How about Dynamic range?

[Clark666]

Thank you. That answered my question. I will just work with RAW unless my CF card is full and I need the storage space.
--
http://www.flickr.com/photos/75673106@N00/

 

[coreyh]

I know what dynamic range is. I also don't care much about high iso dynamic range. 95% of my shooting is at ISO 100. While the 7D has more dynamic range, no doubt about that, I would much prefer a camera that has the dynamic range of Fuji's S5pro.

That comes at a price; the DR is increased, but the quality at various levels is inferior. For stable subjects, you're better of taking multiple exposures.

P.S. Doesn't Highlight headroom also play into dynamic range to a certain extent?

DR = headroom + footroom.

Ya, so why did you make the comment about highlight headroom as if it isn't related to dynamic range? In fact, highlight headroom doesn't even make much sense to me because it is a function of exposure. Either detail is captured or it isn't.

--
John

--
Canon 10D
Canon EF 50mm MKII
Sigma 10-20mm EX DC
Canon 70-300mm IS
http://coreyhardcastle.smugmug.com

 

[Adam2]

I took my 7d and 1dmkII out in the woods today around sunset. While the 7d images looked fairly good on the camera display I was quite disappointed once they were viewed on the monitor. Although I shot in RAW and processed in DPP, there was something amiss in the highlights of the 7d images. They appeared a bit clipped though the rest of the image was properly exposed. Compared to the venerable 1dmkII, the differences were striking.

For comparison, the 1dmkII has around 9.2 stops of DR to the 7d's 8.3. Also, it appears from the review that the 7d exhibits less DR headroom in the highlights. No amount of PP could rectify the rather glaring distinctions. Personally, for the $ I would pick up a 5d or 1dmkII (if better AF is needed) rather than a 7d. Alternatively, wait a couple of months and the price of a good used 1dmkIII should be around the same $. While I loved the high ISO performance and improved resolution, and appreciated the improved AF compared to other xxd bodies, the DR issue is a disappointment. The 7d is going back.

I don't know how you're gauging this, but the 7D image, taken as a whole, has more DR than either of these cameras, by the normal definition of DR. You are talking about highlight headroom, it seems, and that is often more a matter of metering mode and contrast settings than it is RAW headroom.
--

What I am using as reference are images shot under similar lighting with similar settings processed in DPP. Now, perhaps the 7d RAW converter needs to be optimized, but the differences exist. There is something about the 7d images that don't look right. Unlike the 5d or 1dmkII images, the 7d produces images with abrupt highlight transitions and which demonstrate highlight clipping. The tonal transitions and highlights appear better managed in the 1dmkII and 5d images when converted from RAW. Again, this may be a software issue rather than a sensor DR or RAW headroom deficiency.

 

[John Sheehy]

What I am using as reference are images shot under similar lighting with similar settings processed in DPP. Now, perhaps the 7d RAW converter needs to be optimized, but the differences exist. There is something about the 7d images that don't look right. Unlike the 5d or 1dmkII images, the 7d produces images with abrupt highlight transitions and which demonstrate highlight clipping. The tonal transitions and highlights appear better managed in the 1dmkII and 5d images when converted from RAW. Again, this may be a software issue rather than a sensor DR or RAW headroom deficiency.

You seem to be talking about highlight headroom, rather than DR. The 7D is a little short in highlight headroom at ISO 100, because the RAW data clips at a lower level than it does at ISO 200.

ISOs 160, 320, 640, 1250, and 2500 may be a little short, too, because they are pulled from higher ISOs, and their RAW values are scaled down. And there may be other factors, but headroom and DR are two different things.

--
John

 

[Canochrome]

What I am using as reference are images shot under similar lighting with similar settings processed in DPP. Now, perhaps the 7d RAW converter needs to be optimized, but the differences exist. There is something about the 7d images that don't look right. Unlike the 5d or 1dmkII images, the 7d produces images with abrupt highlight transitions and which demonstrate highlight clipping. The tonal transitions and highlights appear better managed in the 1dmkII and 5d images when converted from RAW. Again, this may be a software issue rather than a sensor DR or RAW headroom deficiency.

You seem to be talking about highlight headroom, rather than DR. The 7D is a little short in highlight headroom at ISO 100, because the RAW data clips at a lower level than it does at ISO 200.

ISOs 160, 320, 640, 1250, and 2500 may be a little short, too, because they are pulled from higher ISOs, and their RAW values are scaled down. And there may be other factors, but headroom and DR are two different things.

--
John

Interesting. Interesting because I've noted a bit harder "highlight transition" as well. I'll be experimenting for quite a while. It's different for me, because I'm coming from 4 years of the 5-series which have some pretty malleable files.

--
Tom

 

[Jair Rivera]

I mean, he has have SO much experience

http://forums.dpreview.com/forums/read.asp?forum=1025&message=25379945

almost 2 years of experience since his first studio shot, so he knows about this magical apparatus called "camera"

(FOR THE OP: the 50D has about 2/3 stop more headroom in RAW, but a clear disadvantage on JPEG, about 2 stops, but thats not a sensor issue, its a software problem, easily corrected when converting RAW).
--
http://xclusix.blogspot.com

 

[Steen Bay]

What I am using as reference are images shot under similar lighting with similar settings processed in DPP. Now, perhaps the 7d RAW converter needs to be optimized, but the differences exist. There is something about the 7d images that don't look right. Unlike the 5d or 1dmkII images, the 7d produces images with abrupt highlight transitions and which demonstrate highlight clipping. The tonal transitions and highlights appear better managed in the 1dmkII and 5d images when converted from RAW. Again, this may be a software issue rather than a sensor DR or RAW headroom deficiency.

You seem to be talking about highlight headroom, rather than DR. The 7D is a little short in highlight headroom at ISO 100, because the RAW data clips at a lower level than it does at ISO 200.

Doesn't that indicate that the base ISO is higher than ISO 100?

ISOs 160, 320, 640, 1250, and 2500 may be a little short, too, because they are pulled from higher ISOs, and their RAW values are scaled down. And there may be other factors, but headroom and DR are two different things.

So, then ISO 125, 250, etc. must have more headroom, because they are pushed from lower ISOs! (?) But don't you usually say that it's ISO 125, 250, etc. that should be avoided? (because of relatively high read noise?)

 

[ejmartin]

What I am using as reference are images shot under similar lighting with similar settings processed in DPP. Now, perhaps the 7d RAW converter needs to be optimized, but the differences exist. There is something about the 7d images that don't look right. Unlike the 5d or 1dmkII images, the 7d produces images with abrupt highlight transitions and which demonstrate highlight clipping. The tonal transitions and highlights appear better managed in the 1dmkII and 5d images when converted from RAW. Again, this may be a software issue rather than a sensor DR or RAW headroom deficiency.

You seem to be talking about highlight headroom, rather than DR. The 7D is a little short in highlight headroom at ISO 100, because the RAW data clips at a lower level than it does at ISO 200.

Doesn't that indicate that the base ISO is higher than ISO 100?

Yes; Canons have been that way for a while now.

ISOs 160, 320, 640, 1250, and 2500 may be a little short, too, because they are pulled from higher ISOs, and their RAW values are scaled down. And there may be other factors, but headroom and DR are two different things.

So, then ISO 125, 250, etc. must have more headroom, because they are pushed from lower ISOs! (?) But don't you usually say that it's ISO 125, 250, etc. that should be avoided? (because of relatively high read noise?)

They are pushed, so they have less headroom; some of the highlight headroom from the ISO they were pushed from (100, 200, etc) has been thrown away by the push.

--
emil
--



http://theory.uchicago.edu/~ejm/pix/20d/

 

[DRG]

I mean, he has have SO much experience

almost 2 years of experience since his first studio shot, so he knows about this magical apparatus called "camera"

Perhaps since your own experience here is less than one year old, you may not realize that Gabor is (along with J Sheehy, E Martin and a few others) among a handful of posters to this forum who have actually performed extensive analysis on the raw data of many cameras.

David

 

[John Sheehy]

What I am using as reference are images shot under similar lighting with similar settings processed in DPP. Now, perhaps the 7d RAW converter needs to be optimized, but the differences exist. There is something about the 7d images that don't look right. Unlike the 5d or 1dmkII images, the 7d produces images with abrupt highlight transitions and which demonstrate highlight clipping. The tonal transitions and highlights appear better managed in the 1dmkII and 5d images when converted from RAW. Again, this may be a software issue rather than a sensor DR or RAW headroom deficiency.

You seem to be talking about highlight headroom, rather than DR. The 7D is a little short in highlight headroom at ISO 100, because the RAW data clips at a lower level than it does at ISO 200.

Doesn't that indicate that the base ISO is higher than ISO 100?

Sort of, but the amount of RAW headroom at ISO 100 on the 7D is actually generous compared to the base ISO and even higher ISOs of many cameras. If the ISO of the 7D were rated like historical P&S sensors at all ISOs, or even MF cameras and backs at the lowest supported ISO, it might start at 64 or so.

Canon has allowed RAW headroom to fluctuate a bit with models and ISOs within models with recent models, but 3.5 stops of RAW headroom above metered gray was pretty much standard for a while.

Out of a maximum of 14335 levels (16383 minus 2048), my 7D gives clipping points of 11534, 11532, 10601, and 13256 at ISOs 100, 125, 160, and 200, respectively, measured just now. I have had higher values in the past, though, so it may not even be constant within a camera. There have been strange cases of Canons pushing RAWs by 1/3 stop with wide-open wide lenses ...

ISOs 160, 320, 640, 1250, and 2500 may be a little short, too, because they are pulled from higher ISOs, and their RAW values are scaled down. And there may be other factors, but headroom and DR are two different things.

So, then ISO 125, 250, etc. must have more headroom, because they are pushed from lower ISOs! (?) But don't you usually say that it's ISO 125, 250, etc. that should be avoided? (because of relatively high read noise?)

Past comments and current impressions are made from past cameras. Based on the numbers I just showed, however, it seems Canon may be clipping more headroom in the 125/250 group than previously (they used the full values up to 16383 before subtracting blackpoint), and now the 160/320 group have more DR (but less headroom) than the "main" ISOs. I gotta go now, I'll look into it more when I get home later.

--
John

 

[Steen Bay]

What I am using as reference are images shot under similar lighting with similar settings processed in DPP. Now, perhaps the 7d RAW converter needs to be optimized, but the differences exist. There is something about the 7d images that don't look right. Unlike the 5d or 1dmkII images, the 7d produces images with abrupt highlight transitions and which demonstrate highlight clipping. The tonal transitions and highlights appear better managed in the 1dmkII and 5d images when converted from RAW. Again, this may be a software issue rather than a sensor DR or RAW headroom deficiency.

You seem to be talking about highlight headroom, rather than DR. The 7D is a little short in highlight headroom at ISO 100, because the RAW data clips at a lower level than it does at ISO 200.

Doesn't that indicate that the base ISO is higher than ISO 100?

Yes; Canons have been that way for a while now.

ISOs 160, 320, 640, 1250, and 2500 may be a little short, too, because they are pulled from higher ISOs, and their RAW values are scaled down. And there may be other factors, but headroom and DR are two different things.

So, then ISO 125, 250, etc. must have more headroom, because they are pushed from lower ISOs! (?) But don't you usually say that it's ISO 125, 250, etc. that should be avoided? (because of relatively high read noise?)

They are pushed, so they have less headroom; some of the highlight headroom from the ISO they were pushed from (100, 200, etc) has been thrown away by the push.

Maybe (or probably) you're right, but this is really a bit confusing! ;-) John said that ISO 160, 320, etc. have less headroom because they are pulled from higher ISOs, and you say that ISO 125, 250, etc. have less headroom because they are pushed from lower ISOs! You can't both be right, can you? (Or maybe.. the ISO 125, 250, etc. RAW values are scaled/pushed 1/3 stop (from ISO 100, 200..), but they are also exposed 1/3 stop less, so isn't the headroom actually the same?)

 

[ejmartin]

What I am using as reference are images shot under similar lighting with similar settings processed in DPP. Now, perhaps the 7d RAW converter needs to be optimized, but the differences exist. There is something about the 7d images that don't look right. Unlike the 5d or 1dmkII images, the 7d produces images with abrupt highlight transitions and which demonstrate highlight clipping. The tonal transitions and highlights appear better managed in the 1dmkII and 5d images when converted from RAW. Again, this may be a software issue rather than a sensor DR or RAW headroom deficiency.

You seem to be talking about highlight headroom, rather than DR. The 7D is a little short in highlight headroom at ISO 100, because the RAW data clips at a lower level than it does at ISO 200.

Doesn't that indicate that the base ISO is higher than ISO 100?

Yes; Canons have been that way for a while now.

ISOs 160, 320, 640, 1250, and 2500 may be a little short, too, because they are pulled from higher ISOs, and their RAW values are scaled down. And there may be other factors, but headroom and DR are two different things.

So, then ISO 125, 250, etc. must have more headroom, because they are pushed from lower ISOs! (?) But don't you usually say that it's ISO 125, 250, etc. that should be avoided? (because of relatively high read noise?)

They are pushed, so they have less headroom; some of the highlight headroom from the ISO they were pushed from (100, 200, etc) has been thrown away by the push.

Maybe (or probably) you're right, but this is really a bit confusing! ;-) John said that ISO 160, 320, etc. have less headroom because they are pulled from higher ISOs, and you say that ISO 125, 250, etc. have less headroom because they are pushed from lower ISOs! You can't both be right, can you? (Or maybe.. the ISO 125, 250, etc. RAW values are scaled/pushed 1/3 stop (from ISO 100, 200..), but they are also exposed 1/3 stop less, so isn't the headroom actually the same?)

Actually, we're both right. The 160-320-etc series have less headroom than a hardware ISO 160 etc, since they are pulled from ISO 200 etc; they meter for 160 and expose at 200, so it is ETTR and the lower headroom that this entails, relative to a hardware ISO 160. I suspect this is what John was referring to.

The 125-250-etc series are pushed from hardware ISO 100-200-etc. My point was that since the camera only does hardware ISO 100-200-etc, then you would have more headroom shooting that ISO in RAW and underexposing by 1/3 stop, instead of the camera doing that and throwing away the top 1/3 stop of highlights (since the camera can only record up to 16383 in the RAW data, anything that gets pushed past that by the faux ISO boost is set to clipping; if you shoot at the lower ISO with -1/3 stop EC then you keep those levels at your disposal). Then you have the choice of doing a straight +1/3 stop EC in the RAW converter, as the camera would have done for you had you shot at ISO 125 etc; or you can apply a tone curve and recover the 1/3 stop of highlights that you kept by underexposing at the next lower ISO 100 etc. So my point was, why have the camera throw away the highlight data rather than giving yourself the choice whether to do that later on in PP?

--
emil
--



http://theory.uchicago.edu/~ejm/pix/20d/

 

[Steen Bay]

What I am using as reference are images shot under similar lighting with similar settings processed in DPP. Now, perhaps the 7d RAW converter needs to be optimized, but the differences exist. There is something about the 7d images that don't look right. Unlike the 5d or 1dmkII images, the 7d produces images with abrupt highlight transitions and which demonstrate highlight clipping. The tonal transitions and highlights appear better managed in the 1dmkII and 5d images when converted from RAW. Again, this may be a software issue rather than a sensor DR or RAW headroom deficiency.

You seem to be talking about highlight headroom, rather than DR. The 7D is a little short in highlight headroom at ISO 100, because the RAW data clips at a lower level than it does at ISO 200.

Doesn't that indicate that the base ISO is higher than ISO 100?

Yes; Canons have been that way for a while now.

ISOs 160, 320, 640, 1250, and 2500 may be a little short, too, because they are pulled from higher ISOs, and their RAW values are scaled down. And there may be other factors, but headroom and DR are two different things.

So, then ISO 125, 250, etc. must have more headroom, because they are pushed from lower ISOs! (?) But don't you usually say that it's ISO 125, 250, etc. that should be avoided? (because of relatively high read noise?)

They are pushed, so they have less headroom; some of the highlight headroom from the ISO they were pushed from (100, 200, etc) has been thrown away by the push.

Maybe (or probably) you're right, but this is really a bit confusing! ;-) John said that ISO 160, 320, etc. have less headroom because they are pulled from higher ISOs, and you say that ISO 125, 250, etc. have less headroom because they are pushed from lower ISOs! You can't both be right, can you? (Or maybe.. the ISO 125, 250, etc. RAW values are scaled/pushed 1/3 stop (from ISO 100, 200..), but they are also exposed 1/3 stop less, so isn't the headroom actually the same?)

Actually, we're both right. The 160-320-etc series have less headroom than a hardware ISO 160 etc, since they are pulled from ISO 200 etc; they meter for 160 and expose at 200, so it is ETTR and the lower headroom that this entails, relative to a hardware ISO 160. I suspect this is what John was referring to.

The 125-250-etc series are pushed from hardware ISO 100-200-etc. My point was that since the camera only does hardware ISO 100-200-etc, then you would have more headroom shooting that ISO in RAW and underexposing by 1/3 stop, instead of the camera doing that and throwing away the top 1/3 stop of highlights (since the camera can only record up to 16383 in the RAW data, anything that gets pushed past that by the faux ISO boost is set to clipping; if you shoot at the lower ISO with -1/3 stop EC then you keep those levels at your disposal). Then you have the choice of doing a straight +1/3 stop EC in the RAW converter, as the camera would have done for you had you shot at ISO 125 etc; or you can apply a tone curve and recover the 1/3 stop of highlights that you kept by underexposing at the next lower ISO 100 etc. So my point was, why have the camera throw away the highlight data rather than giving yourself the choice whether to do that later on in PP?

Yes, that seems a bit silly. Why isn't this 16.383 value always scaled with the ISO, so that it's reflecting the sensors actual clipping point? 32.767 at ISO 200, 65,535 at ISO 400, etc. That way ISO 800 would have 3 stops more headroom than ISO 100, which could be useful sometimes.(Or implemented in another way. Should be possible. I don't think that more bits are needed)

 

[ejmartin]

Yes, that seems a bit silly. Why isn't this 16.383 value always scaled with the ISO, so that it's reflecting the sensors actual clipping point? 32.767 at ISO 200, 65,535 at ISO 400, etc. That way ISO 800 would have 3 stops more headroom than ISO 100, which could be useful sometimes.(Or implemented in another way. Should be possible. I don't think that more bits are needed)

16383=2^14 - 1 is the largest 14-bit integer: 11111111111111 in base 2 notation. So you can't have 32767 without adding an extra bit. The sensible way to do things is to implement the ISO for the 125-250-500 etc series as metadata -- an instruction to the converter as to how to work with the RAW data. HTP works this way if you shoot RAW -- the metering is done for ISO X but the actual hardware gain used is X/2. Also, the extended ISO's should be done as metadata, on all Canons, because all that is done is to take the highest hardware amplification and throw away some number of stops of highlight headroom. But, the team there doesn't seem to have thought carefully about what they're doing with their firmware.

BTW, the reason that ISO 200-400-800 etc is not done this way is that the ADC has less DR than the sensor, so the utility of the hardware ISO is to find the best match of the DR window of the ADC to the range of recorded data from the sensor that one wants to fit in that window. Otherwise, if the ADC had as much DR as the sensor electronics (so that the noise floor of the ADC were as low or lower than the noise floor of the sensor), there would indeed be little need for hardware ISO, and all ISO could be implemented as metadata, since the hardware gain would do the same thing as software gain -- push more data off the top end of the ADC output, without gaining anything in shadows. But since the ADC isn't clean enough, it pays at lower ISO to amplify the signal from the sensor to get the sensor data above the noise floor of the ADC.

--
emil
--



http://theory.uchicago.edu/~ejm/pix/20d/

 

[Steen Bay]

Yes, that seems a bit silly. Why isn't this 16.383 value always scaled with the ISO, so that it's reflecting the sensors actual clipping point? 32.767 at ISO 200, 65,535 at ISO 400, etc. That way ISO 800 would have 3 stops more headroom than ISO 100, which could be useful sometimes.(Or implemented in another way. Should be possible. I don't think that more bits are needed)

16383=2^14 - 1 is the largest 14-bit integer: 11111111111111 in base 2 notation. So you can't have 32767 without adding an extra bit. The sensible way to do things is to implement the ISO for the 125-250-500 etc series as metadata -- an instruction to the converter as to how to work with the RAW data. HTP works this way if you shoot RAW -- the metering is done for ISO X but the actual hardware gain used is X/2. Also, the extended ISO's should be done as metadata, on all Canons, because all that is done is to take the highest hardware amplification and throw away some number of stops of highlight headroom. But, the team there doesn't seem to have thought carefully about what they're doing with their firmware.

BTW, the reason that ISO 200-400-800 etc is not done this way is that the ADC has less DR than the sensor, so the utility of the hardware ISO is to find the best match of the DR window of the ADC to the range of recorded data from the sensor that one wants to fit in that window. Otherwise, if the ADC had as much DR as the sensor electronics (so that the noise floor of the ADC were as low or lower than the noise floor of the sensor), there would indeed be little need for hardware ISO, and all ISO could be implemented as metadata, since the hardware gain would do the same thing as software gain -- push more data off the top end of the ADC output, without gaining anything in shadows. But since the ADC isn't clean enough, it pays at lower ISO to amplify the signal from the sensor to get the sensor data above the noise floor of the ADC.

Isn't the D3x read noise so low that the ISOs just as well (or almost) could have been implemented that way, as metadata?. So maybe we'll get something like that, in a not all too far future. (And maybe even in a Canon camera! ;-))

 

[pixelpepper]

Yes, that seems a bit silly. Why isn't this 16.383 value always scaled with the ISO, so that it's reflecting the sensors actual clipping point? 32.767 at ISO 200, 65,535 at ISO 400, etc. That way ISO 800 would have 3 stops more headroom than ISO 100, which could be useful sometimes.(Or implemented in another way. Should be possible. I don't think that more bits are needed)

16383=2^14 - 1 is the largest 14-bit integer: 11111111111111 in base 2 notation. So you can't have 32767 without adding an extra bit. The sensible way to do things is to implement the ISO for the 125-250-500 etc series as metadata -- an instruction to the converter as to how to work with the RAW data. HTP works this way if you shoot RAW -- the metering is done for ISO X but the actual hardware gain used is X/2. Also, the extended ISO's should be done as metadata, on all Canons, because all that is done is to take the highest hardware amplification and throw away some number of stops of highlight headroom. But, the team there doesn't seem to have thought carefully about what they're doing with their firmware.

BTW, the reason that ISO 200-400-800 etc is not done this way is that the ADC has less DR than the sensor, so the utility of the hardware ISO is to find the best match of the DR window of the ADC to the range of recorded data from the sensor that one wants to fit in that window. Otherwise, if the ADC had as much DR as the sensor electronics (so that the noise floor of the ADC were as low or lower than the noise floor of the sensor), there would indeed be little need for hardware ISO, and all ISO could be implemented as metadata, since the hardware gain would do the same thing as software gain -- push more data off the top end of the ADC output, without gaining anything in shadows. But since the ADC isn't clean enough, it pays at lower ISO to amplify the signal from the sensor to get the sensor data above the noise floor of the ADC.

Isn't the D3x read noise so low that the ISOs just as well (or almost) could have been implemented that way, as metadata?.

Nearly, but not quite. The Sony column ADC system seems to have about 50% more read noise at low ISO than at high ISO (which compares with 10x or more in ost conventional systems). My suspicion would be that the ADC's are good enough to operate without a variable gain amplifier, but not big enough. Given that they are essentially counters, making them bigger would mean either raising the clock frequency of the chip, which might not be possible, or slowing down the capture rate, which might not be marketable.

So maybe we'll get something like that, in a not all too far future. (And maybe even in a Canon camera! ;-))

Pentax used to use a similar system. They had a 22 bit ADC, and selected the 12 bits for the raw file according to set ISO. John always said it produced low read noise. The analog front end supplier they used has gone out of business. I think they all use Analog Devices or Texas Instruments front ends now, which have a very similar design and very similar characteristics - except, of course, the Sony sensors with on-chip ADC's.

 

[GaborSch]

The Sony column ADC system seems to have about 50% more read noise at low ISO than at high ISO (which compares with 10x or more in ost conventional systems)

Where are these numbers from? How were these measured?

--
Gabor

http://www.panopeeper.com/panorama/pano.htm

 

[John Sheehy]

Past comments and current impressions are made from past cameras. Based on the numbers I just showed, however, it seems Canon may be clipping more headroom in the 125/250 group than previously (they used the full values up to 16383 before subtracting blackpoint), and now the 160/320 group have more DR (but less headroom) than the "main" ISOs. I gotta go now, I'll look into it more when I get home later.

I just did the math, and it seems 200 and 160 have exactly the same DR, so no change there.

--
John

 

[John Sheehy]

Also, the extended ISO's should be done as metadata, on all Canons, because all that is done is to take the highest hardware amplification and throw away some number of stops of highlight headroom. But, the team there doesn't seem to have thought carefully about what they're doing with their firmware.

Not only would you have more headroom if the extended ISOs were all done with the normal digitization of the highest gain, but the files would be smaller; much smaller in some cases.

It is ironic (maybe sadistic!) that HTP is disabled above ISO 3200; that is where it makes the most sense, because typical low-light situations such as interiors and streets at night, tend to have light sources within the frame, and they lose their detail and color with the unnecessary clipping.

--
John

 

[John Sheehy]

The Sony column ADC system seems to have about 50% more read noise at low ISO than at high ISO (which compares with 10x or more in ost conventional systems)

Where are these numbers from? How were these measured?

According to DXOmark's DR, the D3X is pretty linear above 400, and 200 has about 1/4 stop more read noise, and 100 about 1/2.

--
John