falconeyes: This will become the camera with best image quality ever made up to today!
(in terms of DxOMark, it should score around 105 Overall Score and 50MP, even though the jump from 35mm is small (0.79 crop)).
I hope that Ricoh releases the marketing money needed to make this a bold statement for Pentax being overall leader in IQ. The brand needs a bit of hype to make a come back, esp. in the US.
Yes, Credo 80 is CCD, no chance to lead the scores. IQ280 is the same sensor, so chances are high Pentax deploys their famous magic to lead the score by a point or two again. Ask Nikon and Sony ;)
The 645Z isn't a good low light choice. Performance at high iso will be good (ISO 12800 will be like ISO 8000 on fullframe). But not good enough to compensate for a lack of fast glass.
The 645Z will be outperformed in low light by full frame with lenses F/2 or faster.
tinetz: Some nice additional info:http://www.camera-pentax.jp/645z/en/
Nice indeed. However, I cannot like the exxagerations being made, such as: "PENTAX 645Z realizes high-definition and stereoscopic depiction far exceeding cameras equipped with 35 mm-format sensor."
In reality, 645Z with a f/2.8 performs exactly as a full frame with a f/2.2 lens and is surpassed by FF with an f/1.8 lens (except maybe for up to 17% better spacial resolution). After all, the 645Z has a much smaller sensor than the 645 format would suggest.
One should NOT call this "far exceeding" :(
Bold marketing style once created in the US really is everywhere now, including modest Japan.
This will become the camera with best image quality ever made up to today!
1/50 at f/2.8 and ISO 409600 is LV -3 (light value). That's about full moon light. Some AF systems still work, most won't. Human eye can still see well, people are casting a visible shadow, it is NOT completely dark.
People are amazed. However, this is something EVERY recent Nikon or Sony full frame camera with Sony sensor is capable of in still. Just push ISO from 6400 by 6 EV in post, apply heavy NR and downscale to video resolution.
The thing new is the missing line skipping which doesn't loose you anymore 2 EV or so in video mode. And the gradation options help overcome the limitations of the missing raw file.
EOSHD: You can't judge noise levels on YouTube. Compression removes it. ISO 1600 looks worse than 25,600. There's also the matter of rolling shutter on this camera, it isn't pretty! http://www.eoshd.com/content/12631/sony-a7s-rolling-shutter-test
Wrt rolling shutter caused by pixels or surface ...
I think it is a resolution effect, not chip surface. E.g., the A7s can output 720p @ 120fps IIRC which is very fast. Therefore, applying line skipping the A7s should be able to reduce the rolling shutter effect. Just not in 4k though.
Overall, the A7s chip appears to have faster readout than its smaller brothers, not slowlier. Or at least on par.
Moreover, from a point of physics, a signal needs less than 1 ns to travel from center to border of a full frame chip, at more than 30000 ns for a 30fps readout this should really be an irrelevant delay. Moreover, full frame chips have more chip real estate enabling wider on chip and off chip busses and better heat dissipation. If anything, larger chips should be faster. They are often slowlier because they are expensive enough and would become even more expensive if produced in the same state-of-the-art process. Economy slows them down, not technology.
Hungary passes this new bill in accordance with a world-wide consensus that NSA and other surveilance bodies should be the only ones keeping photos or videos for memories ...
Thos Olympus patent seems to be useful only in scenarios like the firework, where you can afford fuzzy regions of under- or overexposure.
True progress (I mentioned it many times before) will come from digital sensors: sensors which continously use, e.g., column-parallel ADC, to read out lines and add up their values in digital pixel registers. This will allow for digitally implemented well capacities and native iso level as low as ISO 1, with accompanying huge input dynamic range.
In the future. Today, the required memory on sensor-chip is too expensive for the amount of pixels we got.
The matter already is regulated in many areas of the world.
Where I live (Bavaria), you need an "Aufstiegsgenehmigung" which is given for a small fee to commercial photographers (valid one year).
Private persons without a plan to commercially exploit their model flights can pursue thier hobby w/o this permission.
ok, this is interesting, #rear-lens-modules/smartphone jumping from 1 to 2.
The real party will start as soon as we see 3x3 lens arrays or 9 cameras. As this will be the barrier where the smartphone crop factor disadvantage (from the thin housings) will be egalized and even big cameras will be challenged to provide a reason to exist.
peevee1: If, instead of getting 24 fps video in 5:1 and then basing their tests on that, they got just one frame of 1 fps video (and RED is capable of that) in 1:1, the results MIGHT have some validity. As it is, they measured the results of both temporal (interframe) and 5:1 intraframe noise reduction.Just shows that DxO methods do not detect and take into account loss of color resolution coming with noise reduction (and it would mean that their results for X-trans suffering from the same problem would also be too high).
How I am tired from all the incompetent "testers" who really should be McDonalds burger flippers...
Between testers and burger flippers, a majority of jobs can be found ;)
falconeyes: This is both interesting and isn't.
It is interesting because the RED can capture 19 MP images at 85 fps, quite a beast. And of course, at 1/42th s, you just combine two frames into one, thereby halving effective base ISO for slow shutter speeds. For such combined double frames, DxO's measured performance is disappointing actually.
It isn't interesting because for stills with a focal plane or global shutter or flash photography, you can't use this technique without embedding a digital frame memory on the sensor. And this is nowhere near in time.
DxO knows it, which is why they don't list the RED in their rankings.
DPR's headline is WRONG: D800E still is #1 in the DxO ranking, it hasn't changed.
I didn't talk about the HDRx mode, just ordinary temporal noise reduction. You may find the discussion in the DxO forum.
This is both interesting and isn't.
Am I missing something?
Nikon acknowledges the issue but shows no signs of sincere apopolgy?
Because "the presence of dust particles cannot be completely avoided"?
Can anybody beat them in arrogance?
In comparison, we can only now appreciate how Pentax apologized in an almost identical situation:-> http://support.pentaximaging.com/node/1214
D1N0: Won't this make all subjects really short and fat?
It is interesting that even with Chinese workers, this seems to make a factor 2 cost difference. Interesting as we aren't normally offered such a direct 1:1 comparison.
Maybe, it is wether assembly takes place in a clean room or not ... Probably, if I think it through. And probably, they have to rent the clean room for batches of special edition lenses. My guess.
zimarrio: Does anamorphic lenses still make sense in digital cameras?Isn't the image going to lose quality after up-sampling the pixels when converting to 2.35:1?
zimarrio, I think mgrum said the opposite to what you understood.
Scaling/upsizing, then downsampling provides better quality than cropping. Less noise and better resolution. Assuming the anamorphic lens is good enough.
Moreover, just assume that folks at personal-view.com forum and eoshd.com know what they have been talking about.
falconeyes: RGWB bayer pattern (Sony), X-trans Bayer pattern (Fuji), SuperCCD (Fuji), Foveon (Sigma) and now Foveon Quattro (Sigma) -- all are just minor variations of the de-facto standard RGGB bayer pattern sensor.
All variations (which include variations in spectral width of filter colours) are just a mathematically described trade-off between luminosity and color resolution and noise. With the standard RGGB bayer pattern being already very close to an optimum (taking the reduced color resolution of the human eye into account).
The problem is:
Who on earth makes the vendors believe that their variations of the standard RGGB bayer pattern will be paralleled by software makers in their raw converter algorithms which are highly tuned for the standard pattern?
Of course, they won't and rightly so. And after Sony abandoned their approach, Fuji and Sigma are now alone. No way their proprietary color schemes will survive. Just like SuperCCD didn't.
Thanks for all the replies but I meant what I said. I may have a more abstract view on the topic but that's a good thing.
E.g., you may think that Foveon is fundamentally different. It is not. You can achieve a similiar characteristics with very wide spectral color filters where each sensels would possibly capture every color, but with different probabilities. The higher layers in a Foveon sensor act as a sort of color filters for the lower layers.
As I said, it is just a mathematical optimization of luminance vs. chrominance signals. Vendors highlight the differences for marketing reasons. But we all are paying the prize as algorithms aren't as sophisticated as they ought to be.
Btw, I can show that X-trans has less probability for color moiré near the Nyquist frequency at the expense of higher probability for color moiré outside the Nyquist frequency. It is a zero sum game.
RGWB bayer pattern (Sony), X-trans Bayer pattern (Fuji), SuperCCD (Fuji), Foveon (Sigma) and now Foveon Quattro (Sigma) -- all are just minor variations of the de-facto standard RGGB bayer pattern sensor.
A good scientific paper about DFD and an evaluation showing that phase detect and DFD should have similiar performance can be found here:
It is 14 years old ;)