PIX 2015

XF56 f1.2 released and specs

Started Jan 6, 2014 | Discussions thread
Shop cameras & lenses ▾
a l b e r t
Senior MemberPosts: 1,357Gear list
Re: XF56 f1.2 released and specs
In reply to malcolml1, Jan 8, 2014

malcolml1 wrote:

Ok - in the second question I actually forgot to say that I meant the same number of pixels (though I realize this would result in a lot of dead space on the sensor - it was a hypothetical question).

Now I've thought a bit more, I think there is another source of confusion....I have convinced myself again that, whilst I accept that noise levels would be different, if I took a photo with my Fuji and the 56/1.2 at ISO200 and did the same with my (now sold) 5D mk2 with the 85/1.2 at ISO 200, both in aperture priority mode, they would both select much the same shutter speed (I assumed that both Canon and Fuji agree on definition of ISO).

To you (I think) equivalence also includes achieving equivalent noise levels. Pixels are buckets for collecting photons and converting them into electron-hole pairs (as you said). Obviously a bigger bucked can collect more photons (and generate more electron hole pairs). But then the actual number of electron-hole pairs is not used to form the image - rather the signal is digitized when it is read. If both the APS and FF sensor are 14 bit, then in underlying analog signal is converted into one of 2 ** 14 levels. Further, I assume that the levels (at base ISO or amplification) are spread (non-linearly, I know) between 0 and 'full' (max electron-hole pair capacity for that pixel).

To me, a given exposure would result in of the same output level regardless of sensor format - I think this is because the supposedly common definition of ISO.

The advantage of a larger pixel size is the confidence level you have in the level you measure after conversion. More exactly (I work with statistics...), you would be confident that if you repeated the measurement with 10 different pixels they would report the same level for that actual exposure. For example, a uniform area of blue sky - with bigger pixels, you have much more confidence that adjacent pixels would measure the same level after conversion to digital. You have less confidence with smaller pixels - and in fact this lower confidence manifests itself in adjacent pixels reporting more variation in the level they measure. We see this as noise.

Even simpler example - if the read error (I think this is what it is called) is + or - 10 electron-hole pairs for both APS and FF sized pixels, and the APS pixel has a maximum capacity of 100 vs the FF pixel of 1,000 (making the numbers up for illustration), then the likelihood of the analog signal being converted to exactly the same level in the blue sky example is lower for the APS sensor (+ or - 10 out of a total number of 100 is a much bigger error than + - 10 out of 100).

The comparison of APS-C sensor and FF sensor is a complicated one. It is because the sensor performance from one generation to the next can be very different.

For example, the high ISO performance of the X-Trans CMOS I sensor is on the same level of the FF sensor in 5D2, even though the sensor is 2.25x larger.

Moving along to the sensor in 5D3, high ISO noise is much lower. So when one has to compare, the comparison needs to be made to roughly the same generation sensor.

Then there is the number of pixels on the sensor. If you have a 36MPix sensor such as the one in D800, the pixel size is the same as the pixel size of a 16MPix sensor. It is because 36 / 2.25 = 16. So the number of pixels within an APS-C crop of D800 sensor is the same of that of a X-Trans sensor.

Since pixel size is the same, and the D800 sensor is roughly the same generation as X-Trans I, the quantum efficiency of the two sensors should roughly be the same.

Yet if you examine at the pixel level, you'll see that the noise of a 36MPix D800 sensor is actually noisier than the X-Trans I even at ISO 1600. I believe having two extra green pixels in a 6x6 grid helps lower the luminance noise.

Now if you compare to the next generation 36MPix sensor in A7r, you'll see that the high ISO noise is comparable to the X-Trans CMOS II.

Another reason why full frame sensor appears to have lower noise when viewed on the screen is due to pixel binning. You scaled down a 36MPix sensor to fit in your 3MPix screen. It is going to look better than a 16MPix sensor scaled down to 3MPix, both in terms of noise and details.

So all it boils down to is how large of a print are you going to make. It is not going to make a whole lot of difference even if you print it as 16x20.

 a l b e r t's gear list:a l b e r t's gear list
Fujifilm FinePix F31fd Fujifilm FinePix X100 Fujifilm X10 Fujifilm X-E2 Nikon AF-S Micro-Nikkor 105mm f/2.8G IF-ED VR +5 more
Reply   Reply with quote   Complain
Post (hide subjects)Posted by
(unknown member)
(unknown member)
(unknown member)
(unknown member)
(unknown member)
(unknown member)
(unknown member)
(unknown member)
(unknown member)
(unknown member)
(unknown member)
(unknown member)
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark post MMy threads
Color scheme? Blue / Yellow