Build 15 update...

[Dj Joofa]

Granted the D3 can't do 24FPS but it can do 9FPS and do windowing to DX and 4:3 aspects ratios.

It is easier to have better noise immunity at a lower frame rate, and one is many times surprised to see that with night-time shots, or shots with low illumination. The integration time (exposure time) of a sensor may not have a direct relation with the current frame rate, but is bounded by the frame rate. You don't want your integration time to be larger than desired frame rate and hence, lowering the frame rate.

Assuming the integration time has not saturated the signal enough for the clipping to happen, the noise smoothing gets significantly better with larger integration times. For lower frame rates, hence, this naturally helps. For night-time shots, shadow details, larger integration time helps; for high-lights it is not very helpful as it would clip them.

In addition, another very important factor is the temporal filtering. In a still camera image there are no motion issues to consider. However, for a video camera it is not easy to determine that which portion of a moving image is actually temporal noise, and which is valid motion. There is always some compromise. If temporal filtering is done in camera, then the real-time delivery limits it to simple filters. If the co-efficients of the temporal filter are not properly adjusted, or they were calculated at different internal gain, and are not adaptive, if the internal gain before ADC changes, etc., then you would see noise on the images from a video camera.

 

[Evin Grant]

Nikon, along with Canon, have had years of practice on this stuff... give us a couple of months to catch up.

Jim

Don't worry jim, that wasn't a dig at the Mysterium. Neither Nikon nor Canon have a sensor that can do 24FPS so it's of little consequence if they have more aggressive noise control. Also we have no idea how that noise, however pleasing in a still frame would look when passing by at 24 FPS. I think we all have faith that the Mysterium will reach it's full potential.

 

[Sanjin Jukic]

If Build 15 and the next builds give us a quality close to the photos below

that were shot on Nikon D3 I would be really, really satisfied>>>

A narrow street in Florence at night.
Digital Photography by Dennis Hissink
Nikon D3 Test Photo
Resolution : 4256 x 2832
Aperture : f/6,3
Exposure time : 1/50 sec
ISO : 6400
Focal length : 24mm
Metering : Aperture Priority
Compensation : 0 step

Download RAW NEF file>>>

Ponto Vecchio at night, Florence.
Digital Photography by Dennis Hissink
Nikon D3 Test Photo
Resolution : 4256 x 2832
Aperture : f/4,5
Exposure time : 1/10 sec
ISO : 6400
Focal length : 18mm
Metering : Aperture Priority
Compensation : 0 step

Download RAW NEF file>>>

Also find more and download original NEF RAW files at

LINK>>>

 

[Graeme Nattress]

A new sensor is an interesting question, I just bought a Nikon D3 and I'm a little flabbergasted by it's noise control.

As far as I can see from images on the web, most of the D3's "low noise" is heavy chroma noise reduction in RAW.

Graeme

 

[Dj Joofa]

If Build 15 and the next builds give us a quality close to the photos below


Exposure time : 1/10 sec

Please see my comment regarding longer integration times (exposure time) above.

 

[Greg Syverson]

That D3 really rocks. I need the Nikon D3 pluse the RED CAM. So many nice things to work with.

 

[Evin Grant]

As far as I can see from images on the web, most of the D3's "low noise" is heavy chroma noise reduction in RAW.

Graeme

Some interesting things I've noticed playing with the D3 files...
•Noise seems to live more in the Red channel Vs. Blue in the D2x.
•There is some on chip pre RAW NR (Possibly chroma) going on but it preserves detail very well.
•The noise "Character" is more grain-like than any previous DSLR I've seen
•Shadow detial and shadow noise is more controlled than any previous DSLR

Graeme, if you like I can shoot some test frames for you to play with.
Let me know.

 

[Sanjin Jukic]

Please see my comment regarding longer integration times (exposure time) above.

Also it's very interesting to read what Joofa and Graeme wrote above and compare it with Nikon's D3 RAW files that are available via links at my post (also above).

 

[Alexander Nikishin]

Just give me 11.3 stops of latitude, multiple outputs, and 100fps in 2k windowed.

I will be happy for life.

Jim.....

Speaking of 100fps over-cranked, is that still a possibility or are we limited to 75?

 

[Dj Joofa]

Sanjin,

It is not easy for a video camera to produce high-quality high-fidelity images for night-time shots or scenes with low-illumination without comprising either (1) frame rate, or (2) exposure time. For daylight shots the relation of exposure time to frame rate does not enter into picture as the exposure time is a small fraction of the frame rate. However, for night-time shots, the situations starts becoming deteriorating and there reaches a point for many a cameras that they have to pick either between dropping frame rate or dropping exposure time -- dropping exposure time results in higher visibility of temporal noise.

 

[Dj Joofa]

Just give me 11.3 stops of latitude, multiple outputs, and 100fps in 2k windowed.

I will be happy for life.

Jim.....

Speaking of 100fps over-cranked, is that still a possibility or are we limited to 75?

Sustaining a 100 fps at daylight is much easier than at doing it at night-time. It will be a significant challenge to come up with 100 fps while keeping a high dynamic range for the type of night-time shots that Sanjin posted.

 

[Patrick Tresch]

dropping exposure time results in higher visibility of temporal noise.

What Sanjin is showing us are pictures shot @ 6500 asa. IMHO I don't think it changes anything if it's 1 second aperture time or 1/1000th. The preset of the CMO is set to 6500asa and therefore apertrure time and T stop are calculated en relation with this setting and therefore the resulting picture will have a noise of 6500 asa.

Interestingly, noise in redcode raw is not evenly spread out in a picture where you have low lights, mid and highlights.

Patrick

 

[Dj Joofa]

Ponto Vecchio at night, Florence.
Digital Photography by Dennis Hissink
Nikon D3 Test Photo
Resolution : 4256 x 2832
Aperture : f/4,5
Exposure time : 1/10 sec
ISO : 6400
Focal length : 18mm
Metering : Aperture Priority
Compensation : 0 step

Download RAW NEF file>>>

Also find more and download original NEF RAW files at

LINK>>>

This is a wonderful image. Please note that the exposure time is 1/10 sec. hence, assuming a very standard shutter angle of 180 degrees that would mean 5 frames per second. Hence, it will be difficult to go over 5 frames per second to get such an image assuming the noise characteristics on the sensor, and other parameters stay constant.

 

[Dj Joofa]

What Sanjin is showing us are pictures shot @ 6500 asa. IMHO I don't think it changes anything if it's 1 second aperture time or 1/1000th. The preset of the CMO is set to 6500asa and therefore apertrure time and T stop are calculated en relation with this setting and therefore the resulting picture will have a noise of 6500 asa.

Interestingly, noise in redcode raw is not evenly spread out in a picture where you have low lights, mid and highlights.

Patrick

Unlike a film camera where some characteristics are determined by the ASA of the film, a digital camera does not have the ability to swap out the sensor and replace it with one with higher/lower equivalent ASA. The parameters available to a digital camera are integration time (exposure time), the size of the aperture, the characteristics of sensor noise, and the bit resolution of the ADC determined by its bit depth. Hence, all ASA equivalents of a digital cameras are worked out using these basic set of parameters.

Noise characteristics and ADC bit depth can't be changed on the digital camera. Hence, the other two free variables left are exposure time and the size of the shutter. The ADC is agnostic to discriminate between the two. It only cares about what would be the Exposure Value (EV), which is a relation of the exposure time and the shutter size.

For the same exposure value, one can have different shutter sizes and exposure times. However, as I said the ADC does not even know that, it is too much at the back of the chain. Hence, the exposure value derives the ADC output, and assuming clipping is not happening, a higher exposure value can result in better noise immunity, whether you get it by increasing the size of the aperture, or increasing the exposure time.

 

[Deanan]

This is a wonderful image. Please note that the exposure time is 1/10 sec. hence, assuming a very standard shutter angle of 180 degrees that would mean 5 frames per second. Hence, it will be difficult to go over 5 frames per second to get such an image assuming the noise characteristics on the sensor, and other parameters stay constant.

Its 2+stop overexposed compared to 180 deg @ 24fps so its more directly comparable to ~1300 iso.

 

[Dj Joofa]

Its 2+stop overexposed compared to 180 deg @ 24fps so its more directly comparable to ~1300 iso.

Deanan, kindly see my remark about iso equivalent on a digital camera above. Thanks.

 

[Deanan]

yeah, i quoted you because we are saying the same thing but in different terms.

 

[Jannard]

Keep in mind that the full frame DSLR's (D3 for example) have the same pixel count (12MP) in a sensor that is much larger (about 3x the area). Which means the pixels are much larger. Which means better low light capability and less noise. And much harder to do high frame rates...

Jim

 

[Dj Joofa]

yeah, i quoted you because we are saying the same thing but in different terms.

Okay, then we are in agreement. I am sorry I did not get you the first time.

 

[Patrick Tresch]

Noise characteristics and ADC bit depth can't be changed on the digital camera.

Your right... but presetting your camera to 6500asa "tells your camera" in wich portion of the sensitivity of your censor you want to grab informations. This is in the low light zone where usually you'll see a lot of noise and where usually, when preset to 200 asa, you should only get deep blacks.

So changing asa, even if it's only a relique of the ancient times where different film speed were used, does in the raw world actually affect the calculation of T stops/exposure times and, at the end ,the overall look of the picture.

Pat:sarcasm: