Luminous Landscape Forum
Equipment & Techniques => Medium Format / Film / Digital Backs – and Large Sensor Photography => Topic started by: ErikKaffehr on April 12, 2008, 11:13:18 am
-
Hi,
I don't really understand why MFDB seem to have low ISO sensivity like 50 ISO and maximum like 800 on the P30. DSLRs seem to get base ISO about 200 and can easily go up to 1600 ISO, possibly more.
Quantum efficiency should be by and large similar between both types of systems.
Someone who has some good explanation?
Best regards
Erik Kaffehr
-
The dSLRs have microlenses, which gains them one stop over the backs as far as sensitivity is concerned, and maybe the Bayer filters are a bit more orthogonal which means better colors but more noise under certain conditions. However my tests of a few backs indicate that they would be quite decent at higher ISO level if they were good samples. What kills them is not noise, it's various types of non-uniformity. For instance my first P45+ had horizontal black streaks and a centerfold issue, the replacement back has uneven zones and sometimes has vertical white streaks. It seems to be impossible to get a decent sample, I guess the back makers sell the cheapest quality sensors they can get away with.
Edmund
Hi,
I don't really understand why MFDB seem to have low ISO sensivity like 50 ISO and maximum like 800 on the P30. DSLRs seem to get base ISO about 200 and can easily go up to 1600 ISO, possibly more.
Quantum efficiency should be by and large similar between both types of systems.
Someone who has some good explanation?
Best regards
Erik Kaffehr
[a href=\"index.php?act=findpost&pid=188954\"][{POST_SNAPBACK}][/a]
-
Digital backs use CCD sensors - which generally have better performance traits at low ISO's compared to CMOS. CCD's require more energy, thus run hotter - so high ISO performance is not their strength. CMOS runs cooler and does better (generally) at the higher ISO's compared to their CCD counterparts.
-
Hi!
OK, I buy that. I guess that things add up.
No microlenses one stop
Running less cool
Perhaps some agressive noise reduction on CMOS?
Erik
Digital backs use CCD sensors - which generally have better performance traits at low ISO's compared to CMOS. CCD's require more energy, thus run hotter - so high ISO performance is not their strength. CMOS runs cooler and does better (generally) at the higher ISO's compared to their CCD counterparts.
[a href=\"index.php?act=findpost&pid=189029\"][{POST_SNAPBACK}][/a]
-
http://www.dalsa.com/markets/ccd_vs_cmos.asp (http://www.dalsa.com/markets/ccd_vs_cmos.asp)
&
http://www.shortcourses.com/sensors/ (http://www.shortcourses.com/sensors/)
&
http://en.wikipedia.org/wiki/Active_pixel_sensor (http://en.wikipedia.org/wiki/Active_pixel_sensor)
Thierry
-
Thanks for the links. They are quite interesting, do not answer my question really, however. Lot of good info anyway. Thanks a lot!
Erik
http://www.dalsa.com/markets/ccd_vs_cmos.asp (http://www.dalsa.com/markets/ccd_vs_cmos.asp)
&
http://www.shortcourses.com/sensors/ (http://www.shortcourses.com/sensors/)
&
http://en.wikipedia.org/wiki/Active_pixel_sensor (http://en.wikipedia.org/wiki/Active_pixel_sensor)
Thierry
[a href=\"index.php?act=findpost&pid=189150\"][{POST_SNAPBACK}][/a]
-
Hi!
OK, I buy that. I guess that things add up.
No microlenses one stop
Running less cool
Perhaps some agressive noise reduction on CMOS?
Erik
[a href=\"index.php?act=findpost&pid=189125\"][{POST_SNAPBACK}][/a]
Correct. The Canon CMOS design has some breakthrough NR features.
Edmund
-
CCD's require more energy, thus run hotter - so high ISO performance is not their strength.
[a href=\"index.php?act=findpost&pid=189029\"][{POST_SNAPBACK}][/a]
As far as I know, sensor power consumption is irrelevant to noise except with continuously operated sensors (Live View or video cameras.) Where CMOS has an advantage, when used well, is doing on-pixel pre-amplification including possible adjustment for fixed pattern noise.
At high ISO speeds, an active CMOS sensor amplifies the weak signal at each photosite, and this amplified signal travels accross the sensor to the A/D unit. With a CCD instead, the weak high ISO speed signal from each photosite is passively transmitted across the sensor before being pre-amplified and converted to digital.
During the travel from photosite to A/D converter, the stronger CMOS signal is less prone to noise arising in transit. This seems to help good CMOS (and NMOS) sensors at high ISO speeds.
I agree about other factors like micro-lenses, absent on most MF CCDs (the 31MP Kodak has micro-lenses.)
There might also be design trade-offs reflecting different priorities: medium format is perhaps overall less interested in high ISO speed operation, and far more interested in getting the possible results under more controlled conditions, meaning mostly low to moderate ISO, using a tripod, studio lights, flash if necessary to avoid high ISO speeds. Dynamic range at low ISO speeds has traditionally been higher for the Full Frame Transfer type CCD's used in medium format than for CMOS sensors.
Finally, Michael R. has also suggested (if I recall correctly) that the costs of changing to CMOS is more justifiable with the higher sales volumes of smaller formats, while in the very low volume market for larger sensors, cost minimization favor staying with the existing full frame transfer CCD technology, at least up till now.
-
Thanks a lot to everyone providing input!
As far as I understand it seems that there are a lot of factors adding up more than a single factor explaining everything.
Best regards
Erik
Hi,
I don't really understand why MFDB seem to have low ISO sensivity like 50 ISO and maximum like 800 on the P30. DSLRs seem to get base ISO about 200 and can easily go up to 1600 ISO, possibly more.
Quantum efficiency should be by and large similar between both types of systems.
Someone who has some good explanation?
Best regards
Erik Kaffehr
[a href=\"index.php?act=findpost&pid=188954\"][{POST_SNAPBACK}][/a]
-
The dSLRs have microlenses, which gains them one stop over the backs as far as sensitivity is concerned, and maybe the Bayer filters are a bit more orthogonal which means better colors but more noise under certain conditions. However my tests of a few backs indicate that they would be quite decent at higher ISO level if they were good samples. What kills them is not noise, it's various types of non-uniformity. For instance my first P45+ had horizontal black streaks and a centerfold issue, the replacement back has uneven zones and sometimes has vertical white streaks. It seems to be impossible to get a decent sample, I guess the back makers sell the cheapest quality sensors they can get away with.
Edmund
[a href=\"index.php?act=findpost&pid=189026\"][{POST_SNAPBACK}][/a]
So Edmund, did you dump your P45+ or are you still waiting a a decent sample?
db