Welcome, Guest. Please login or register.

Username: Password:
Pages: [1]   Go Down

Author Topic: 210609 iso invariance discussion  (Read 2632 times)

TomT

  • Administrator
  • Sr. Member
  • *****
  • Karma: +0/-0
  • Offline Offline
  • Posts: 171
    • View Profile
    • Photos of SBAU activities on FLICKR, albums latest to older
210609 iso invariance discussion
« on: July 09, 2021, 04:55:17 PM »

On 6/10/2021 6:06 PM, bkm wrote:
> Mike,
> Thanks for reminding me of the avalanche photodiode, and your nice explanation.
> Like you, its been many years since I used them (I've been retired 32 years).  They are fast and sensitive, but excess noise is a problem.  Wikipedia has a nice writeup: https://en.wikipedia.org/wiki/Avalanche_photodiode
> In a camera detector, it is the capacitance of the photodiode that converts the photoelectric electrons to a voltage (V = Q/C).  This process is often referred to as filling the well, and when the well gets full, the pixel saturates (a voltage problem).
> Let's keep dreaming!
> Bruce
> They talk about poisson
>
> On 6/10/2021 1:21 PM, Mike Chibnik wrote:
>> That along with a new development on a new type of imaging sensor which is supposed to be more sensitive than what we have today.  Evidently they were stating something similar to avalanche diode technology.  It’s something they’ve been doing with fiber optics for a long time and we had used them at Canoga Perkins back in 2000.  With normal photo diodes one photon gets converted into 0.2 to 0.9 electron .    This number is referred to the detector’s quantum efficiency.  Avalanche diodes utilize an additional mechanism were the initial electron is accelerated in an electric field by a high voltage applied to it. As a result additional electrons are generated effectively amplifying the detected optical input.  It’s been 20 years so I’m fuzzy one the gain but as I remember it was from about 20 to 1000 depending on the construction and operating voltages which were around 100 volts.  These diodes were super expensive and tricky to use.  They were also a bit noisy.  In contrast nearly all optical sensors are photovoltaic sensors which in simplistic terms uses photons to generate a current of electrons which generates a charge that becomes a voyage which is the signal that is detected.  If it took about 20 years to combine one expensive photo diode in to a chip with millions of inexpensive high performance devices then it shouldn’t be to unreasonable to assume that this will happen with avalanche photo diode technology. 
>> I think history with image sensors will repeat itself as something very similar happened to television camera tube in the early fifties when the old Iconoscopes were replaced by the more sensitive image orthicons.
>> So it might be possible in the future where people will hold up a thick version of what might look like an iPad which will have that new optical technology with sensor technology to view the heavens.
>> We can dream can’t we?
>> Mike
>>
>>> On Jun 10, 2021, at 12:17 PM, Robert C R wrote: 
>>> Thanks, Chuck...Wow!  This looks like transformative stuff when it comes to optics.  Amazing to watch how discoveries like this will affect the world of photography and telescope optics.  We do live in amazing times!
>>> Bob R.
>>>
>>> From: macpuzl
>>> To: Bruce M
>>> Cc: Jerry W Mike C; Web Master <webmaster@sbau.org>; Ron H; Tom W; Tom T; Bob R; Dick B
>>> Sent: Thu, Jun 10, 2021 12:00 pm
>>> Subject: Re: camera iso invariance
>>> Bruce (et al) -
>>> This may be interesting:
>>> https://phys.org/news/2021-06-goodbye-camera-miniaturized-optics-counterpart.html
>>> Hasta nebula - Chuck

On 6/10/2021 3:06 AM, bkm wrote:
> Jerry,
> I forgot to mention that I also use the hat trick.  I have a piece of cardboard that is flat black painted on both sides.  I hold it in front of the telescope (not touching) when I remote trip the shutter, count to 2, then lift the cardboard away.
> The "hat trick" name arises from old time photographers who used their hat to block the camera lens as they were manually releasing the shutter on the longer time exposures needed for ordinary photography of the day (iso 25 film, or less).
> Bruce
>
> On 6/10/2021 2:37 AM, bkm wrote:
>> Thanks Jerry,
>> Accidental dithering is an apt description!
>> I normally take astrophotos with the mirror locked up, use a 2-second shutter delay, and use a remote trip
>> If I average 10 or more images in Deep Sky Stacker, I don't see pattern noise, but do see non pinpoint stars due to atmospheric turbulence.  Brighter stars are bigger smeared circles.
>> Nikon went all out with the D500 shutter.  It and the mirror holder are made from carbon fiber.  The camera can take 10-pictures per second for a 200 frame interval before it slows down.  Between each frame, it does an autofocus and an exposure calculation.  Both autofocus and exposure require the mirror to be down.
>> I know that this fast frame rate is not used in astrophotography, but the very light components should minimize shutter shake, especially if I use the quiet shutter release mode.
>> My images are usually non guided.  Typically 30 second exposures, or less.
>> Thanks for the info.  It gives me a lot of food for thought.
>> Bruce
>>
>> On 6/9/2021 6:17 PM, Jerry wrote:
>>> Bruce
>>> I depended on the accidental dithering provided by the environment, until I realized it really indicated a weakness in my set up. When I went to non DSLR cameras which did not provide a mechanical shutter shock, my dithering disappeared and my images went down the tubes.
>>> I now do dithering on purpose using software designed for it, Nebulosity and PHD Guiding working together. It gives me better control of my imaging.
>>> Jerry
>>>
>>>> On Jun 9, 2021, at 5:51 PM, bkm < wrote:
>>>> Mike,
>>>> Thanks for the reminder.  I figured that the minute movement of the image between pictures (turbulence, wind, etc.) coupled with the small sensor pixel dimensions (microns), pretty much guarantees the same Bayer filtered pixels will not overlay from one exposure to the next.  When I use Deep Sky Stacker, the pattern noise generally disappears.
>>>> Regarding your other email on dynamic range, the same website has other displays that might interest you.  See: https://www.photonstophotos.net/Charts/PDR_Area_scatter.htm
>>>> You can turn off the various camera manufacturers  colored symbol by clicking on them in the legend.  I find my Nikon D500 and D5200 have very similar dynamic ranges, and are among the best.
>>>> Bruce
>>>>
>>>> On 6/9/2021 2:48 PM, Mike Chibnik wrote:
>>>>> Hi Bruce:
>>>>> The way you get around the issue with fixed pattern color noise is to use dithering.  My friend Mark said that the best way he found was to dither for every single exposure.  Otherwise he found out that there was not enough randomization.
>>>>> Mike

Hi Bruce:
I went to the website and plugged in the dynamic range page for various cameras I had/have in addition the the RA which is Canon's latest Astroimaging camera.
It appears that dynamic range increases as ISO is reduced.  However, this might result in losing faint nebulosity.  However, if you want to image colors best for objects like colorful globulars or the Orion Nebula then to me it appears a lower ISO is the way.
image.png
Mike

On Wed, Jun 9, 2021 at 11:03 AM bkm > wrote:
    Jerry and Mike,
    Thanks for your responses.  I look forward to more information you might
    find.
    Jerry,  yes, fixed pattern noise in color pictures can be a problem with
    Bayer interpolation.  I've noticed it.  When Deep Sky Stacker averages a
    number of photos, the fixed pattern noise appears to be much reduced.
    Probably caused by the images not being aligned exactly on the sensor, a
    good thing.
    Bob Richard has noticed and complained about this artifact in his
    pictures with the Atik cameras he uses.
    Bruce

    On 6/9/2021 10:43 AM, Mike Chibnik wrote:
    > Thanks Bruce
    > I was going over other websites afterwards and found links in some
    > sites where they refer to others that have differing findings.  I’m
    > led to believe that as technology changes certain aspects such as read
    > noise, dark noise, pixel size and quantum efficiency have lead to
    > differing optimum solutions.
    > Mike
 
    >> On Jun 9, 2021, at 10:10 AM, Jerry  wrote:
    >> Hi Bruce
    >> It’s an interesting article, thanks. The author is comparing dynamic
    >> range to photon noise and uses the proper physics definition of
    >> noise. Usually when astro imagers talk about noise in DSLRs they
    >> really mean fixed pattern noise which spatial nonuniformity. He does
    >> not address that aspect, at least at my first read. I’ll take a
    >> deeper look when I get some time.
    >> Jerry

    >>> On Jun 9, 2021, at 2:54 AM, bkm wrote:
    >>>  In today's telescope workshop, I related there was a website
    >>> showing digital camera iso invariance i.e., the signal to noise is
    >>> the same over a wide iso range.  See:
    >>> https://www.photonstophotos.net/Charts/PDR_Shadow.htm#Nikon%20D500
    >>> For the D500 I have, which has a Sony sensor, is iso invariant from
    >>> iso 400 to iso 102400.  This fact is refreshing to know, in that I
    >>> can shoot pictures at higher isos and faster shutter speeds and not
    >>> worry about noise.
    >>> The Canon 6D that astrophotographers like to use is not iso invariant.
    >>> Notice the vertical axis scale is logarithmic.
    >>> Even though the camera names are grayed out, you can select other
    >>> cameras to see their input-referred read noise vs. iso
    >>> Here are the plots from the above website:
    >>> <sensor nosie collage.jpg>
    >>> Enjoy,
    >>> Bruce
« Last Edit: July 09, 2021, 05:15:30 PM by TomT »
Logged
Pages: [1]   Go Up