On 4/11/2021 2:21 am, Whisky-dave wrote:
> On Tuesday, 2 November 2021 at 14:48:57 UTC, Savageduck wrote:
>> On Nov 2, 2021, Whisky-dave wrote
>> (in article<8ec0fee5-b272-42a7...@googlegroups.com>):
>>> On Monday, 1 November 2021 at 13:53:31 UTC, Incubus wrote:
>>>> On 2021-10-29, Alfred Molon<alfred...@yahoo.com> wrote:
>>>>> Am 29.10.2021 um 16:34 schrieb Incubus:
>>>>>> I think newer sensors don't have the issue with propellors, rotor blades
>>>>>> etc. I don't know how LED lighting will work.
>>>>>
>>>>> I think the Z9 doesn't have a global shutter, only a fast readout
>>>>> sensor. Fast moving things could be a problem.
>>>> I can't imagine they would release if it fast moving things are a
>>>> problem given that it is designed for fast moving things.
>>>
>>> I guess it depends on how fast a thing is actually moving as to whether it affects the final image.
>>> But it's only 120 fps my 6 yearv old iphone can do 240 fps lower resolution sure.
>>>
>>> fastest shutter speed is 1/32000 fast but not so incredable that nothing will be have motion blur.
>> The issue isn’t motion blur that is the problem with an electronic shutter.
>
> I think the problem will have the same effect but the term shutter is at fault here, which is why there is confusion in terms.

Agreed.

Probably has already been sated in this thread, but the rolling or
global 'scanning' of the CCD is a completely separate issue to whether
or not there is (also) a mechanical shutter.

geoff

On Wednesday, 3 November 2021 at 17:49:48 UTC, Alan Browne wrote:
> On 2021-11-03 12:57, Whisky-dave wrote:
> > On Wednesday, 3 November 2021 at 16:40:03 UTC, Alan Browne wrote:
> >> On 2021-11-03 12:24, Whisky-dave wrote:
> >>> On Wednesday, 3 November 2021 at 15:29:08 UTC, Alan Browne wrote:
> >>>> On 2021-11-03 09:21, Whisky-dave wrote:
> >>>>
> >>>>> My 'image' of a global shutter must be different then.
> >>>>> I assumed a global shutter would suffer from Heisenberg princible just like in star trek.
> >>>> Whatever happens in Star Trek wrt to the Heisenberg principle is likely
> >>>> (at best) a misconstrued attempt at putting "science" into space opera.
> >>>
> >>> No it's only mention regarding the Heisenberg compensator after the scientest got involved
> >> A Heisenberg compensator violates ....
> >
> > There's nothing wrong with a bit of violation between consenting adults..
> >
> >>> in keeping the next generation series more credible from a technology POV.
> >> When I was a child I knew there was 0 technology credibility in ST. Han
> >> Solo's "in 17 parsecs" line was more credible where heaps of silliness
> >> comes in (and yes, they "compensated" for that goof in the space opera
> >> "Solo".
> >
> > Hans solo wasn't in star trek and parsecs is a distanace not a speed I knew that too, and it's
> > why of the reaseans I think star wars is science fantasy and star trek is science fiction.
> > Some people can't tell the differnce between the two.
> >
> >>>>> You just can;t have such a sensor you can't sample ~30meg pixels instantaneously, presently anyway so I doubt Nikon has achived it
> >>>>> even with the fastest proccessor we have today.
> >>>> An array can theoretically be devised that samples all 30Mpixels
> >>>> simultaneously for an arbitrary shutter period.
> >>>
> >>> Theoretically is not practically though.
> >> It's quite feasible. It's just more transistors, "traces" and passive
> >> components. About 60 .. 120M more which is absolutely trivial in
> >> today's chips, esp, a camera sensor.
> >
> > But that won't do the job, the new macbooks M1 Max processor has 57 billion transistors
> > but it won't be able to sample a sensor fast enough for ever pixel to have the same timestamp.
> This is NOT a CPU issue. It is a devoted device sampling gate issue.

And what does that actually mean in real terms it's something that can sample
a 'signal' from a pixel and what is that signal ?

Is it a hello world I'm here ?

> As long as you can get the signal to where it is needed the sample is
> started (and ended when the sampling signal goes to the opposite state).

So what is this magic signal ?

>
> A strobe (sampling signal) is akin to the processor timing signals all
> over a CPU chip. So while they won't be perfectly simultaneous, they
> will be at close to 0 lag pixel to pixel. The device does not "address"
> each pixel, each pixel receives the sample signal simultaneously (with
> minor variation due to propagation - sub ns level).

But this magic device would need at least 30 million inputs one for each pixel.

> >
> >>> Why use a shutter anyway ?
> >>>
> >>> Better to sample each pixel for a nano second why have a shutter at all?
> >> Depends on the shutter period. So even if the inter pixel sampling
> >> interval is 1 ns, the exposure still needs to be much longer for a
> >> viable (low noise) period.
> >
> > Which is partially while you get a blur effect.
> Not at all. Rolling shutter is due to the CPU sampling one row at a
> time; in turn each row being off loaded 1 pixel at a time.

we had similar with a slit used for a shutter now since the 70s, well when I bought my first SLR
proabbbly longer. In fact the fiorst plate camera had this effect too, but it wasn;t noticable with exposures in miniutes or hours.

A camera with a vertical plane shutter had a different effect from a horizonal plane shutter.

>
> What I'm describing is trigger the senor to sample all pixels at once
> (frozen for the same exposure period), and then offloading the entire image.

but how do you do this is where the problem starts, especailly at a high enough quality.
This si why cameras are limited by frame rates.
Not sure what the best is currently but 8k at 120 fps isn't that amazing.

Now 30MP at 10k fps would actually be impressive, but why isn't it done ?

> >>>> It would require that
> >>>> all pixel sites be triggered by a single sampling signal ("strobe")
> >>>
> >>> that signal would take a difernt time to reach each pixel, although you could possibley compensate
> >>> in a similar way to the Heisenberg compensator in star trek.
> >>>
> >>>> (There may be a trivial propagation delay in the nano second or less
> >>>> scale of the sampling signal).
> >>>
> >>> Nano second is no longer that trival as light moves a whole foot every nano second electric current somewhat slower.
> >> In an electronic circuit the propagation is somewhat slower than that,
> >> but devised correctly, you would get the sample trigger everywhere
> >> needed with negligible delay wrt the shutter period (which is in the
> >> many microseconds and slower domain).
> >
> > But not yet.
> Of course "yet". It's just more costly to do.

Yes well outside the budget of people and most companies so little point as yet in making such a camera.

> >>> Trival but you'll need something similar to rotating mirrors to achieve it as did the ultra FPS camera used for photographing a light beam 'traveling'
> >> You're way out there ...
> >
> > if it;'s so easy to do with justa few million more transistors they;d have done it.
> $

Yep same old story I'd like a lamborghini sian fkp 37 if it wasn't for the $
they'd need to make more than just 63 of them too, so sold out so I can't have one.

> >>>> This is independent of the time to offload the data which would define
> >>>> the frame rate (or specifically the inter-frame delay). Thus
> >>>> independent of the processor speed.
> >>>
> >>> The clock speed of the processor normally dictates the speed.
> >> Capturing the image is completely independent of processor speed.
> >>
> >> Processor speed goes to offloading, displaying, storing the image after
> >> the fact.
> >
> > but you need the rest of the circuit to read all teh pixels at exactly the same time.
> > File could do that.
> A DMA transfer of 30M pixels to a DMA array of some number would take
> little time.

But you need far more than just 30M, that wouldn't even give you a monochrome image just B&W.
Just is the pixel on or off.

> >
> >>> That why faster processors are used, we are pretty close to maxium presently at ~5Ghz
> >>> the way we increase so called speed is by adding more cores.
> >> Camera processors have no need to be very fast at all. Certainly not up
> >> in the 5GHz range. Because: battery life.
> >
> > They;d need to be fast , processor in camera are getting faster.
> Again, image sampling ≠ image transfer / storage. It is an independent
> of the CPU operation.

No it's not in the real world.
But what is image sampling then ?

> --
> "...there are many humorous things in this world; among them the white
> man's notion that he is less savage than the other savages."
> -Samuel Clemens

On 2021-11-04 11:07, Whisky-dave wrote:
> On Wednesday, 3 November 2021 at 17:49:48 UTC, Alan Browne wrote:
>> On 2021-11-03 12:57, Whisky-dave wrote:
>>> On Wednesday, 3 November 2021 at 16:40:03 UTC, Alan Browne wrote:
>>>> On 2021-11-03 12:24, Whisky-dave wrote:
>>>>> On Wednesday, 3 November 2021 at 15:29:08 UTC, Alan Browne wrote:
>>>>>> On 2021-11-03 09:21, Whisky-dave wrote:
>>>>>>
>>>>>>> My 'image' of a global shutter must be different then.
>>>>>>> I assumed a global shutter would suffer from Heisenberg princible just like in star trek.
>>>>>> Whatever happens in Star Trek wrt to the Heisenberg principle is likely
>>>>>> (at best) a misconstrued attempt at putting "science" into space opera.
>>>>>
>>>>> No it's only mention regarding the Heisenberg compensator after the scientest got involved
>>>> A Heisenberg compensator violates ....
>>>
>>> There's nothing wrong with a bit of violation between consenting adults.
>>>
>>>>> in keeping the next generation series more credible from a technology POV.
>>>> When I was a child I knew there was 0 technology credibility in ST. Han
>>>> Solo's "in 17 parsecs" line was more credible where heaps of silliness
>>>> comes in (and yes, they "compensated" for that goof in the space opera
>>>> "Solo".
>>>
>>> Hans solo wasn't in star trek and parsecs is a distanace not a speed I knew that too, and it's
>>> why of the reaseans I think star wars is science fantasy and star trek is science fiction.
>>> Some people can't tell the differnce between the two.
>>>
>>>>>>> You just can;t have such a sensor you can't sample ~30meg pixels instantaneously, presently anyway so I doubt Nikon has achived it
>>>>>>> even with the fastest proccessor we have today.
>>>>>> An array can theoretically be devised that samples all 30Mpixels
>>>>>> simultaneously for an arbitrary shutter period.
>>>>>
>>>>> Theoretically is not practically though.
>>>> It's quite feasible. It's just more transistors, "traces" and passive
>>>> components. About 60 .. 120M more which is absolutely trivial in
>>>> today's chips, esp, a camera sensor.
>>>
>>> But that won't do the job, the new macbooks M1 Max processor has 57 billion transistors
>>> but it won't be able to sample a sensor fast enough for ever pixel to have the same timestamp.
>> This is NOT a CPU issue. It is a devoted device sampling gate issue.
>
> And what does that actually mean in real terms it's something that can sample
> a 'signal' from a pixel and what is that signal ?

The CPU ( or some other arbitrary origin signal ) tells the device
(sensor array to begin sampling. It does. All pixels at once (in this
one shot version).

So each sensor "well" is cleared and begins accumulating charge from
that moment until the trigger says "stop" (change of state). So all
wells accumulate over the same period of time.

When the period ends, then the entire array is read by the CPU. This
takes non-zero time. But the information at each site is from the same
period of time. No rolling shutter.
>
>> As long as you can get the signal to where it is needed the sample is
>> started (and ended when the sampling signal goes to the opposite state).
>
> So what is this magic signal ?

Trigger, strobe, whatever name you want. No magic.

>
>>
>> A strobe (sampling signal) is akin to the processor timing signals all
>> over a CPU chip. So while they won't be perfectly simultaneous, they
>> will be at close to 0 lag pixel to pixel. The device does not "address"
>> each pixel, each pixel receives the sample signal simultaneously (with
>> minor variation due to propagation - sub ns level).
>
> But this magic device would need at least 30 million inputs one for each pixel.

No magic.

>> What I'm describing is trigger the senor to sample all pixels at once
>> (frozen for the same exposure period), and then offloading the entire image.
>
> but how do you do this is where the problem starts, especailly at a high enough quality.

It's just a matter of a lot of traces on the silicon. Lots. And each
pixel well would need 2 .. 4 transistors to take in the signal, clear
the well, and wait until the signal clears.

Added complexity. Not "magic".

> This si why cameras are limited by frame rates.
> Not sure what the best is currently but 8k at 120 fps isn't that amazing.
>
> Now 30MP at 10k fps would actually be impressive, but why isn't it done ?

That makes for very short exposures per frame. Aka: high noise.

Another issue : it's one thing for the sensor to capture the image in
situ. Another thing to read it off.

>>> But not yet.
>> Of course "yet". It's just more costly to do.
>
> Yes well outside the budget of people and most companies so little point as yet in making such a camera.

Markets are for testing.

>
>
>>>>> Trival but you'll need something similar to rotating mirrors to achieve it as did the ultra FPS camera used for photographing a light beam 'traveling'
>>>> You're way out there ...
>>>
>>> if it;'s so easy to do with justa few million more transistors they;d have done it.
>> $
>
> Yep same old story I'd like a lamborghini sian fkp 37 if it wasn't for the $
> they'd need to make more than just 63 of them too, so sold out so I can't have one.

They are crapily built cars that devalue very quickly (for the most
part) and that are extremely costly to operate and maintain. Only a
rare model will climb in value. Esp. if you don't actually use it.
>> A DMA transfer of 30M pixels to a DMA array of some number would take
>> little time.
>
> But you need far more than just 30M, that wouldn't even give you a monochrome image just B&W.
> Just is the pixel on or off.

RGB arrays are not a novel concept ...

>>>>> That why faster processors are used, we are pretty close to maxium presently at ~5Ghz
>>>>> the way we increase so called speed is by adding more cores.
>>>> Camera processors have no need to be very fast at all. Certainly not up
>>>> in the 5GHz range. Because: battery life.
>>>
>>> They;d need to be fast , processor in camera are getting faster.
>> Again, image sampling ≠ image transfer / storage. It is an independent
>> of the CPU operation.
>
> No it's not in the real world.

There are thousands of sensor types that operate independent of the
system CPU. The CPU might set up the sampling, gain, etc., but other
than triggering a sample period or reading data, setting parameters,
they don't need to be part of the actual sampling.

> But what is image sampling then ?

What happens on the image sensor chip independently of the CPU.

Done with this.

--
"...there are many humorous things in this world; among them the white
man's notion that he is less savage than the other savages."
-Samuel Clemens

On Thursday, 4 November 2021 at 21:30:20 UTC, Alan Browne wrote:
> On 2021-11-04 11:07, Whisky-dave wrote:
> > On Wednesday, 3 November 2021 at 17:49:48 UTC, Alan Browne wrote:
> >> On 2021-11-03 12:57, Whisky-dave wrote:
> >>> On Wednesday, 3 November 2021 at 16:40:03 UTC, Alan Browne wrote:
> >>>> On 2021-11-03 12:24, Whisky-dave wrote:
> >>>>> On Wednesday, 3 November 2021 at 15:29:08 UTC, Alan Browne wrote:
> >>>>>> On 2021-11-03 09:21, Whisky-dave wrote:
> >>>>>>
> >>>>>>> My 'image' of a global shutter must be different then.
> >>>>>>> I assumed a global shutter would suffer from Heisenberg princible just like in star trek.
> >>>>>> Whatever happens in Star Trek wrt to the Heisenberg principle is likely
> >>>>>> (at best) a misconstrued attempt at putting "science" into space opera.
> >>>>>
> >>>>> No it's only mention regarding the Heisenberg compensator after the scientest got involved
> >>>> A Heisenberg compensator violates ....
> >>>
> >>> There's nothing wrong with a bit of violation between consenting adults.
> >>>
> >>>>> in keeping the next generation series more credible from a technology POV.
> >>>> When I was a child I knew there was 0 technology credibility in ST. Han
> >>>> Solo's "in 17 parsecs" line was more credible where heaps of silliness
> >>>> comes in (and yes, they "compensated" for that goof in the space opera
> >>>> "Solo".
> >>>
> >>> Hans solo wasn't in star trek and parsecs is a distanace not a speed I knew that too, and it's
> >>> why of the reaseans I think star wars is science fantasy and star trek is science fiction.
> >>> Some people can't tell the differnce between the two.
> >>>
> >>>>>>> You just can;t have such a sensor you can't sample ~30meg pixels instantaneously, presently anyway so I doubt Nikon has achived it
> >>>>>>> even with the fastest proccessor we have today.
> >>>>>> An array can theoretically be devised that samples all 30Mpixels
> >>>>>> simultaneously for an arbitrary shutter period.
> >>>>>
> >>>>> Theoretically is not practically though.
> >>>> It's quite feasible. It's just more transistors, "traces" and passive
> >>>> components. About 60 .. 120M more which is absolutely trivial in
> >>>> today's chips, esp, a camera sensor.
> >>>
> >>> But that won't do the job, the new macbooks M1 Max processor has 57 billion transistors
> >>> but it won't be able to sample a sensor fast enough for ever pixel to have the same timestamp.
> >> This is NOT a CPU issue. It is a devoted device sampling gate issue.
> >
> > And what does that actually mean in real terms it's something that can sample
> > a 'signal' from a pixel and what is that signal ?
> The CPU ( or some other arbitrary origin signal ) tells the device
> (sensor array to begin sampling. It does. All pixels at once (in this
> one shot version).
>
> So each sensor "well" is cleared and begins accumulating charge from
> that moment until the trigger says "stop" (change of state). So all
> wells accumulate over the same period of time.
>
> When the period ends, then the entire array is read by the CPU. This
> takes non-zero time. But the information at each site is from the same
> period of time. No rolling shutter.
> >
> >> As long as you can get the signal to where it is needed the sample is
> >> started (and ended when the sampling signal goes to the opposite state).
> >
> > So what is this magic signal ?
> Trigger, strobe, whatever name you want. No magic.
> >
> >>
> >> A strobe (sampling signal) is akin to the processor timing signals all
> >> over a CPU chip. So while they won't be perfectly simultaneous, they
> >> will be at close to 0 lag pixel to pixel. The device does not "address"
> >> each pixel, each pixel receives the sample signal simultaneously (with
> >> minor variation due to propagation - sub ns level).
> >
> > But this magic device would need at least 30 million inputs one for each pixel.
> No magic.
> >> What I'm describing is trigger the senor to sample all pixels at once
> >> (frozen for the same exposure period), and then offloading the entire image.
> >
> > but how do you do this is where the problem starts, especailly at a high enough quality.
> It's just a matter of a lot of traces on the silicon. Lots. And each
> pixel well would need 2 .. 4 transistors to take in the signal, clear
> the well, and wait until the signal clears.
>
> Added complexity. Not "magic".
> > This si why cameras are limited by frame rates.
> > Not sure what the best is currently but 8k at 120 fps isn't that amazing.
> >
> > Now 30MP at 10k fps would actually be impressive, but why isn't it done ?
> That makes for very short exposures per frame. Aka: high noise.
>
> Another issue : it's one thing for the sensor to capture the image in
> situ. Another thing to read it off.
> >>> But not yet.
> >> Of course "yet". It's just more costly to do.
> >
> > Yes well outside the budget of people and most companies so little point as yet in making such a camera.
> Markets are for testing.
> >
> >
> >>>>> Trival but you'll need something similar to rotating mirrors to achieve it as did the ultra FPS camera used for photographing a light beam 'traveling'
> >>>> You're way out there ...
> >>>
> >>> if it;'s so easy to do with justa few million more transistors they;d have done it.
> >> $
> >
> > Yep same old story I'd like a lamborghini sian fkp 37 if it wasn't for the $
> > they'd need to make more than just 63 of them too, so sold out so I can't have one.
> They are crapily built cars that devalue very quickly (for the most
> part) and that are extremely costly to operate and maintain. Only a
> rare model will climb in value. Esp. if you don't actually use it.
> >> A DMA transfer of 30M pixels to a DMA array of some number would take
> >> little time.
> >
> > But you need far more than just 30M, that wouldn't even give you a monochrome image just B&W.
> > Just is the pixel on or off.
> RGB arrays are not a novel concept ...
> >>>>> That why faster processors are used, we are pretty close to maxium presently at ~5Ghz
> >>>>> the way we increase so called speed is by adding more cores.
> >>>> Camera processors have no need to be very fast at all. Certainly not up
> >>>> in the 5GHz range. Because: battery life.
> >>>
> >>> They;d need to be fast , processor in camera are getting faster.
> >> Again, image sampling ≠ image transfer / storage. It is an independent
> >> of the CPU operation.
> >
> > No it's not in the real world.
> There are thousands of sensor types that operate independent of the
> system CPU. The CPU might set up the sampling, gain, etc., but other
> than triggering a sample period or reading data, setting parameters,
> they don't need to be part of the actual sampling.
> > But what is image sampling then ?
> What happens on the image sensor chip independently of the CPU.
>
>
> Done with this.

Yeah well, it's still not a global shutter it's a rolling shutter, which doesn't sample all
the sensor data at the same time.

> --
> "...there are many humorous things in this world; among them the white
> man's notion that he is less savage than the other savages."
> -Samuel Clemens

On 2021-11-09 08:18, Whisky-dave wrote:
> On Thursday, 4 November 2021 at 21:30:20 UTC, Alan Browne wrote:

>> Done with this.
>
> Yeah well, it's still not a global shutter it's a rolling shutter, which doesn't sample all
> the sensor data at the same time.

The discussion was a hypothetical. But you knew that. Right?

Really done.

--
"...there are many humorous things in this world; among them the white
man's notion that he is less savage than the other savages."
-Samuel Clemens

On Tuesday, 9 November 2021 at 13:53:22 UTC, Alan Browne wrote:
> On 2021-11-09 08:18, Whisky-dave wrote:
> > On Thursday, 4 November 2021 at 21:30:20 UTC, Alan Browne wrote:
>
> >> Done with this.
> >
> > Yeah well, it's still not a global shutter it's a rolling shutter, which doesn't sample all
> > the sensor data at the same time.
> The discussion was a hypothetical. But you knew that. Right?
>
> Really done.

As I said not currently possible not even with the latest Nikon Z9 or whatever it's called.
It;'s just ditched a mechanical shutter, I can not use my mechanical shutter in my EOS M6 mkII.

As for hyperthetical well star trek is like that with a star trek camera hyperthetically
you could go back in time and take a photo of something that happend before phtography came about.
hyperthetically all you need is to get really close to a black hole, or exceed the speed of light,
but not sure if that's possibe
But thre's is the terminator method but you have to be naked for that to work.

hyperthetical is very close to fiction, and fantasy is even further away from current reality.

Hyperthetically sucha sensor could be deleloped but it would need a much better and father processor,
to store whatever the sensor senses, the most efficint way would be to store that digitaly,
which would require a really fast A-D converter for each pixel, otherwise it would take a significant to
before you could take another picture perhaps a second+ or with current tech.

Which is why sampling 8K of 30M sensor can only just be done fast enough
for 60/120 fps.
But you can buy camera with global shutters but tehy are limited to about 4K,
which is quite low for a photograph and they cost about £30k.

> --
> "...there are many humorous things in this world; among them the white
> man's notion that he is less savage than the other savages."
> -Samuel Clemens