Pages: [1] 2   Go Down

Author Topic: In 10 days we'll know what ISO400,000 looks like!  (Read 8133 times)

Petrus

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 952
Re: In 10 days we'll know what ISO400,000 looks like!
« Reply #1 on: February 25, 2014, 04:36:54 am »

ISO numbers are getting uncomfortably large, how about starting to use the good old logarithmic DIN again? 400 000 ISO would be 57 DIN, yes?

(edit: ASA changed to DIN… Trying to be too clever at this age...)
« Last Edit: February 25, 2014, 11:55:15 am by Petrus »
Logged

bdp

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 266
    • http://www.bendearnleyphotography.com
Re: In 10 days we'll know what ISO400,000 looks like!
« Reply #2 on: February 25, 2014, 05:42:17 am »

I thought ASA and ISO were the same values. Wasn't it DIN that had the small numbers? So 15DIN = 25ASA from memory, and went up one number for each third of a stop. So 50 ASA/ISO would be DIN18, 100ISO would be DIN21 etc. ISO 400,000 would be DIN57, would it?! Lost count.....
« Last Edit: February 25, 2014, 05:46:49 am by bdp »
Logged

Petrus

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 952
Re: In 10 days we'll know what ISO400,000 looks like!
« Reply #3 on: February 25, 2014, 07:39:32 am »

Dang! DIN it is !!!!!!!

too old...
Logged

Christoph C. Feldhaim

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2509
  • There is no rule! No - wait ...
Re: In 10 days we'll know what ISO400,000 looks like!
« Reply #4 on: February 25, 2014, 08:57:50 am »

ASA=100*2(DIN-21)/3

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
Re: In 10 days we'll know what ISO400,000 looks like!
« Reply #5 on: February 25, 2014, 10:57:38 am »

At a guess, the official range 100-25,600 could be fairly close to what ISO 12232 recommends as "ISO speed latitude", with 100 being about the well-saturation based minimum or "Ssat" value, and 25,600 being around the upper limit based on barely acceptable 10:1 SNR in the mid-tones, "Ssnr10". Basic photon counting says that beyond about 25,600, the mid-tone SNR level will be horribly low at full 16MP resolution, but down-sampling to about 1MP or less for publishing in the sports section of a newspaper or website could give quite usable results.  Roughly, a 16-to-1 downsampling could increase the per-pixel Ssnr10 speed by a factor of 16; with low enough read noise from the sensor, SNR=10:1 just needs about 100 photons counted per pixel (not per photosite).  In this context, note the new "small raw option".

ISO numbers are getting uncomfortably large, how about starting to use the good old logarithmic DIN again? 400 000 ISO would be 57 DIN, yes?
Yes, I like to think in terms of "stops above ISO 100", so the official range of 100-25,600 is like "0 to 8 stops" and the expanded range of 50-409,600 is "-1 to 12".  The former German DIN standard is roughly this, but in one-third stop increments, counting from a bit below ISO speed of 1.


P. S. To combine two previous comments, the exposure index scales are ISO = ASA =100*2(DIN-21)/3 and "stops above ISO100" = (DIN-21)/3.
« Last Edit: February 25, 2014, 11:14:46 am by BJL »
Logged

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
still "only" 11fps?! (Same as the Sony A6000)
« Reply #6 on: February 25, 2014, 11:12:57 am »

Another note: the maximum frame rate stays at 11fps as in the D4, due I guess to the mechanical speed limits of Nikon's mirror-flipping technology. This is "only" the same as the frame rate with subject tracking continuous AF offered by the $650 Sony A6000 with the 179 PDAF points of its 24MP "APS-C" sized sensor.  Given that high frame rate action shooting with continuous AF subject tracking is one of the major remaining advantages of SLRs over mirrorless system cameras, it would be fun to see a test of Nikon D4s vs Canon 1D-X vs Sony a6000 on fast moving subjects.  More so when Sony gets it latest on-sensor PDAF technology into a 35mm format FE mount body.
« Last Edit: February 25, 2014, 11:15:47 am by BJL »
Logged

Vladimirovich

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1311
Re: still "only" 11fps?! (Same as the Sony A6000)
« Reply #7 on: February 25, 2014, 05:50:58 pm »

More so when Sony gets it latest on-sensor PDAF technology into a 35mm format FE mount body.
it is there already - the only difference is the coverage... a6000 pdaf sensels (or rather CFA filters modified for PDAF purposes) cover more space in A6000 than they do in NEX6 or A7... otherwise there is no difference.
Logged

Eddy M

  • Newbie
  • *
  • Offline Offline
  • Posts: 7
Re: In 10 days we'll know what ISO400,000 looks like!
« Reply #8 on: February 25, 2014, 10:52:24 pm »

It looks like ISO 51200 pushed 3 stops further, hehehehe.
Logged

bjanes

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3387
Re: In 10 days we'll know what ISO400,000 looks like!
« Reply #9 on: February 26, 2014, 09:32:04 am »

At a guess, the official range 100-25,600 could be fairly close to what ISO 12232 recommends as "ISO speed latitude", with 100 being about the well-saturation based minimum or "Ssat" value, and 25,600 being around the upper limit based on barely acceptable 10:1 SNR in the mid-tones, "Ssnr10". Basic photon counting says that beyond about 25,600, the mid-tone SNR level will be horribly low at full 16MP resolution, but down-sampling to about 1MP or less for publishing in the sports section of a newspaper or website could give quite usable results.  Roughly, a 16-to-1 downsampling could increase the per-pixel Ssnr10 speed by a factor of 16; with low enough read noise from the sensor, SNR=10:1 just needs about 100 photons counted per pixel (not per photosite).  In this context, note the new "small raw option".
Yes, I like to think in terms of "stops above ISO 100", so the official range of 100-25,600 is like "0 to 8 stops" and the expanded range of 50-409,600 is "-1 to 12".  The former German DIN standard is roughly this, but in one-third stop increments, counting from a bit below ISO speed of 1.

To carry out the analysis further, I assumed the full well of the D4s to be 100,000 e-, which is a generous guess. Each doubling of ISO reduces the electron count by a factor of 2. SNR (considering only shot noise) is the square root of the number of electrons. Values for the SNR at the maximal electron count and at the midtones (18%) are shown.

Examples of various SNRs are shown in Figure 13 of Emil's treatise.

Bill
Logged

Theodoros

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2454
Re: still "only" 11fps?! (Same as the Sony A6000)
« Reply #10 on: February 26, 2014, 09:57:25 am »

Another note: the maximum frame rate stays at 11fps as in the D4, due I guess to the mechanical speed limits of Nikon's mirror-flipping technology. This is "only" the same as the frame rate with subject tracking continuous AF offered by the $650 Sony A6000 with the 179 PDAF points of its 24MP "APS-C" sized sensor.  Given that high frame rate action shooting with continuous AF subject tracking is one of the major remaining advantages of SLRs over mirrorless system cameras, it would be fun to see a test of Nikon D4s vs Canon 1D-X vs Sony a6000 on fast moving subjects.  More so when Sony gets it latest on-sensor PDAF technology into a 35mm format FE mount body.
I very much doubt that the possible customers of D4 or 1DX (photojournalists, sports/action photographers, war/crisis photo reporters) would consider …Sony A6000 for an alternative. In fact, I don't doubt it… I'm pretty sure they won't…  ??? Neither those who would choose an "A6k" (street, all around, travel, ….my aunt (!)) would ever choose a D4 or 1DX…. the rich ones may consider a DF (and the even richer a Leica M), but D4 and the like, …no way! :P ;D
Logged

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
Re: still "only" 11fps?! (Same as the Sony A6000)
« Reply #11 on: February 26, 2014, 11:09:04 am »

I very much doubt that the possible customers of D4 or 1DX (photojournalists, sports/action photographers, war/crisis photo reporters) would consider …Sony A6000 for an alternative.
Of course not; that's why I used the word "fun":
Quote
it would be fun to see a test of Nikon D4s vs Canon 1D-X vs Sony a6000 on fast moving subjects.
This would just be a look at how one aspect of technology that formerly contributed to the great advantage to high end DSLRs is being reduced by newer technologies; it does not come close to nullifying all the advantages in low-light/high-speed photography that come from a larger sensor format, fast professional lenses for that format, Nikon's (and Canon's) expertise in high speed operation, etc.  It might however hint at the progress that could be coming when Nikon or Canon merge the best of on-sensor AF technology with their other assets to produce a high-end mirrorless system for integrated stills and video PJ/sports photography.
Logged

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
Re: In 10 days we'll know what ISO400,000 looks like!
« Reply #12 on: February 26, 2014, 11:24:21 am »

To carry out the analysis further, I assumed the full well of the D4s to be 100,000 e-, which is a generous guess. Each doubling of ISO reduces the electron count by a factor of 2. SNR (considering only shot noise) is the square root of the number of electrons. Values for the SNR at the maximal electron count and at the midtones (18%) are shown.
Thanks Bill; that 100,000e- is a reasonable if optimistic estimate, being about 2,000e- per square micron, whereas the measurements I have seen top out at about 1600. Also, thanks for linking to Emil's illustration, which suggests that at low as SNR=8 can be tolerable for reportage in extreme lighting situations.

One slight change though: if the exposure index is calibrated as the ISO defines, the mid-tone (image from an 18% gray card) is placed at 12.75% of maximum, not 18%, to allow room for the average reflectivity of a scene being a bit less than 18%. So overall, your values might be optimistic by up to one stop, but good enough to get the main idea: full resolution images might be tolerable for some news/sports reporting at up to EI=12,800 or even 25,600, but beyond that the files will be mainly good for use down-sampled from 16MP (or displayed at very high PPI for massive dithering).  No criticism of that from me, as that is probably quite a useful and common usage with such a camera.  For downsampling, each halving of linear resolution about quadruples the photon count per pixel, and so quadruples the EI at which a given SNR is achieved. So reducing by  a linear factor of 16, to about 1MP, could make EI=409,600 tolerable.
« Last Edit: February 26, 2014, 02:44:12 pm by BJL »
Logged

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
Re: still "only" 11fps?! (Same as the Sony A6000)
« Reply #13 on: February 26, 2014, 11:35:45 am »

it is there already - the only difference is the coverage... a6000 pdaf sensels (or rather CFA filters modified for PDAF purposes) cover more space in A6000 than they do in NEX6 or A7... otherwise there is no difference.
Yes, maybe the AF firmware technology is the same, and is ready for 11fps. But for whatever reason the A7 only goes to 6fps (the same as in the A99) whereas the A6000 tracks AF at 11fps.
Could it be that not enough of Sony's full 35mm format lenses can be "slewed" fast enough to keep up with the signals from the new AF system?
Or that Sony has not yet upgraded its focal plane shutter designs in the larger format to handle the higher frame rate, while it now has with the smaller, lighter shutter of the A6000?
« Last Edit: February 26, 2014, 12:25:33 pm by BJL »
Logged

MrSmith27

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 79
Re: In 10 days we'll know what ISO400,000 looks like!
« Reply #14 on: February 27, 2014, 07:07:44 am »

ISO numbers are getting uncomfortably large, how about starting to use the good old logarithmic DIN again? 400 000 ISO would be 57 DIN, yes?

(edit: ASA changed to DIN… Trying to be too clever at this age...)

ISO numbers are bullsh*t just as Richter scale numbers: I'm willing to bet that 9 out of 10 people think that ISO 12000 is "a thousand times more light sensitive than ISO 100" and I know for a fact that 10 out of 10 people think that a Richter 10 earthquake is "twice as bad" as a Richter 5 earthquake.

Of course camera makers won't change the metric because it makes them look good: "Oh my god! It's ISO half a million! My old camera could only do 1600 and... now... it's A THOUSAND TIMES BETTER!!!"




Logged

Petrus

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 952
Re: In 10 days we'll know what ISO400,000 looks like!
« Reply #15 on: February 27, 2014, 07:22:36 am »

ISO numbers are bullsh*t just as Richter scale numbers: I'm willing to bet that 9 out of 10 people think that ISO 12000 is "a thousand times more light sensitive than ISO 100" and I know for a fact that 10 out of 10 people think that a Richter 10 earthquake is "twice as bad" as a Richter 5 earthquake.

Of course camera makers won't change the metric because it makes them look good: "Oh my god! It's ISO half a million! My old camera could only do 1600 and... now... it's A THOUSAND TIMES BETTER!!!"

Well, one million ISO is thousands times more sensitive than 1000 ISO, as ISO uses a linear scale. Twice the sensitivity, twice the number, like from 200 to 400 ISO.

Richeter scale is logarithmic just like DIN sensitivity scale, except that DIN uses deciBels where 3 equals double (0.3 in straight logarithmic units), i.e. the numbers are multiplied by 10 (deci- in metric usage means one tenth). With Richter scale 0.3 is double and 1 is ten times the severity of the quake (basically speaking).
Logged

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
ISO Exposure Index setting has no necessary connection to "ISO speed"
« Reply #16 on: February 27, 2014, 10:20:49 am »

Well, one million ISO is thousands times more sensitive than 1000 ISO, as ISO uses a linear scale.
Using an ISO Exposure Index setting on a camera of one million instead of one thousand simply means that:
  • In auto-exposure modes, it will choose a shutter speed one thousand times faster (or other equivalent combination of shutter speed and f-stop) so that the sensor receives one thousand time less light, and
  • in default conversions to JPEG, a given level of light received by the sensor will be sent to an output brightness level one thousand times higher, to compensate for the above reduction in how much exposure the sensor gets, and so produce appropriate luminance levels in the final output.
This choice of exposure index is unrelated to whether the sensor is more sensitive in the sense of detecting light better, by counting a larger fraction of incoming photons and/or having less dark noise. Of course one expects and hopes that camera makers only increase the maximum ISO EI setting offered in response to improvements in sensor performance, but there is no direct or guaranteed connection.

Sorry to repeat myself, but the "ISO" setting on a camera is a measure of Exposure Index, which is not at all the same thing as the ISO speed of a sensor (or of film), as measured for example by the noise-based ISO standard Snoise40, which requires a SNR of at least 40:1 at mid tones (e.g. on gray card test shot).  The ISO speed [Snoise40] of a camera with 7.2 micron pixel pitch and Bayer CFA like the D4s cannot be more than about 400, due to the basic quantum nature of light (Poisson distributed photon shot noise). The more generous measure Snoise10 (SNR of 10:1 at midtones) can be at most about 6400.

For more details on the persistent confusion of the multiple photographic measures with the letters "ISO" in their name, see my rant at http://www.luminous-landscape.com/forum/index.php?topic=87439.msg711760#msg711760
« Last Edit: February 27, 2014, 10:43:06 am by BJL »
Logged

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
D4s samples at all the way up to ISO 409,600
« Reply #17 on: February 27, 2014, 11:41:06 am »

There are many D4s samples at http://www.clubsnap.com/forums/showthread.php?t=1375310 and some comparisons to D4 images taken side-by-side.

As expected, the 100% pixel view crops fall apart very obviously above about EI=6,400, but "web-sized" JPEGs look better, and fairly good up to EI=25,600 which is as far as the full-image samples go.
« Last Edit: February 27, 2014, 11:44:31 am by BJL »
Logged

LKaven

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1060
Re: In 10 days we'll know what ISO400,000 looks like!
« Reply #18 on: February 27, 2014, 11:41:31 am »

To carry out the analysis further, I assumed the full well of the D4s to be 100,000 e-, which is a generous guess.

The D4 has a full-well capacity of 120k e-, according to Marianne Oelund's (and others') measurements.  I'd assume that D4s is at least that.

One of the ways they achieved this was by using a multiplexed readout.  Two adjacent columns share a single readout, thereby reducing the amount of active electronics at the photosite.
« Last Edit: February 27, 2014, 11:44:10 am by LKaven »
Logged

LKaven

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1060
Re: D4s samples at all the way up to ISO 409,600
« Reply #19 on: February 27, 2014, 11:49:20 am »

There are many D4s samples at http://www.clubsnap.com/forums/showthread.php?t=1375310 and some comparisons to D4 images taken side-by-side.

The problem with those samples is that it appears that the D4s samples have been heavily stepped-on with noise reduction while the D4 samples haven't.  The D4s samples all have zero chroma noise, and a lot of smearing, all evidence of heavy noise filtering.
Pages: [1] 2   Go Up