratings. - TF

Thanks for posting this. The DxOMark content has been very impressive, right from the start. I should really write an article about the new lens ratings...

Evidently I forgot the link. Here it is.

http://www.dxomark.com/index.php/en/...ngs/Peak-Score

Somewhat of a surprise was that, with the <$3000 bodies (this would cover even my 'dream' cameras), the body didn't much matter. The lenses have the same score regardless of body.

TF

That really isn't a surprise, I always thought lens quality surpasses sensor resolution until you get well into the Pro body range.

I really wished they would hold 8-bit standard measurements, understandably its not 'peak performance' but its more practical and relevant, actually means something to the end user. I am surprised how well the Sony combo's fair in their tests though considering the 12 to 14 bit disparity.

*edit - not saying holding 8-bit exclusively, but cross reference 8-bit in their testing. Either by doing jpeg engine, converting in raw, or compressed 8-bit raw standards.

Why don't they do that anyway?

Quote:

---

Originally Posted by Anbesol

I really wished they would hold 8-bit standard measurements, understandably its not 'peak performance' but its more practical and relevant, actually means something to the end user. I am surprised how well the Sony combo's fair in their tests though considering the 12 to 14 bit disparity.

*edit - not saying holding 8-bit exclusively, but cross reference 8-bit in their testing. Either by doing jpeg engine, converting in raw, or compressed 8-bit raw standards.

Why don't they do that anyway?

---

This is an interesting idea that at first glance I didn't think was a great idea. However after a second or third thought I'd add or amend this by saying that a rating for in-camera default JPEG conversion might be a good idea.

The most recent cameras do a very good job and any weaknesses are usually user induced or are white balance related. At least one brand in one model has a very specific color cast that is noticeable. There are certainly other items as well including de-mosaicing implementations that could be looked at too.

A big drawback to this type of evaluation is so much data is lost in the 8-bit data that in many cases differences are literally in the noise and not quantifiable to a meaningful level.

Of course this evaluation doesn't exactly relate to DxO's lens/body combination ratings either!

Still could be intriguing, but how meaningful I am not sure. Some of the DxO work now is already in the realm of the esoteric.

Yeah I agree with you there, thats why I dont think they should do the JPEG instead, but as well as.

Its also more realistic to print quality, considering most prints aren't printed beyond 8-bit gamut anyway, even if you send it to the printer as 16-bit, it still converts to 8-bit. And how many use monitors capable of displaying the full 16-bit gamut? The difference of 12-bit and 14-bit is a lot more in math than it is in real world too, a maximum cap of 68.7 Billion colors vs the 4.4 trillion of 14 bit. And our eyes are accustomed to imaging almost exclusively limited to 16.8 million colors. Hence, there lab findings are subject to this difference, a difference that is more even-leveled on a Jpeg engine. *edit - and more practical to real world results.

Quote:

---

Originally Posted by Anbesol

Yeah I agree with you there, thats why I dont think they should do the JPEG instead, but as well as.

Its also more realistic to print quality, considering most prints aren't printed beyond 8-bit gamut anyway, even if you send it to the printer as 16-bit, it still converts to 8-bit. And how many use monitors capable of displaying the full 16-bit gamut? The difference of 12-bit and 14-bit is a lot more in math than it is in real world too, a maximum cap of 68.7 Billion colors vs the 4.4 trillion of 14 bit. And our eyes are accustomed to imaging almost exclusively limited to 16.8 million colors. Hence, there lab findings are subject to this difference, a difference that is more even-leveled on a Jpeg engine. *edit - and more practical to real world results.

---

There are more and more 12/14/16 bit printers coming on the market all the time. Many are large professional inkjets (i.e. Canon 9xxx series) but they are appearing. There is a very visible difference between 8 and 12 bit implementations. 12 - 16 so far not as much though with certain images it is noticeable in certain color ranges. The ink sets can be revealed as good bad or indifferent with the higher bit images!

So much of this though comes from post process manipulation and refinement that the camera is only part of the final output.

I'd still like to see an additional real world quantification, particularly with the lens/body combos including CA tendencies, color clipping or stair-stepping (from JPEG reduction/conversions), and general overall color accuracy of the in body JPEG algorithms!

Woah! I didn't know 16-bit printing was available to consumers yet. Thanks for catching me up to speed.

I agree, the jpeg engine is different enough that the results would be very different too, some of us use JPEG a lot too, and would find it useful.