The New CMOS Sensor May Not Be As Bad As People Imagine

[ramosa]

Count me out. I still am no fan of their ratings.

Jaap,

I agree completely. Part of me is oddly happy that a Leica sensor was rated well, but I'm still not sure what such a result means ... Much better for me is the chance to use a camera and see its output.

Best ...

 

[willie_901]

It means the sensor has an excellent signal to noise ratio. Because SNR is highly correlated to dynamic range, it means the dynamic range is excellent as well.

You should be happy as SNR is one important aspect of performance.

Of course using the camera in conditions important to your work is the ultimate test. At the same time the new sensor's SNR performance can only increase the probability you will be pleased.

Jaap,

I agree completely. Part of me is oddly happy that a Leica sensor was rated well, but I'm still not sure what such a result means ... Much better for me is the chance to use a camera and see its output.

Best ...

 

[semilog]

Jaap,

I agree completely. Part of me is oddly happy that a Leica sensor was rated well, but I'm still not sure what such a result means ... Much better for me is the chance to use a camera and see its output.

Best ...

It means that Leica's engineers, unlike much of the Leica user base, live in the reality-based community.

 

[dcsang]

FYI, in limited crappy lighting testing of my brand new M, there is indeed a 1 stop improvement over the M9, but still about 1 stop behind the Nikons above 1250. So, yes the M has a usable ISO 3200, but you have to "shoot to the right" to mitigate banding in the shadows (as Ming Thein reported, the M preserves highlights pretty well, so no need to use -1 exposure comp as I did on my M9/8).

Thanks for this info.
I can deal with "over exposing" a wee bit (expose to the right) if there's preservation happening in the highlights. I can always re-adjust the image in post to get the look/feel I want out of the photo.

Cheers,
Dave

 

[icebear]

...

This means raw shooters never have to change ISO from the minimum setting. Increasing the Exposure slider in one's preferred raw rendering software during post processing should produce identical results to increasing the camera ISO.

Has anyone ever tried this with their M8 or M9?

I haven't wasted my time to read the LL article but this is utter nonsense. If you seriously underexpose and try to rescue by pushing the exp. slider to the right, you will dramatically increase noise. If on the other side you try to work with base ISO in darkness you will not get usable speeds for handheld work. So what would be the point of not adjusting your ISO setting as required by EV ? Doesn't make any sense to me.

 

[willie_901]

Some cameras only amplify the signal at a single gain value inbetween the the sensor and the analog to digital converter. With these cameras it is impossible to increase the electronic gain. Until recently the only cameras with CCD sensors worked this way. Recently SONY has developed a CMOS sensor has a single gain.

In these designs the integer values produced by the ADC are increased to simulate values that would have been obtained if electronic amplification had been used. I was curious if the M8 and M9 use this method. The Phase One digital backs use this design and they produce excellent results. In fact increasing the camera's ISO value only tells the software what multiplication factor to use.

I would be interested in hearing how you would decscribe what happens to the signal inbetween the time it leaves the sensor and the data is finally written to the in-camera raw file. For instance cameras that increase the electronic gain between the sensor and the ADC also amplify the signal and the noise. Or perhaps you can explain how only the signal is amplified without amplifying the noise?

 

[ampguy]

Hi Willie

Hi Willie

I understand what you are saying, and I only have some experience with the M8, and some info. that I am not sure of (would love to be corrected) on the D7000 (IMX701 sensor).

The M8's gain increases up to 2500 shows the noise over lower settings, that can display the noise associated with gain. Plus there's the issue that with some that had vertical lines or uneven exposures between parts of the image, that the fix was actual capacitors in the L or R image side, etc.

re: the D7000 sensor, I think the on-chip IMX701 Prog. Gain Amp. (PGA) is always on up to 800 or 1600. Whether it can or vendor firmware ever has bypassed it, is not likely, since vendors like Pentax who use the same sensor, but not the Nikon expeed, etc. firmware show the same e noise levels. My guess is that the basic gain balance per color channel are needed to some point. But would be curious if anyone had other info. I hope to get a refurb D5100 at some point for my daughter, to replace her broken D40x.

I can understand how a pro raw only (remember a NEF isn't raw sensor data, it has sensor data + expeed info/fixes added) system can use the sensor's optimal base gain for low noise, and then let a full pc do the developing.

But a dslr, or even high-end RF that does JPGS/DNGs, needs to have something with semi-decent images (e.g. the black images of the older CCD Nikons (D100?), that could actually be recovered if shot RAW, aren't likely coming back...) coming out of the camera media. The PGAs gains have been increasing from 24 to 36ish db (from 4 to 6 stops), but consumer expectations of low-noise high iso, plays a part on where a manufacturer will place that gain, and how much, if any, control they can give the user.

Some cameras only amplify the signal at a single gain value inbetween the the sensor and the analog to digital converter. With these cameras it is impossible to increase the electronic gain. Until recently the only cameras with CCD sensors worked this way. Recently SONY has developed a CMOS sensor has a single gain.

In these designs the integer values produced by the ADC are increased to simulate values that would have been obtained if electronic amplification had been used. I was curious if the M8 and M9 use this method. The Phase One digital backs use this design and they produce excellent results. In fact increasing the camera's ISO value only tells the software what multiplication factor to use.

I would be interested in hearing how you would decscribe what happens to the signal inbetween the time it leaves the sensor and the data is finally written to the in-camera raw file. For instance cameras that increase the electronic gain between the sensor and the ADC also amplify the signal and the noise. Or perhaps you can explain how only the signal is amplified without amplifying the noise?

 

[willie_901]

amp guy,

Thank you for answering my question about the M8/M9. I was curious about how the CCD signal path was handled. This is not a pure academic question. Different cameras require different strategies to make the most of their potential. Understanding differences in what happens between the sensor and writing the raw integer data can improve the technical results.

My initial comments regarding ISO were based on a post by Emil Martin on LuLa.

http://www.luminous-landscape.com/forum/index.php?topic=56906.0

Please note the plots in his response on Aug 13 2011.

Re: ETTR: ISO vs shutter speed
« Reply #9 on: August 13, 2011, 10:26:43 AM »

His comments on increasing ISO with read-noise constant cameras are at the end of the post.

Thanks again for sharing your experience with the Leica CCD sensor.

 

[furcafe]

Following up, here are some shots (+ 35/1.4 'lux ASPH & all ISO 3200) from a recent show, w/typically horrible club lighting (I used Noise Ninja at the same levels I used for ISO 1250 for the M9):

You can compare these shots in a mix w/X-Pro1 (+ 35/1.4 & ISO 3200) shots from the same show:

http://brightestyoungthings.com/articles/photos-black-angels-elephant-stone-allah-las-black-cat.htm

Thanks for this info.
I can deal with "over exposing" a wee bit (expose to the right) if there's preservation happening in the highlights. I can always re-adjust the image in post to get the look/feel I want out of the photo.

Cheers,
Dave

 

[dcsang]

Chris,

Thanks for those!! I'll review them when I'm at home later today - but they look really good for ISO 3200 (especially compared to what we've come to expect - "high" ISO of maybe 2500).

This is making me convinced

Cheers,
Dave

 

[eleskin]

M vs X Pro

M vs X Pro

furcafe: How did the X Pro compare to the M? Were they close in performance? I would love to hear your observations on this.

 

[furcafe]

The X-Pro1 is more like the D700 (the only other digital I currently use; R-D1 is basically retired). Now that I've used all 3 cameras in bad lighting in concerts, restaurants, & bars (my primary use for digital, BTW), I think both the X-Pro1 & D700 are @ least 1 stop better than the M in the high ISO department. However, the M is still 1 stop better than my old M9 & its AWB is definitely superior (for some reason, of all of these cameras, the X-Pro1 seems to have the best AWB). All of this seems to be consistent w/the reviews I've read, so I'm happy.

The X-Pro1 & D700 don't have RF focusing, which is important to me as I still consider it to be the best manual focus system for available darkness shooting. The X-Pro1 is small & light but the AF is dodgy in low light, even w/the latest firmware. The D700's AF is pretty good in low light, definitely better than the X-Pro1, but that's usually outweighed by the fact that it's your typical dSLR brick (& really bad for manually focusing in low light).

A Leica M body w/full-frame Fuji or Sony/Nikon sensor would appear to be my dream camera, but the 240 is good enough for me.

furcafe: How did the X Pro compare to the M? Were they close in performance? I would love to hear your observations on this.

 

[furcafe]

If you must have RF focusing, have a decent investment in Leica mount glass, need decent high ISO performance, & have the spare cash (usually the biggest "if"), it's a no-brainer. If you're part of the daylight shooting brigade & need an optical RF, a used M9 or new M-E is a better choice; if you can live w/AF, there are all the other mirrorless options.

Chris,

Thanks for those!! I'll review them when I'm at home later today - but they look really good for ISO 3200 (especially compared to what we've come to expect - "high" ISO of maybe 2500).

This is making me convinced

Cheers,
Dave

 

[willie_901]

For what it's worth I also find the overall sensor performance of the X-Pro 1 and D700 to b very similar. With raw the D700 recovers shadows better while the Fuji recovers highlights better. In the end there isn't that much difference because you just expose the two cameras a bit differently. I prefer the Fuji results overall. The final images are cleaner.

I prefer the Fuji fast primes because they have much less longitudnal CA than the new G Nikkors.

 

[furcafe]

Agreed.

I think I posted some of these shots in another, X-Pro-centric, thread, but here are 2 similar shots from the same show last year, the 1st w/the X-Pro1 + 18/2, the 2nd w/the D700 + 28/1.4 AF-D, both @ ISO 3200 & f/2:

For what it's worth I also find the overall sensor performance of the X-Pro 1 and D700 to b very similar. With raw the D700 recovers shadows better while the Fuji recovers highlights better. In the end there isn't that much difference because you just expose the two cameras a bit differently. I prefer the Fuji results overall. The final images are cleaner.

I prefer the Fuji fast primes because they have much less longitudnal CA than the new G Nikkors.

 

[dcsang]

It's interesting because my intent is to replace both the D700 and X-Pro1 if I get the M (240).

I don't want/need to be carrying around 3 systems so I would like to concentrate on one - of course the issue is there are benefits/drawbacks/sacrifices one has to make depending on which you go with. Each of the cameras has their strengths.. and their weaknesses.

Cheers,
Dave

 

[furcafe]

As you know, I've gone full circle, too. My progression for low light shooting was getting the D700 to replace the M9 when I needed ISO 3200 & above. Then I found the D700 to be annoyingly unwieldy, so I got the X-Pro1, which turned out to have annoyingly erratic AF just when I most need accurate AF. So now I'm back in the Leica fold w/the 240.

I guess I could have saved a lot of money if I wasn't so easily annoyed. The journey was kind of fun, though, as the D700 was my 1st really modern SLR & the X-Pro1 is an interesting take on current camera tech.

It's interesting because my intent is to replace both the D700 and X-Pro1 if I get the M (240).

I don't want/need to be carrying around 3 systems so I would like to concentrate on one - of course the issue is there are benefits/drawbacks/sacrifices one has to make depending on which you go with. Each of the cameras has their strengths.. and their weaknesses.

Cheers,
Dave

 

[dcsang]

I hear you Chris - I've tried really hard to make the X-Pro1 work for "work" (weddings/portraits) and while it is great for standard portraits (i.e. "Sit still and look here") it bites for any sort of work where there's movement. The continual AF on the camera is useless. The D700, as you have said, is awesome but you have to be prepared to wield that beast (i.e. workout your arms, shoulders and back muscles )

I'm willing to give up the "ability" to shoot Macros. I'm willing to give up continual AF (D700) or any AF as long as I can zone focus and get the bride/groom during a processional/recessional etc. I've done it before using my Leicas and film so the concept/plan of attack should be the same. All I'm asking for is decent ISO3200 - doesn't have to be D700 or D800 clean but it's gotta be a lot better than what ISO2500 on the M8/M9 in dirty light produces.

Cheers,
Dave

 

[furcafe]

Word.

At this point, I'm going to keep the D700 for those rare occasions when I need things like continuous/matrix focusing & TTL flash & the X-Pro1 to use as my M backup/buddy.

I hear you Chris - I've tried really hard to make the X-Pro1 work for "work" (weddings/portraits) and while it is great for standard portraits (i.e. "Sit still and look here") it bites for any sort of work where there's movement. The continual AF on the camera is useless. The D700, as you have said, is awesome but you have to be prepared to wield that beast (i.e. workout your arms, shoulders and back muscles )

I'm willing to give up the "ability" to shoot Macros. I'm willing to give up continual AF (D700) or any AF as long as I can zone focus and get the bride/groom during a processional/recessional etc. I've done it before using my Leicas and film so the concept/plan of attack should be the same. All I'm asking for is decent ISO3200 - doesn't have to be D700 or D800 clean but it's gotta be a lot better than what ISO2500 on the M8/M9 in dirty light produces.

Cheers,
Dave

 

[airfrogusmc]

A good read and compared to the D800.

Sorry if this had already been posted here.
http://www.luminous-landscape.com/reviews/cameras/i_will_not_buy_that_camera_i_promise.shtml