OFFICIAL: Canon 1D X announced

theviper wrote in post #14048296
Is there no way to reverse the phase of the noise to cancel it out?

Sadly, no.

"Noise canceling" headsets and such aren't actually canceling noise, they're canceling an external signal that is measured in realtime.

Noise is not an external signal. It is a variation in the signal. The only type of noise that can be canceled is fixed pattern noise, because the pattern can be ascertained and accounted for (doing this may be simple or complicated, depending on how the noise is generated, and may even be impossible if there isn't enough in the way of telltales to determine how the noise was generated).

Random noise can't generally be canceled, it can only be reduced at it source (e.g., by using higher quality components, a better design, etc.).

kcbrown wrote in post #14051359
Sadly, no.

"Noise canceling" headsets and such aren't actually canceling noise, they're canceling an external signal that is measured in realtime.

Noise is not an external signal. It is a variation in the signal. The only type of noise that can be canceled is fixed pattern noise, because the pattern can be ascertained and accounted for (doing this may be simple or complicated, depending on how the noise is generated, and may even be impossible if there isn't enough in the way of telltales to determine how the noise was generated).

Random noise can't generally be canceled, it can only be reduced at it source (e.g., by using higher quality components, a better design, etc.).

Well, coming from a recording studio background, there are lots of tricks used to cancel out noise with phase reversal.
You can place a microphone precisely between a set of PA speakers if a vocalist does not like wearing headphones to sing. You can blast the singer with a mono backing track through these speakers and by reversing the phase of one of the speakers, none of the backing track will end up on the vocals going down the mic. The backing track simply gets cancelled out.
Another method would be when stacking tracks and tracks of backing vocals, the spill from the singers headphones ends up going down the mic and builds up to the point it's making a racket between vocal lines. If the singer stays completely still while singing, every alternate track can be put out of phase to pretty much remove every bit of headphone spill.
Different things but the same method. Phase cancellation is phase cancellation.
All noise cancelling headphones do is mic the outside noise and feed an identical signal to each headphone of this noise but flip the phase on one side. This removes the majority of the noise.
It's all about feeding the original and a copy of it with the phase reversed at the same level and they will cancel each other out.

theviper wrote in post #14052025
Well, coming from a recording studio background, there are lots of tricks used to cancel out noise with phase reversal.
You can place a microphone precisely between a set of PA speakers if a vocalist does not like wearing headphones to sing. You can blast the singer with a mono backing track through these speakers and by reversing the phase of one of the speakers, none of the backing track will end up on the vocals going down the mic. The backing track simply gets cancelled out.
Another method would be when stacking tracks and tracks of backing vocals, the spill from the singers headphones ends up going down the mic and builds up to the point it's making a racket between vocal lines. If the singer stays completely still while singing, every alternate track can be put out of phase to pretty much remove every bit of headphone spill.
Different things but the same method. Phase cancellation is phase cancellation.
All noise cancelling headphones do is mic the outside noise and feed an identical signal to each headphone of this noise but flip the phase on one side. This removes the majority of the noise.
It's all about feeding the original and a copy of it with the phase reversed at the same level and they will cancel each other out.

Right. The problem is that the source of noise in electronics isn't something that, itself, generally acts as a signal. Instead, it is just random variation inside the device that, itself, has inputs and outputs. Semiconductors are quantum devices, so there is naturally fundamental random variation within them that affects the output, on top of the usual noise sources (e.g., thermal noise). This affects analog to digital converters, amplifiers, etc.

Phase cancellation works only if you can somehow treat the noise as a signal, but that implies that you can somehow sample the noise itself in such a way that the noise is not altered. But the nature of most random noise in electronics is that you can't sample it in a useful way, because the very act of sampling either invalidates the value you get (see, e.g., quantum tunneling) or changes the value that affects the signal.

kcbrown wrote in post #14053027
Right. The problem is that the source of noise in electronics isn't something that, itself, generally acts as a signal. Instead, it is just random variation inside the device that, itself, has inputs and outputs. Semiconductors are quantum devices, so there is naturally fundamental random variation within them that affects the output, on top of the usual noise sources (e.g., thermal noise). This affects analog to digital converters, amplifiers, etc.

Phase cancellation works only if you can somehow treat the noise as a signal, but that implies that you can somehow sample the noise itself in such a way that the noise is not altered. But the nature of most random noise in electronics is that you can't sample it in a useful way, because the very act of sampling either invalidates the value you get (see, e.g., quantum tunneling) or changes the value that affects the signal.

Yes. I understand what you are saying. If they could extract just that data from the signal, I suppose they would just eliminate it there and then
All the very best.
Vinny.

a couple of good shots w/the 400 2.8 + 1.4x & 2x TCs.

Frits van Eldik on shooting an F1 weekend with the EOS-1D X

theviper wrote in post #14052025
Well, coming from a recording studio background, there are lots of tricks used to cancel out noise with phase reversal.
You can place a microphone precisely between a set of PA speakers if a vocalist does not like wearing headphones to sing. You can blast the singer with a mono backing track through these speakers and by reversing the phase of one of the speakers, none of the backing track will end up on the vocals going down the mic. The backing track simply gets cancelled out.
Another method would be when stacking tracks and tracks of backing vocals, the spill from the singers headphones ends up going down the mic and builds up to the point it's making a racket between vocal lines. If the singer stays completely still while singing, every alternate track can be put out of phase to pretty much remove every bit of headphone spill.
Different things but the same method. Phase cancellation is phase cancellation.
All noise cancelling headphones do is mic the outside noise and feed an identical signal to each headphone of this noise but flip the phase on one side. This removes the majority of the noise.
It's all about feeding the original and a copy of it with the phase reversed at the same level and they will cancel each other out.

Techniques such as these, and others in the audio space, such as the old Dolby noise reduction, only work when you have access to the "pure" signal separated from the "noise". In your example, you have the "noise" as a separate signal, so you merely reverse the phase so it cancels itself out. Dolby has access to the "pure" signal (since the noise intended to be reduced was tape hiss) and artificially amplifies the signal in the frequency range of tape hiss on recording, and then reduces that same signal on playback. Since the tape hiss was not amplified on recording, it is reduced on playback when the originally recorded signal is reduced (simplified explanation, but that is the basic idea).

Some of the noise from a camera sensor could be reduced in similar ways IF that noise originates somewhere other than the sensor itself (e.g. amplifier-induced noise due to the very low signal levels coming off the sensor), but when the noise is random and starts out embedded with the good data, and is in the same frequency as the good data, there is not much that can be done externally that will not also affect the good data.

Stone 13 wrote in post #14057065
a couple of good shots w/the 400 2.8 + 1.4x & 2x TCs.

Frits van Eldik on shooting an F1 weekend with the EOS-1D X

Atleast from the shots in the Canon CPN network they look pretty good. The high ISO's seem to hold good detail. We'll have to see the actual images before we make a judgement, but these seem lot better than the samples posted on the Canon's Japan website.

RTPVid wrote in post #14057103
Techniques such as these, and others in the audio space, such as the old Dolby noise reduction, only work when you have access to the "pure" signal separated from the "noise". In your example, you have the "noise" as a separate signal, so you merely reverse the phase so it cancels itself out. Dolby has access to the "pure" signal (since the noise intended to be reduced was tape hiss) and artificially amplifies the signal in the frequency range of tape hiss on recording, and then reduces that same signal on playback. Since the tape hiss was not amplified on recording, it is reduced on playback when the originally recorded signal is reduced (simplified explanation, but that is the basic idea).

Some of the noise from a camera sensor could be reduced in similar ways IF that noise originates somewhere other than the sensor itself (e.g. amplifier-induced noise due to the very low signal levels coming off the sensor), but when the noise is random and starts out embedded with the good data, and is in the same frequency as the good data, there is not much that can be done externally that will not also affect the good data.

Thanks Tom. I agree. I am sure Canon have looked into all the possibilities. Like you said, if it's noise caused externally there is something they can do. I am sure someone will eventually stumble across something.
The new Formula 1 pictures that CPS have just released look good. Glad that pictures are starting to trickle out. Can't wait to get mine.

More Canon samples, including some high ISO.

http://web.canon.jp/im​aging/eosd/samples/eos​1dx/

I brought both of these shots up by almost a full stop in post as well... so at least a stop of underexposure, at these high of ISOs, and still walk away with a pretty useable shot, is very impressive.

I also applied my normal PP routine, with just a touch of noise reduction added over the top... but I'm very excited to see what this camera can do.

12800

Those images look a little flat but a heck of a lot better than most things I've seen shot at those ISO speeds.

K6AZ wrote in post #14078563
Those images look a little flat but a heck of a lot better than most things I've seen shot at those ISO speeds.

Yeah, I agree. I tried not to really alter the "punch" factor much, preferring to just get them exposed and then a base sharpening, levels, etc...

For 6400 and 12800, I am going to be very pleased with this level of quality from the 1DX. A completely useable 6400 is going to rock my world. I truly can't wait to get one of these out in the field. Coupled with a fast lens, I feel like I'll be able to suck parts of the milky way out of the sky.

I am already starting to plan re-shoots of places that I've shot, but wasn't happy with the noise results from the 5D2.

Todd Lambert wrote in post #14078648
Yeah, I agree. I tried not to really alter the "punch" factor much, preferring to just get them exposed and then a base sharpening, levels, etc...

For 6400 and 12800, I am going to be very pleased with this level of quality from the 1DX. A completely useable 6400 is going to rock my world. I truly can't wait to get one of these out in the field. Coupled with a fast lens, I feel like I'll be able to suck parts of the milky way out of the sky.

I am already starting to plan re-shoots of places that I've shot, but wasn't happy with the noise results from the 5D2.

The 1D4 was very usable at 6400, the bar for this is going to be 12800. Same goes for the 5D3. It's going to be very interesting to see exactly what the difference is between the two sensors at 12800.

K6AZ wrote in post #14078563
Those images look a little flat but a heck of a lot better than most things I've seen shot at those ISO speeds.

Just needs proper contrast and color.

HTTP response: NOT FOUND | MIME changed to 'image/png'

Does seem very usable at 12800...

I am beginning to think that the 1dx is probably the camera to go to, to get that ISO leap many of us want.

I suspect the 5d3 is not going to give a major step up from the 5d2.

But this looks like it will.

Add in the cost of a grip for the 5d3 and you might want to start pushing the budget to this anyway...