Andrew2468

Joined on Aug 23, 2012

Comments

Total: 7, showing: 1 – 7
On article Sony a7R III added to studio scene comparison (453 comments in total)

My 2p / 2cents worth...
The studio scene tool is more about lenses than sensors and electronics. How about some additional new tests aimed at the latter. E.g.
1. How resistant to electro-magnetic interference is the sensor/electronics. E.g. does the DR reduce if someone standing next to you is talking on their phone when you fire the shutter? And if you have to shoot near heavy electrical gear?
2. Temperature and humidity. E.g. is image quality the same shooting in the arctic as the tropics. How well does the sensor cope?
3. I think I read light hitting the sensor obliquely can be an issue. Can evenness be tested, preferably without a lens, e.g. direct light beams at the sensor in varying places and at varying angles and analyse the numbers.

Perhaps something like the above might reveal there are real-world differences between makes and models that we don't know about.

Link | Posted on Nov 19, 2017 at 16:23 UTC as 9th comment | 3 replies
In reply to:

Andrew2468: ...continuing...
Take as example the A7RII/III sensor at 7952 pixels in 35.9mm.
This gives a pixel width of 4.51E-06 m (dimensions in metres from now..)
Take a tripod leg as 1.5 long and made of aluminium.
Coefficient of expansion is 22.5E-06. (Source: https://www.engineeringtoolbox.com/linear-expansion-coefficients-d_95.html)
So a 1 degK temp change causes a change in length of 3.38E-05.
This equates to 7.48 pixels.
So the temp change for a quarter pixel is 0.033 degK.

(Ok, I've not done the trig given the tripod is a triangle, roughly, but you get the idea hopefully)

Imagine you rock up at the frozen lake for a pixel shift photo and take your tripod out of the warm car. How long do you wait before it's cooled down enough? Perhaps someone else can work that out...!

hi @Lars V, I've already suggested to DPR ppl lower down this item that they could do a stacking test to see the difference between that and PS. I don't understand your infinity comment. I'm assuming the point of the successive 1 pixel movements is that the de-bayering algorithm can make use of that knowledge and do a better job than just averaging 4 shots. But if the sensor movement is a random number of pixels for whatever reason, then surely that advantage goes out of the window, and I don't see what difference the tripod height/infinity bit makes. (The temperature thing is a quantifiable example but I reckon there are various reasons why the camera can move a thousandth of a millimetre in the time between shots)

Link | Posted on Nov 15, 2017 at 11:35 UTC

Headline: 1/25 degree K temperature change is enough to mess up pixel shift!

I'm still intrigued by the idea of keeping a camera so still that the 1 pixel internal movements are not invalidated. Some of the factors are hard to quantify, like how much does the camera move in a slight breeze, but one factor is quite easy - the change in length of tripod legs due to changing temperature. So I've done some simple maths. How much can you allow the camera to move? Being generous, let's say up to a quarter of a pixel in order not to make a mockery of the internal sensor movement. So what temperature change can occur between exposures? I reckon less than 1/25th degree kelvin. Here's my working in the post below...

Link | Posted on Nov 14, 2017 at 11:46 UTC as 37th comment

...continuing...
Take as example the A7RII/III sensor at 7952 pixels in 35.9mm.
This gives a pixel width of 4.51E-06 m (dimensions in metres from now..)
Take a tripod leg as 1.5 long and made of aluminium.
Coefficient of expansion is 22.5E-06. (Source: https://www.engineeringtoolbox.com/linear-expansion-coefficients-d_95.html)
So a 1 degK temp change causes a change in length of 3.38E-05.
This equates to 7.48 pixels.
So the temp change for a quarter pixel is 0.033 degK.

(Ok, I've not done the trig given the tripod is a triangle, roughly, but you get the idea hopefully)

Imagine you rock up at the frozen lake for a pixel shift photo and take your tripod out of the warm car. How long do you wait before it's cooled down enough? Perhaps someone else can work that out...!

Link | Posted on Nov 14, 2017 at 11:45 UTC as 38th comment | 6 replies
In reply to:

Andrew2468: @Rishi Sanyal, please could you perhaps do a further test on the A7RIII to see to what extent the improvements are simply due to "signal averaging" i.e. merging multiple images, I think this is quite interesting. Something like a) pixel shift photo, vs. b) 4 separate photos with the tripod moved very slightly each time so different sensor sites are most definitely being used, then stack and align the images appropriately. It will be interesting to see how the detail compares.

@SmilerGrogan, I've not used photoshop for a good while but I think it has a function for aligning a bunch of layers.

Link | Posted on Nov 13, 2017 at 17:38 UTC

@Rishi Sanyal, please could you perhaps do a further test on the A7RIII to see to what extent the improvements are simply due to "signal averaging" i.e. merging multiple images, I think this is quite interesting. Something like a) pixel shift photo, vs. b) 4 separate photos with the tripod moved very slightly each time so different sensor sites are most definitely being used, then stack and align the images appropriately. It will be interesting to see how the detail compares.

Link | Posted on Nov 13, 2017 at 11:33 UTC as 81st comment | 9 replies
On article Sigma 50mm F1.4 DG HSM review (592 comments in total)

Re. the bokeh - and looking at the pic of the boy hanging off the climbing frame - the point highlights, esp. to the right of the boy, appear as rings, a bit like a mirror lens. I'm no expert but if a bright point is out of focus, shouldn't the resulting blur still be brighter in the middle than the edges? People have commented on distracting straight lines. The above suggests to me the lens will tend to add a secondary line outside the original, spoiling a smooth blur.

Link | Posted on Apr 23, 2014 at 10:31 UTC as 128th comment
Total: 7, showing: 1 – 7