My Musing on the 6D vs D600... And Canon in General right now

quickben wrote in post #15094086
And completely different to DPreview's controlled studio test.

DPReview doesn't correct for sensor resolution - it compares 100% crops, so higher-resolution sensors are always disadvantaged (since the 100% crop is taken using a smaller percentage of the sensor than the 100% crop from a lower-resolution sensor).

They were the same website which had the 5D2 looking worse than the D700 and 5Dc, and the 50D looking worse than the 40D, since their samples were all viewed at 100% rather than at the same final output size.

Shadowblade wrote in post #15097569
DPReview doesn't correct for sensor resolution - it compares 100% crops, so higher-resolution sensors are always disadvantaged (since the 100% crop is taken using a smaller percentage of the sensor than the 100% crop from a lower-resolution sensor).

That sounds like a major miscalculation. Something they need to address urgently if they hope to be taken seriously.

Shadowblade wrote in post #15097569
DPReview doesn't correct for sensor resolution - it compares 100% crops, so higher-resolution sensors are always disadvantaged (since the 100% crop is taken using a smaller percentage of the sensor than the 100% crop from a lower-resolution sensor)....

Lowner wrote in post #15097758
That sounds like a major miscalculation. Something they need to address urgently if they hope to be taken seriously.

I rather prefer that they publish the unmodified sensor resolution as that is what the sensor delivers. Once you start to introduce resizing it adds another variable and you are no longer seeing what the sensor is delivering but something entirely else.

gjl711 wrote in post #15097915
I rather prefer that they publish the unmodified sensor resolution as that is what the sensor delivers. Once you start to introduce resizing it adds another variable and you are no longer seeing what the sensor is delivering but something entirely else.

Unless you take total size into context, though, the unmodified file doesn't tell you very much. An 800x800 100% crop of a 12MP sensor represents a much greater proportion of the final image than an 800x800 100% crop of a 46MP sensor, so you're effectively looking at the higher-resolution sensor at a greater magnification. If the image from the 12MP sensor looks better (which it most probably will), all you can really say from that is that a 20x30" print from the 12MP sensor will look better at 12" viewing distance than a 40x60" print from the 46MP sensor at the same 12" viewing distance.

Shadowblade wrote in post #15097924
Unless you take total size into context, though, the unmodified file doesn't tell you very much. An 800x800 100% crop of a 12MP sensor represents a much greater proportion of the final image than an 800x800 100% crop of a 46MP sensor, so you're effectively looking at the higher-resolution sensor at a greater magnification. If the image from the 12MP sensor looks better (which it most probably will), all you can really say from that is that a 20x30" print from the 12MP sensor will look better at 12" viewing distance than a 40x60" print from the 46MP sensor at the same 12" viewing distance.

But if one is intelligent enough to realize that, then I'd much rather see the totally untouched file and make my own judgement.

It's like that Fred Miranda review of the D800 vs. 5D3; he enlarges the 5D's file size to match the D800 (to be sure not to give the D800 too big an advantage), then knocks the 5D for showing noticeably less detail. Who knows what program was used to enlarge, but no program out there can magically fill in the pixels that weren't there to begin with. That's the same thought I've got when I see images resized for comparrison (and part of my knock on DxO, but that's for another thread). Downsizing gives certain advantages, and upsizing gives certain disadvantages.

Knowing how to sort your way through it all is the real trick.

bjyoder wrote in post #15098279
But if one is intelligent enough to realize that, then I'd much rather see the totally untouched file and make my own judgement.

Knowing how to sort your way through it all is the real trick.

I certainly would not be aware of the implications, even if they admitted what they do, or fail to do. They need to explain and maybe show the comparison both ways.

gjl711 wrote in post #15097915
I rather prefer that they publish the unmodified sensor resolution as that is what the sensor delivers. Once you start to introduce resizing it adds another variable and you are no longer seeing what the sensor is delivering but something entirely else.

See my comment about how DPReview should be using Zeiss primes, Not Canon/Nikon ones to remove the variable that the lens could be adversely affecting resolution.. As I said, I swear i saw some nasty CA on the Canon samples that were NOT in the Nikon ones...

Lowner wrote in post #15098334
I certainly would not be aware of the implications, even if they admitted what they do, or fail to do. They need to explain and maybe show the comparison both ways.

that being the case I don't value the Fred Miranda review as much because he has now changed the 5Dmk3 file. all files should be untouched and at 100%. This will allow both images to be viewed at it's native resolution.

Enlarging the file is going to create pixels that wasn't there during the press of the shutter button. The same goes if someone reduced the D800 image to be the same file size as the 5Dmk3. By reducing the image size your throwing away pixels. in both cases your creating or removing pixels that may contain noise or detail so it's not a 100% accurate representation of the output of a camera.

I would trust dpreview.com review of both cameras than the Fred Miranda review.

KenjiS wrote in post #15099276
See my comment about how DPReview should be using Zeiss primes, Not Canon/Nikon ones to remove the variable that the lens could be adversely affecting resolution.. As I said, I swear i saw some nasty CA on the Canon samples that were NOT in the Nikon ones...

that's not a bad idea I don't think they would do that. One could say that the output is what you would expect if your using canon camera with canon lenses.