The Future of CCD Sensors

[ErikKaffehr]

Hi,

Just to make it clear, I did compare what I have. The P45+ may be old, but it still costs 5X the Alpha 99. It also has 2.1X times the area of the Sony Alpha 99 sensor, giving it a significant advantage.

Canon doesn't do noise suppression on raw images, BTW. They do CDS but that is not noise reduction. Anyway I don't have any Canon.

That said, Doug did publish a comparison between IQ 260, 280 and 250 in the library shots.

Chris Barret was kind enough to share a comparison between Sony A7r and IQ 260.

All those files were available as raws.

My interpretation was that the CMOS sensors outperformed the CCD-s by a clear margin, but in both cases it was contrast lighting, making demands on DR.

Here are the threads:
http://www.luminous-landscape.com/forum/index.php?topic=87187.80

http://www.luminous-landscape.com/forum/index.php?topic=84842.0

If you have better data, you are most welcome to share!

Best regards
Erik

The ideal look is not important.
What is important is the real look under best conditions.
You can also make comparisons under harsh conditions, but then they apply only to those specific conditions.

The models you quoted are not best of breed, so I would not use them to come to a conclusion.
P45 is using the old Kodak sensor, and we know very well that the DALSA sensors produce better images.
We also know that the IQ/Credo produces better images than P/Aptus, so that is what needs to be compared.

Comparing 5-6 years old technologies maybe gives us a picture of what was right then, but we want to know the answer that is applicable today, considering what the relevant companies are actually making.

I believe the plastic wrap look is a result of exaggerated noise reduction, not of naturally low noise in CMOS.
When you are comparing a Canon image to a Leaf image, you are not only comparing CCD to CMOS, but also the two companies very different approaches to noise reduction and general rendering.

If Doug would feel generous, he could release to us a couple of images by IQ CMOS and IQ CCD using controlled lighting and the lowest ISO each back does, and then we will know what is best right now.

[sbernthal]

You you please show me where those files are available?

[Paul2660]

You you please show me where those files are available?

DT about 1 month ago did a sweep series on the IQ250 and posted the results on their blog.  You can find the 250 sweep here:

https://digitaltransitions.com/blog/dt-blog/iq250-iso-sweep

On the same day DT did the same series with the IQ260, I don't believe these images were posted, but you can email Doug for them.  It's basically what you are looking for, and these are the raw files, so you can apply the amounts of noise reduction you wish. 

Paul

[sbernthal]

The point Doug was making is that these files are available on request to his clients only, meaning not for the clients of the other guy.
Also the file posted is a compound jpg from several different exposures.
I think a couple of 1-1 jpgs posted here would put this argument to rest.

[Paul2660]

I was able to look at the files being a client.  The images are pretty telling, in that you can see the IQ250 will easily hold up to 1600 ISO before you really start to see a lot of detail loss.  Where as the IQ260 pretty much was maxed out by iso 400, in the shadows.  ISO 800 was pretty much gone there.  However I was able to get some of the image back using some of the topaz filters.

After spending wayyy to much time on this lately, I have no doubt that the IQ250 far exceeds any of the current IQ CCD backs, in regards to DR range, highlight to shadows.  You can pull about as much detail from the IQ250 iso 800 image as you can from a IQ260 iso 100 image.  The IQ260 does go to one lower stop, 50, but pushing a CCD much past 1.5 stops from base ISO seems pretty pointless.

I had hoped that Phase One would be able to do more with the IQ260 images in the long exposure mode, when using ISO 400 and even 800, but from the use of my 260, it's just not there.  You might see just a tad of difference, but really most times not enough to matter.

I will let the experts talk to color as both chips, Dalso in the IQ260 and Sony in the 250 seem to do a very good job here.  But on noise and shadow detail, the 250 wins hands down.

Paul

[sbernthal]

I don't see the point in comparing iso 1600 - it's pretty much a foregone conclusion CMOS will be better.

What I would like to know is how IQ iso 35 compares to CMOS iso 100, both properly lit.
Meaning - is there any loss moving to CMOS under the best conditions?

[ErikKaffehr]

Hi,

This is from Doug's test.

Left IQ250 at 100 ISO right IQ280 at 35 ISO

Exposure pushed 3.0 EV on both in C1, no noise reduction on either and no sharpening.

Best regards
Erik

I don't see the point in comparing iso 1600 - it's pretty much a foregone conclusion CMOS will be better.

What I would like to know is how IQ iso 35 compares to CMOS iso 100, both properly lit.
Meaning - is there any loss moving to CMOS under the best conditions?

[sbernthal]

1. These files are pushed!
2. This is a comparison intended for noise aspects.
3. Screen capture is not a complete file.

If you want to settle this discussion, you need one jpg (save as jpg 12) from each back, no pushing or any modification, under the best conditions - in studio lowest iso perfect lighting.

This gives you the baseline - the best from one back compare to the best from the other back.
After that you can compare pushing or high iso or highlights or lowlights or any other extreme that you like.

But where is the best image from 1, and the best image from 2 - each in its own jpg?
I would expect a studio photographer to get the lighting right in most cases, so you want the "best" comparison, but also you want to know what happens if you don't get the best lighting - that is a second and different comparison. I'm seeing only the second comparison.
If I went only by the second comparison, I would buy a Canon (until now, maybe now they can beat Canon/Nikon).
The high iso/pushing comparison should be against D800, not against IQ160 which will lose for sure.

If they made a back that can do =IQ160 in perfect conditions, and =D800 in bad conditions, then it is a winner.
Maybe it is, I'm not saying it's not, but I'm not seeing any files yet to tell me.

[ErikKaffehr]

Hi,

Pushing the image shows the capability of the sensor to reproduce shadow detail.

Not everyone is shooting in studio in perfect lighting, that is the easiest case for any camera. Anyway these are the best comparisons I have seen between the IQ series backs.

Best regards
Erik

1. These files are pushed!
2. This is a comparison intended for noise aspects.
3. Screen capture is not a complete file.

If you want to settle this discussion, you need one jpg (save as jpg 12) from each back, no pushing or any modification, under the best conditions - in studio lowest iso perfect lighting.

This gives you the baseline - the best from one back compare to the best from the other back.
After that you can compare pushing or high iso or highlights or lowlights or any other extreme that you like.

But where is the best image from 1, and the best image from 2 - each in its own jpg?
I would expect a studio photographer to get the lighting right in most cases, so you want the "best" comparison, but also you want to know what happens if you don't get the best lighting - that is a second and different comparison. I'm seeing only the second comparison.
If I went only by the second comparison, I would buy a Canon (until now, maybe now they can beat Canon).

[sbernthal]

Not everyone is shooting in studio in perfect lighting, that is the easiest case for any camera. Anyway these are the best comparisons I have seen between the IQ series backs.

Of course not everyone, but let's agree this is a very significant segment of the back clients.
If we compare what is easiest for the camera, between Canon/Nikon and IQ/Leaf, then the latter wins, and this is the reason I own it.

[eronald]

 In a way, this whole discussion is irrelevant.

 Phase and Hassy killed CCD.

 They overpriced their product and underinvested in new chip and body developments, taxed their customers like a monopolist and pressured their chip suppliers for the best price.  The suppliers reacted by recycling their existing designs as far as they could take them without major reinvestment.

 During the same period, Sony used the money from their sensor foundry to create a fountain of new chips which are actually drop-in modules, and are used by the whole industry - nobody was stopping Dalsa and Kodak from going the same way.

 As a result we have a super-modern CMOS all-singing all-dancing design fighting it out with geriatric 5 year old MF CCDs. There is no state of the art all-digital MF CCD with on-chip A/D available - in fact one can argue that Phase and Hassy wanted there to be none because it would have been a drop-in camera. Set me right if I'm wrong. I for one believe that MF-CCD was a perfectly good technology that was pushed into irrelevance by greed.

Edmund

[Doug Peterson]

The point Doug was making is that these files are available on request to his clients only, meaning not for the clients of the other guy.
Also the file posted is a compound jpg from several different exposures.
I think a couple of 1-1 jpgs posted here would put this argument to rest.

We have, and will again in the future, make these (and more) available for public download.

As I said *at the moment* we are changing storage systems from Dropbox to google drive, so the process of sharing is time consuming and hence I can only offer to do this for our clients. We have 300gb of sample and test files spanning nearly every back for every genre, so the switch in storage system for our small company is no overnight thing. When the switch is complete we will update the public links already there and add some new files (including skintone from the weddings I've shot with the iq250). We will never post our entire catalog publicly for any number of practical and business reasons.

[sbernthal]

I think we can wait - it's not exactly a matter of life and death.

[sbernthal]

In a way, this whole discussion is irrelevant.

 Phase and Hassy killed CCD.

 They overpriced their product and underinvested in new chip and body developments, taxed their customers like a monopolist and pressured their chip suppliers for the best price.  The suppliers reacted by recycling their existing designs as far as they could take them without major reinvestment.

 During the same period, Sony used the money from their sensor foundry to create a fountain of new chips which are actually drop-in modules, and are used by the whole industry - nobody was stopping Dalsa and Kodak from going the same way.

 As a result we have a super-modern CMOS all-singing all-dancing design fighting it out with geriatric 5 year old MF CCDs. There is no state of the art all-digital MF CCD with on-chip A/D available - in fact one can argue that Phase and Hassy wanted there to be none because it would have been a drop-in camera. Set me right if I'm wrong. I for one believe that MF-CCD was a perfectly good technology that was pushed into irrelevance by greed.

Edmund

I don't think it makes a very big difference to phase/hass where they get their sensors - other than how much they pay for them, of course. If the sensor is good and the price is good, they will take it. Why should they, or us for that matter, care whether CCD lives or dies? Is there anything special in CCD that isn't in CMOS? (I don't know)

If A/D is on chip or on board, does that make a difference for the end user? (I don't know)

I'm sure what you said about putting the money back into r&d is correct - but what does it matter? Phase is not a charity organization. In their market segment, I am amazed that they could survive and continue to pay salaries to their employees and provide support for their clients. Almost all other MF firms went under. Let's agree  that keeping the company alive is a pretty good goal. Some investors gut the company, sell off the ingredients and ho home.

[eronald]

deleted

[uaiomex]

But many CCD lovers claim is "the ccd look". Whatever that is.
Eduardo

Hi,

It doesn't seem that CCDs have any advantage over CMOS, possibly except astronomical applications where they are cooled with liquid nitrogen.

Well, except they are made in larger sizes… Also, it may be that they may work better with large beam angles, but that may also change.

Best regards
Erik

[Manoli]

Phase is not a charity organisation [...] I am amazed that they could survive

Did they ?
As you knew them, they didn't. Predatory pricing brought its own rewards.

[ErikKaffehr]

Hi,

Regarding on chip AD conversion, they have two advantages:

Noise levels are much lower, the best MFD CCDs perhaps have something like 12 electron charges of readout noise, the best on chip ADCs can reach around 3 electron charges. That means that deep shadows on CMOS with on chip ADCs have four times less shadow noise and ten therefore to have two stops extra dynamic range. A larger sensor has a natural advantage the IQ 280 has 69% larger area than the IQ 250, and that helps a bit.

The other great advantage of on chip AD-s is that the sensor vendor handles everything. The signal from the sensor is a digital one, immune to noise.

Mid tones are dominated by shot noise, that is pretty much dependent on sensor size and not much else, if exposure is optimal.

Highlights can have a noise caused by pixel non-uniformity. This factor would also get better on CMOS, I guess, as more modern technology is used and tighter design rules are employed.

And advantage of CCDs used to be fill factor, but modern CCDs seem to be needing micro lenses (both the IQ-260 and the IQ-280 use microlenses) so CCD fill factor is obviously not near 100%.

As a side note on the sample files. Capture One uses different defaults of sharpening and noise reduction for different sensors. I disabled sharpening and noise reduction on both images I have posted.

Best regards
Erik

I don't think it makes a very big difference to phase/hass where they get their sensors - other than how much they pay for them, of course. If the sensor is good and the price is good, they will take it. Why should they, or us for that matter, care whether CCD lives or dies? Is there anything special in CCD that isn't in CMOS? (I don't know)

If A/D is on chip or on board, does that make a difference for the end user? (I don't know)

I'm sure what you said about putting the money back into r&d is correct - but what does it matter? Phase is not a charity organization. In their market segment, I am amazed that they could survive and continue to pay salaries to their employees and provide support for their clients. Almost all other MF firms went under. Let's agree  that keeping the company alive is a pretty good goal. Some investors gut the company, sell off the ingredients and ho home.

[sbernthal]

Is the only reason there is no AD on chip CCD, that they didn't invest the time in r&d?

Other than the location of the AD, is there any other reason CCD will produce a different image than CMOS? Maybe the famous differences everybody are talking about, come from the different ADC?

[ErikKaffehr]

Hi,

The reason that CCDs don't do on chip ADC is mainly that the chips work differently. CMOS just measures a voltage, while the CCD pops the electrons from cell to cell. So a cell is popped several thousands of times before it is read out. Reading a CCD pixel is destructive, it can only be read once. CMOS can be read any number of times. This is used in CDS (Correlated Double Sampling) which is used on CMOS to eliminate reset noise.

Another problem with CCDs used to be that they heat up. Each pixel is first shifted downwards several thousand times and than shifted horisontally several thousand times. The areas nearest to the preamps feeding the ADC-s heat up.

Another practical advantage of CMOS over CCD is that complex circuitry can be built into the sensor as CMOS construction is normally used for processing type logic. So lot of technology that is practical with CMOS is not practical with CCDs. But, CMOS is more difficult to make, needs better processing facilities and tighter design rules, so CMOS needs higher investments in processing facilities. No small scale business.

Regarding ADCs, they just take an analogue signal like 10 micro volts and convert in a number. There are only two ways to do it, correctly or not. Off sensor ADC-s need to handle a lot of conversions each second. Say we have 6 readout channels and 39 MP. Assuming 1.5 FPS the ADCs need to process 39000000*1.5/6 =975000 pixels per second.

The IQ 250 has 8280 ADCs and 50 MP. Still shooting 1.5 FPS the ADC process 50000000*1.5/8280= 9058 samples per second. So, the IQ 250 can use a lot of simple but accurate (Wilkinson type) raw converters, while the CCD backs need much faster and much more complex converters.

Best regards
Erik

Is the only reason there is no AD on chip CCD, that they didn't invest the time in r&d?

Other than the location of the AD, is there any other reason CCD will produce a different image than CMOS? Maybe the famous differences everybody are talking about, come from the different ADC?