IQ260 native base ISO is at 100 for normal mode, and 200 for long exposure mode

[Shashin]

I don't think so.

Like Ray, I am just having a hard time with your conclusions. It get more confusing when you say you use ETTR as you are just ignoring ISO anyway. It seems the basis of your conclusion is that you prefer the in-camera histogram when the camera is set to ISO 100.

 

[voidshatter]

On cameras which "pull" from base ISO, there's a giveaway in the DxOmark graphs: the pulled ISO point falls above the nominal 1:1 line. Here are the Canon 5DII and 6D for example; see where ISO 50 falls above the line for both cameras, while every other ISO point tracks slightly below the line:

What I am trying to say is that the lowest point lying on the 1:1 line is regarded as the base native ISO, because when you further raise ISO beyond that with the same aperture and shutter speed, you start clipping more highlights.

 

[Shashin]

What I am trying to say is that the lowest point lying on the 1:1 line is regarded as the base native ISO, because when you further raise ISO beyond that with the same aperture and shutter speed, you start clipping more highlights.

But don't you clip fewer highlights at ISO 200? Why isn't that the base ISO?

 

[voidshatter]

Like Ray, I am just having a hard time with your conclusions. It get more confusing when you say you use ETTR as you are just ignoring ISO anyway. It seems the basis of your conclusion is that you prefer the in-camera histogram when the camera is set to ISO 100.

Yes exactly. I prefer the in-camera histogram at ISO 100 to do my ETTR (which is the native base ISO). If I use the in-camera histogram at ISO 50 to do my ETTR (which is the extended ISO), I then risk blowing out highlight details. For the same aperture:

a) "50 ISO 2 seconds" vs "100 ISO 1 second": the former should have less shadow noise but the latter should have more highlight details. The total DR remains the same;

b) "50 ISO 2 seconds" vs "100 ISO 2 seconds": these should have the same shadow noise and the same highlight details. The total DR remains the same;

c) "100 ISO 2 seconds" vs "200 ISO 1 second": the former should have less shadow noise but they both have the same highlight details. The total DR of the latter is less;

d) "100 ISO 2 seconds" vs "200 ISO 2 seconds": the latter physically overexposes by 1 stop, and when pulled down in pp, the latter has 1-stop less highlight details. The total DR of the latter is less;

e) "200 ISO 2 seconds" vs "400 ISO 1 second": the former should have less shadow noise but they both have the same highlight details. The total DR of the latter is less;

f) "200 ISO 2 seconds" vs "400 ISO 2 seconds": the latter physically overexposes by 1 stop, and when pulled down in pp, the latter has 1-stop less highlight details. The total DR of the latter is less;

...

You can see that ISO 100 is the turning point so ISO 100 is called the native base ISO.

 

[voidshatter]

But don't you clip fewer highlights at ISO 200? Why isn't that the base ISO?

Because ISO 200 has less total DR than ISO 100 does, while ISO 100 has the same total DR as ISO 50. Check my explanation above.

 

[Paul2660]

Voidshatter:

I have taken a few shots this morning, will not be able to get to them until this afternoon.

ISO 50 and 100, are very close, and definitively much closer than 50 and 200. I want to shoot this a bit more, but so far, I feel that I still see the following:

1. At iso 50, a bit cleaner shadows, however remember I am using a tech camera, and I have no 1/3 or 1/2 shutter gaps (this make a huge difference IMO). I still feel that 50 gives me more room in highlights, i.e when I expose to the right for the shadows. So, for example on a recent shoot, where I was working with both a ND 0.9 and CLPL at F11, I pushed (I call it that) to the right as much as .75 a stop. On the screen, camera LCD, the water was totally white showing blown out 255 255 255, however in C1, (I was tethered), I was able to get that high light back and still had water detail. OK, now that was at 1 second. I then went to 1.5 and 2 seconds (remember I am using a copol so hitting these speeds is totally hit and miss as I am in blub). But at 1.5 I could not recover my water, it was totally gone, so just .5 of a sec took it out of range.

I tend to skip over 100 and jump to 200, so I don't have much experience there, but I will look at the stuff I took this morning and see.

2. It's anyone's guess as to what the 260 is doing in Long Exposure Mode. I personally have not taken anything longer than 15 seconds and that at iso 50. I have however use 200 in LEM (long exposure Mode) and I don't see any difference in normal exposures. I had hoped that this would NOT be the case and that somehow Phase One could pull some more magic out of the newer chip. The 160 chip is the same as the P65 thus older and pretty much at end of life. So far I have not seen this and really 200 and 400 in LEM mode look pretty much the same as 200 and 400 in normal mode, in that both are pretty much worthless (sorry but it's true) in darker shadow areas. The details just are not there vs the noise. I have not worked much with LEM 140 either as it's just not much more than 100 in normal mode for standard (non long exposures).

3. In the Library tests that Doug did, I was personally shocked by the 6 second shots that were taken by the 260. There were thousands of stuck pixels and by my thinking, the 260 with a mandatory dark frame should have removed this. With my 260 at exposures up to 15 seconds in low light I have been so far pleased with the results at iso 50.

4. You may be right, in that the iso 50 setting is "fake" and the back is really working at 100. I just can't tell enough from my work, as I always use 50 unless pushed to 200 by conditions, (wind/motion). Only Phase knows this and they don't tend to join this forum or others and only they really know. My rule of thumb, is if I have good light then take the 260, (even with windy conditions) as I can use 200 when needed. In theory 1/60th at 200 should be the same as 1/50th at iso 50, since CCD's really only work at base iso from what I understand. However over and over I see better results at 200 (expect in shadows), so leaves sky etc will be better off. I am a photographer who tends to shift everything to create a pano later on, albeit a short pan, so my shifts to 15mm will suffer the worse with noise.

This whole issue of "what iso is a CCD back really using" is one of those things that you just can't find enough info about. I think that Phase One wants the photographer to think that the CCD backs work the same as a CMOS chip, so the iso is a true increase in gain, but it's not. From what I understand, it only a marker to the raw file.

So for me I guess every shoot is an experiment as it's just hard to tell what combination will work. Again, the ability to be tethered to a S Pro 2 and look at the actual raw helps a lot as you can judge much more "true" info about the image in relation to highlights, shadows noise.

I understand your concern, so really the best idea is shoot both backs if you can, and then make the call from that. Long term, by far the 250 is the better investment. I saw this type of recovery 2.5 years ago with the D800 and overnight I switched 14 years of Canon equipment to Nikon, I have never looked back, the differences at base iso are just that much better. CCD backs by the nature of the CCD chip, just won't get there thus you have to shoot around that issue.

Much more info than you want I am sure, I end it with this, when I get it right with the 260, it looks great and I love the output, which is why I always lead with that back.

Paul

 

[ondebanks]

What I am trying to say is that the lowest point lying on the 1:1 line is regarded as the base native ISO,

But none of the IQ180 points lie on the 1:1 line. They all lie under it. And the one which lies closest to the line is the ISO 35 one. That's why it is base ISO. It comes closest to meeting the nominal ISO standard, while also maintaining the maximum available DR.

because when you further raise ISO beyond that with the same aperture and shutter speed, you start clipping more highlights.

You don't start clipping more highlights...not in this back's RAW files at ISO 50 or 100. Not in most other CCD-MFD backs' RAW files, at all. Not even in your own tests! I already gave the reasons above. The DR window slides around...it doesn't start to close down. You are applying thinking from a different camera design paradigm, the one where increasing gain is applied either in hardware or firmware as ISO is increased, and the ISO curve tracks the slope of the 1:1 line. That does happen eventually in the IQ180 from ISO 200 and up, but not at ISO 50 and 100.

Ray

 

[voidshatter]

But none of the IQ180 points lie on the 1:1 line. They all lie under it. And the one which lies closest to the line is the ISO 35 one. That's why it is base ISO. It comes closest to meeting the nominal ISO standard, while also maintaining the maximum available DR.



You don't start clipping more highlights...not in this back's RAW files at ISO 50 or 100. Not in most other CCD-MFD backs' RAW files, at all. Not even in your own tests! I already gave the reasons above. The DR window slides around...it doesn't start to close down. You are applying thinking from a different camera design paradigm, the one where increasing gain is applied either in hardware or firmware as ISO is increased, and the ISO curve tracks the slope of the 1:1 line. That does happen eventually in the IQ180 from ISO 200 and up, but not at ISO 50 and 100.

Ray

Check #24 post of this thread

 

[Wayne Fox]

A CCD based sensor is really ISOless (supposedly so is the nikon d800, although I haven’t tested that yet, so ISO is basically a metadata setting.

One issue is the actual base ISO of each sensor (or perhaps each sensor batch) varies. So when the backs are made some type of average is used. i’ve heard the sensors in the 180 series can actually be as low as 25 and as high as 50, so 35 was the established base.

Perhaps what you are finding is the actual base ISO of your particular sensor if measured would end up more in the 80 range so a little on the high side as compared to other 160 sensors?

personally thats’ one thing I don’t like about the 180 ... the ISO 35 puts me at too long of exposure times or forces me to underexpose. I wish it was 100, I can always use neutral density when I want to drag the shutter.

 

[ondebanks]

Check #24 post of this thread

I'm not sure which part of your post you are referring to, but I guess it is this? - "You can see that ISO 100 is the turning point so ISO 100 is called the native base ISO."

Ok, let's look at some other MFD backs. Here are two from Hasselblad. By your argument above, you would conclude that these backs have no turning points so they have no native base ISO (!!), while I would conclude that the base ISO is clearly ISO 50 for both. Hasselblad users, which would you agree with?

Meanwhile, to further show just how malleable (arbitrary, meaningless, fake - take your pick) the above-base-ISO settings are in MFD CCD systems, let's compare two Phase One ISO curves, P45+ and P40+. These are both CCDs, 39MP and 40MP respectively, and both are from the "P+" generation of Phase One backs:

It's kind of funny how in one back, Phase One decided to initially track the nominal curve, then had a change of heart and went flat. In the other, they did the exact opposite - initially flat, then sloping!

Oh, and BTW, the P45+ uses the exact same KAF-39000 sensor as the H3DII-39 in the plot above. Same sensor, different ISO curve above ISO 50 - by your definition, they have different turning points (one has none at all!) so they have different base ISOs - but how can that be, as this is the same sensor??

The answer is that firstly, base ISO is 50 for both, and secondly, the back manufacturers can do whatever they please above that base ISO, knowing that it won't make any difference - it's all just underexposure with crappier and crappier shadow detail. The smarter ones (like Hasselblad here) leave ISO as a pure flag, and in this way they at least maintain full (but shifted) DR through the complete ISO range.

Ray

 

[voidshatter]

I'm not sure which part of your post you are referring to, but I guess it is this? - "You can see that ISO 100 is the turning point so ISO 100 is called the native base ISO."

Ok, let's look at some other MFD backs. Here are two from Hasselblad. By your argument above, you would conclude that these backs have no turning points so they have no native base ISO (!!), while I would conclude that the base ISO is clearly ISO 50 for both. Hasselblad users, which would you agree with?

Meanwhile, to further show just how malleable (arbitrary, meaningless, fake - take your pick) the above-base-ISO settings are in MFD CCD systems, let's compare two Phase One ISO curves, P45+ and P40+. These are both CCDs, 39MP and 40MP respectively, and both are from the "P+" generation of Phase One backs:

It's kind of funny how in one back, Phase One decided to initially track the nominal curve, then had a change of heart and went flat. In the other, they did the exact opposite - initially flat, then sloping!

Oh, and BTW, the P45+ uses the exact same KAF-39000 sensor as the H3DII-39 in the plot above. Same sensor, different ISO curve above ISO 50 - by your definition, they have different turning points (one has none at all!) so they have different base ISOs - but how can that be, as this is the same sensor??

The answer is that firstly, base ISO is 50 for both, and secondly, the back manufacturers can do whatever they please above that base ISO, knowing that it won't make any difference - it's all just underexposure with crappier and crappier shadow detail. The smarter ones (like Hasselblad here) leave ISO as a pure flag, and in this way they at least maintain full (but shifted) DR through the complete ISO range.

Ray

This is easy to judge

For the H3D case the base native ISO can be called anything since throughout the whole range you do not clip more highlights and in this case the ISO is just a metadata;

For the P45+ case the base native ISO is 50 (since you start to clip highlights beyond 50);

For the P40+ case the base native ISO is 100 (since you start to clip highlights beyond 100).

The point here is to find out when you start to blow out highlight as you raise the ISO while you stay with the same aperture and shutter speed, so you know that you get the perfect shot with ETTR. Yes, of course if you are familiar with, say, the IQ260 digital back working at 50 ISO and know that you can pull back something like 1.5 stops of highlight beyond the default luminance level 240 stuff in the in-camera histogram warning (just the same as 2.5 stops beyond 100 ISO mode) then okey it's totallly fine.

If you still cannot quite follow what I am talking about, then open dxomark, click on the "Dynamic Range" tab in Measurements for a digital back. If you hover your mouse over the left-most data point and read the "Manufacturer ISO", then that is the base native ISO for the digital back.

 

[alajuela]

Hi Ray

I think I get it

To put it in laymen's terms - (or dumber)

To get the best image - best practice (ideal) would be shoot on a tripod at base iso., adjust shutter speed to your exposure requirement - not mentioning aperture as mentally that is preset. -- This is what I do now on the Phase 80 set at 35 or 50 iso.

After that, when hand holding and options are limited by light - changing iso is in effect underexposing and telling C1 to stretch the captured information -- to put "pixels" in different zones? Stretch the pixels to fulfill the DR?

So if you need a certain shutter / aperture moving up the iso does NOT change the amount of information captured - only redistributes it?

OTOH - Stay at "base iso and adjusting the shutter / aperture - will in effect - give more or less information in your file to work with. We all see how much smaller an underexposed file is compared to a correct or overexposed file.

Is Senor Plus any different? I have found that at iso 400 on the 280 - the file is better (less noise and just smoother) in Sensor plus (20 meg) as opposed to full resolution. Is this my imagination?

Thanks

Phil

 

[ondebanks]

This is easy to judge

For the H3D case the base native ISO can be called anything since throughout the whole range you do not clip more highlights and in this case the ISO is just a metadata;

For the P45+ case the base native ISO is also 50 (since you start to clip highlights beyond 50);

For the P40+ case the base native ISO is 100 (since you start to clip highlights beyond 100).

The point here is to find out when you start to blow out highlight as you raise the ISO, so you know that you get the perfect shot with ETTR.

Then you are adopting a definition of base native ISO that I don't think anyone else uses. To say that the base ISO of the old Hasselblad backs could be ISO 400...I mean, really?! Sure, ETTR would not clip highlights at their ISO 400, but if you ETTR with such a back you are doing one of two things - depending on whether the histogram shows the RAW data or the ISO-flag-scaled data:

- If the histogram shows the RAW data, by using ETTR you are not actually shooting at ISO 400 - you are bypassing the ISO stuff altogether and filling the pixels to just under saturation - which requires the same exposure (shutter and aperture) as at the lowest ISO setting of 50. You will get the same ETTR results shooting ISO 50 without exposure compensation as shooting ISO 400 and forcing +3 stops exposure compensation. Isn't it more natural to define base native ISO as the ISO which allows you to ETTR by normal metering, without the need for exposure compensation?

- If OTOH the histogram shows the ISO-flag-scaled data, you are truly shooting at ISO 400, so by using ETTR you are using 3 stops less exposure than at ISO 50...[why else would you pick ISO 400, if not to tell the meter/AE system to underexpose and use a faster shutter speed?]. But then there would be a severe impact on image quality from underexposing everything in the scene by 3 stops. Which is why ISO 400 certainly could not be considered base ISO. Base ISO implies maximum usage of the pixel well depth, for high signal to noise.

And shouldn't the definition of something as fundamental as "native base ISO" also include an expectation of adhering to the ISO standard, marked by the nominal 1:1 line? "ISO 400" on the Hasselblads is huge distance from the line - it's measured as ISO 46, so it's just over 3 stops from its declared value. Whereas "ISO 50" is the same measured ISO 46, so it's only a fraction of a stop off.

Ray

 

[ondebanks]

If you still cannot quite follow what I am talking about, then open dxomark, click on the "Dynamic Range" tab in Measurements for a digital back. If you hover your mouse over the left-most data point and read the "Manufacturer ISO", then that is the base native ISO for the digital back.

Unfortunately that doesn't really help to pin down base native ISO, as hovering over any of the other points also shows a different value for the "Manufacturer ISO".

But are you really admitting that the left-most data point for DBs is always the base native ISO? That has been my contention all along! :chug:

Ray

 

[voidshatter]

Unfortunately that doesn't really help to pin down base native ISO, as hovering over any of the other points also shows a different value for the "Manufacturer ISO".

But are you really admitting that the left-most data point for DBs is always the base native ISO? That has been my contention all along! :chug:

Ray

I guess you still don't quite get it. You are confusing the left-most data point for the "ISO Sensitivity" tab with the "Dynamic Range" tab.

The base native ISO should be the ISO that gives you the best dynamic range result in the "Dynamic Range" tab. If there are multiple values giving you the same (top) dynamic range performance, then the highest ISO is the native base ISO.

 

[Shashin]

Is there such a thing as "native base ISO"? I have seen the ISO standard; they have no such thing. They do have a saturation-based ISO, but you are not using a method to determine that. They also have a S/N ratio based ISO, but you are not using that either.

Film speed - Wikipedia, the free encyclopedia

 

[ondebanks]

I guess you still don't quite get it. You are confusing the left-most data point for the "ISO Sensitivity" tab with the "Dynamic Range" tab.

Nope. I had checked it in the "Dynamic Range" tab.

BTW, DxO screw up when they plot Dynamic Range for ISO-flag backs like the aforementioned Hasselblads. Rather than measuring it at ISOs above 50, they plot "extrapolated" values which show a steady DR decline of 1 eV/stop. If they took the time to actually measure it in the RAW file rather than extrapolating to what they think the processing software would do, they'd find that DR doesn't decline and is constant across the ISO range.

The base native ISO should be the ISO that gives you the best dynamic range result in the "Dynamic Range" tab. If there are multiple values giving you the same (top) dynamic range performance, then the highest ISO is the native base ISO.

And we come full circle. Why the highest? You have not yet given a reason that stands up. In CMOS cameras, we might pick the highest because their readnoise often decreases with increased ISO. In MFD CCDs, there's no such improvement. If you're relying on an ETTR approach, then in my last post I explained why choosing the higher ISO is either pointless, or damaging to image quality. I wish you'd address those points.

Ray

 

[ondebanks]

Is there such a thing as "native base ISO"? I have seen the ISO standard; they have no such thing. They do have a saturation-based ISO, but you are not using a method to determine that. They also have a S/N ratio based ISO, but you are not using that either.

Film speed - Wikipedia, the free encyclopedia

Will, it's a somewhat fuzzy concept alright - a spin-off from the saturation-based ISO you mentioned.

In our MFD CCDs it's very simple though - there's only one real ISO and it's the saturation-based one (or a value close to it, with more headroom) that the manufacturers arrive at. This is always the smallest ISO value offered with the MFD setup and can be termed base ISO, since it's the optimum setting to shoot at, conditions permitting. As shooting at any higher ISO setting means capturing less light, then the *original* max signal part of the scene now drops 1 or more stops below the saturation level, and since readnoise doesn't change with ISO, signal to noise falls right across the scene. DR however stays constant, since something even brighter and previously saturated can now drop into the sub-saturation counts regime. (Although there's no requirement that there be something brighter to avail of this space - the DR definition [Saturation Level/Readnoise] is blind to what's being captured). If I raise ISO, saturation level aka Full Well Capacity ain't changing and neither is Readnoise. So DR stays constant.

With the CMOS cameras, it's more complicated because a change in ISO can change both saturation level and readnoise, and as a consequence, DR and signal to noise would both change too. In that melee, if the very lowest ISO offered causes overexposure according to the saturation-based ISO, and if its readnoise is worse than the next ISO up, then for reasons of either DR, signal to noise, or both, the next ISO up is a better candidate for base or native ISO. It seems to me that voidshatter has zoned in on that fact and is applying it inappropriately to CCD MFD.

Ray

 

[voidshatter]

Nope. I had checked it in the "Dynamic Range" tab.

BTW, DxO screw up when they plot Dynamic Range for ISO-flag backs like the aforementioned Hasselblads. Rather than measuring it at ISOs above 50, they plot "extrapolated" values which show a steady DR decline of 1 eV/stop. If they took the time to actually measure it in the RAW file rather than extrapolating to what they think the processing software would do, they'd find that DR doesn't decline and is constant across the ISO range.



And we come full circle. Why the highest? You have not yet given a reason that stands up. In CMOS cameras, we might pick the highest because their readnoise often decreases with increased ISO. In MFD CCDs, there's no such improvement. If you're relying on an ETTR approach, then in my last post I explained why choosing the higher ISO is either pointless, or damaging to image quality. I wish you'd address those points.

Ray

There is a reason why dxomark ignore the lower values (non native ISO) when plotting the "Dynamic Range" tab. If you look into comcast's plot, you find things like the following:

Anything below 100 ISO on a D800E does not give you more DR range.

Unless you have definitive analysis of dynamic range in terms of highlight details and SNR in shadow to prove that anything below 100 ISO on a D800E or IQ260 could give you better total DR range, I would rather trust comcast or dxomark instead.

 

[ondebanks]

There is a reason why dxomark ignore the lower values (non native ISO) when plotting the "Dynamic Range" tab. If you look into comcast's plot, you find things like the following:


Anything below 100 ISO on a D800E does not give you more DR range.

Unless you have definitive analysis of dynamic range in terms of highlight details and SNR in shadow to prove that anything below 100 ISO on a D800E or IQ260 could give you better total DR range, I would rather trust comcast or dxomark instead.

?? I have never said that you get better DR below ISO 100 on an IQ260. I said that you get constant DR at ISO 35, 50 and 100 on an IQ180. But you get the best SNR at ISO 35, somewhat less at ISO 50 and even less SNR at ISO 100. So the optimum (base) ISO is 35. Do you dispute any of this? Does comcast?

And I have never referred to the D800E at all - it's irrelevant to this exchange, as it is not using a CCD!

You started this thread with a question about the IQ260. I have answered it as helpfully as I could, based on an intimate knowledge of how CCDs work (I lecture in observational astronomy) and on my own experience as a MFD user. Now seeing as DxO have never tested that back, nor comcast - I don't see a single MFD device listed at comcast -, how can you say that you would trust them over my analysis?

Ray