droopy1592 wrote:
Maybe the 12 mp would keep it cheap with the minor upgrades. The sensor is what makes rhe mkii expensive. Take it out and put the old one in and it sounds like the perfect cheap FF.
Hmmmm, the 5D was about 1400-1500 euros at its lowest point just before Canon released the 5D2. Give it some time and the 5D2 will be as cheap too
Honestly, I was a little bit sceptic about the 5D2's sRAW1 capability. But sRAW1 (despite being a little bit softer than when downsizing a 21MP RAW file) produces excellent quality. And it has less noise too (than when downsizing a full RAW to sRAW size).
So, basically you have the 5D incorporated into the 5D2 + all the extras. It would have been nice though if Canon had added all the 7D goodies to the 5D2. That would have made it an excellent all-round performer
Liquidstone wrote:
How about the D3X, DPR must be biased against that too according to your theory?
It is not theory that DPR compares noise between cameras at different magnifications. Go and look at any of their stamp comparisons. When cameras have different numbers of pixels, the size of the Queen's head in the crop varies in proportion.
(1) DPR compares 100% crops in order to make judgements about noise.
(2) that heavily biases the result in favour of the lower resolution camera.
(3) DXO also shows per pixel comparison by default.
(4) most people don't understand why that makes comparison entirely bogus.
brainiac wrote:
It is not theory that DPR compares noise between cameras at different magnifications. Go and look at any of their stamp comparisons. When cameras have different numbers of pixels, the size of the Queen's head in the crop varies in proportion.
(1) DPR compares 100% crops in order to make judgements about noise.
(2) that heavily biases the result in favour of the lower resolution camera.
(3) DXO also shows per pixel comparison by default.
(4) most people don't understand why that makes comparison entirely bogus.
Those are facts, not theory.
That's not the point I was responding to. Let me quote again below the whole paragraph you posted:
brainiac wrote:
I sent dpr a very polite email about this issue, detailing why I thought their noise tests were thoroughly misleading to the average photographer. I never received an acknowledgement, let alone a response. I believe that someone at DPR is a Nikon loyalist and so is quite happy with their noise comparison, since it biases noise comparisons heavily in favour of 12 Mpixels (D3, D700) against 21 (1Ds3, 5D2).
If someone who is a Nikon loyalist at DPR is biased against 21 MP sensors, that bias should also work against the D3x with its even denser pixels? This seems to be a paradox to me.
On a slightly different but related note, I used DXOMark to compare Canon 40D to the Nikon D200.
Canon 40D:
Color Depth = 22.1
Dynamic Range = 11.3
Low Light ISO = 703
Nikon D200:
Color Depth = 22.3
Dynamic Range = 11.5
Low Light ISO = 583
Everyone knows the D200 is a noisy bugger at high ISO and its low light ISO score supports this.
YET, DXOMark assigns an overall score of 64.2 to D200 and 'only' 63.5 to 40D. Is there a reasonable explanation of this or is this pure BIASNESS? If there is no satisfactory explanation for this, I will classify DXOMark as another extremely biased site, not unlike DPReview and Rob Galbraith.
Seems like all the major review sites are up in arms against Canon. Why?
Liquidstone wrote:
If someone who is a Nikon loyalist at DPR is biased against 21 MP sensors, that bias should also work against the D3x with its even denser pixels? This seems to be a paradox to me.
I think brainiac's problem with DPReview is the latter's obsession with pixel level scrutiny instead of comparing images at the same print size.
But make no doubt about this: DPReview is strongly biased against Canon. When Canon was once upon a time the only champion of FF cameras, DPReview questioned the need for existence of FF sensors. They openly challenged and disparaged images from FF cameras by picking on their struggles with vignetting. YET, when the D3 was announced, vignetting was suddenly deemed as being aesthetically pleasing.
thw2 wrote:
I think brainiac's problem with DPReview is the latter's obsession with pixel level scrutiny instead of comparing images at the same print size.
But make no doubt about this: DPReview is strongly biased against Canon. When Canon was once upon a time the only champion of FF cameras, DPReview questioned the need for existence of FF sensors. They openly challenged and disparaged images from FF cameras by picking on their struggles with vignetting. YET, when the D3 was announced, vignetting was suddenly deemed as being aesthetically pleasing.
Now, what do you call that?
I won't argue in this thread for or against DPR's testing protocols. I just wish Brainiac will clarify his paradoxical post.
thw2 wrote:
Is there a reasonable explanation of this or is this pure BIASNESS? If there is no satisfactory explanation for this
I think it's unsatisfactory testing methodology rather than bias [per se. As anyone who knows those cameras will be painfully aware (and I am that soldier - owned and hated the D200, own and really like the 40D) the D200 is so far behind the 40D in any real world IQ context (and that certainly included DR at both ens of the histogram), that the DxO values are nithing short of hysterically funny.
I will classify DXOMark as another extremely biased site, not unlike DPReview and Rob Galbraith.
Liquidstone wrote:
I won't argue in this thread for or against DPR's testing protocols. I just wish Brainiac will clarify his paradoxical post.
Romy's making a fair point.
Richard (rightly IMHO) has argued that DPR testing methodologies favour comparisions of smaller sensors against larger sensors: Romy is pointing out that if this a brand-based bias, the D3X will suffer from it too, which is counter-intuitive if this really is about brand...
Another few FACTS to consider next to Richards facts earlier:
(1) resizing with bicubic normal in PS is the same thing as using NR - it lowers pixel-to-pixel contrast by at least 15% more than the scaling in itself would suggest.
(2) High-pixel cameras do very well in normal circumstances, but if you're looking for maximum latitude in exposure, fewer MP still wins - even with the highMP camera downscaled.
(3) Per pixel quality has in fact almost DECREASED ever since '2004 if you plot the general trend over all models.... Which means that the numerical increase in MP is in fact screwing you with marketing trickery.
This is all irrelevant right now though, if we're to get back on track - since the 7D is the first Canon in a VERY long time that does both things right (a pixel density increase AND an individual pixel quality increase) compared to the previous comparable model - the 50D. Rejoice!
I'll repost an example in this thread (it has nothing to do with the text above), I made a small comparison of the only thing that's currently available for scrutiny - ImagingReview examples at ISO3200. I cannot convert 7D raws yet, so they're all from jpgs. One thing that is immediately noticeable if you analyze the pictures is that even with teh NR set to "off" the 50D and the 7D does A LOT of noise reduction. Chroma smearing is on the level of 5-6 pixels, and more in the horizontal direction (). This starts from quite early in the ISO range.
Do also notice that the 7D cheats even more than the other cameras with the "ISO" specification - it needs 1/3-step longer shuttertimes at all ISO's to achieve the same exposure at the same aperture with the same lens as the 50D. The 50D is at the true value of ~ISO2350 when the camera display says "ISO3200". The 7D is roughly 1/3Ev lower than that - which would give a true ISO of ~2000 or lower when the camera says "3200". This I cannot compensate for in the comparison, not without raw access.
Anyway, here's the comparison. All pictures scaled to the same relative scale (40D and Nikon D5000 scaled up, 50D and 7D scaled down). I chose a comparison resolution somewhere in between the 50D and the D5000.
I've just pre-ordered mine (Warehouse Express in the UK)
I got one of the first 20Ds when they came out and have been continually pleased with it since. The 7D seems like the first worthy upgrade to me.
Sadly I'm 60th on the list, so I don't know if I'll be in the first batch or not...
theSuede wrote:
(2) High-pixel cameras do very well in normal circumstances, but if you're looking for maximum latitude in exposure, fewer MP still wins - even with the highMP camera downscaled.
Huh? Have you looked at 24 MP D3X vs 12 MP D3/D700?
theSuede wrote:
(3) Per pixel quality has in fact almost DECREASED ever since '2004 if you plot the general trend over all models.... Which means that the numerical increase in MP is in fact screwing you with marketing trickery.
Are you referring to Canon alone or across all brand names?
Make no mistake about this: the 12 MP D300s/D90/D5000 (released in 2008-2009) are better than 10 MP D200/D80 (released in 2006). Similarly, the 10 MP 40D (released in 2007) is better than 8 MP 30D (released in 2006).
theSuede wrote:
I'll repost an example in this thread (it has nothing to do with the text above), I made a small comparison of the only thing that's currently available for scrutiny - ImagingReview examples at ISO3200. I cannot convert 7D raws yet, so they're all from jpgs. One thing that is immediately noticeable if you analyze the pictures is that even with teh NR set to "off" the 50D and the 7D does A LOT of noise reduction. Chroma smearing is on the level of 5-6 pixels, and more in the horizontal direction (). This starts from quite early in the ISO range.
Do also notice that the 7D cheats even more than the other cameras with the "ISO" specification - it needs 1/3-step longer shuttertimes at all ISO's to achieve the same exposure at the same aperture with the same lens as the 50D. The 50D is at the true value of ~ISO2350 when the camera display says "ISO3200". The 7D is roughly 1/3Ev lower than that - which would give a true ISO of ~2000 or lower when the camera says "3200". This I cannot compensate for in the comparison, not without raw access...Show more →
There have been some rigorous analysis of 7D RAW files. Most indicate very high signal to noise for the 7D.
However, no one has talked about chroma noise and actual ISO sensitivity. We'll see how true your conclusions are.
Edit: Went over your posting history and realize you have a fondness for LOW pixel count (whether it's D40 or D700). No wonder you seem so defensive about 6 MP APS-C sensors. Mind you, they have NOT always been good. Look at the awful D100/D70 6 MP sensor.
