Luminous Landscape Forum
Equipment & Techniques => Motion & Video => Topic started by: lowep on October 06, 2013, 07:28:43 pm
-
3MOS vs. Micro 4/3 vs. APS-C vs. cropped FF vs… ???
How do different sizes and types of sensors impact on dynamic range?
For example, what difference would you expect to see in dynamic range of video shot with a prosumer camcorder like the Panasonic AC90 that has 1/4.7 type 3MOS image sensor compared to the AF100 that has 4/3-type image sensor?
Or how does dynamic range of video recorded with a Panasonic GH3 that also has a Micro 4/3 sensor compare with the Sony A99 that has a full frame sensor but apparently crops the FF image down to ???? when image stabilizing is switched on in video mode?
No doubt this is like trying to compare Eva Peron and the Easter Bunny and also depends on a lot of other esoteric stuff like codec, bit rate, etc etc etc…. But at the end of the day blown highlights and blocked up shadows look the same whatever setup is used, right?
So does this mean the only way to figure out which one delivers the best dynamic range is to try everything?
-
For comparable sensor technology the bigger the sensor the greater the dynamic range captured.
Tony Jay
-
Sure but how about 3 small sensors (3CCD, 3MOS) compared to one bigger one (micro 4/3s)?
After all Gulliver didn't last long when he was attacked by all those Lilliputians :o
-
Hi,
Dynamic range is more limited by the recording format than by the sensor it self, I would think.
Best regards
Erik
3MOS vs. Micro 4/3 vs. APS-C vs. cropped FF vs… ???
How do different sizes and types of sensors impact on dynamic range?
For example, what difference would you expect to see in dynamic range of video shot with a prosumer camcorder like the Panasonic AC90 that has 1/4.7 type 3MOS image sensor compared to the AF100 that has 4/3-type image sensor?
Or how does dynamic range of video recorded with a Panasonic GH3 that also has a Micro 4/3 sensor compare with the Sony A99 that has a full frame sensor but apparently crops the FF image down to ???? when image stabilizing is switched on in video mode?
No doubt this is like trying to compare Eva Peron and the Easter Bunny and also depends on a lot of other esoteric stuff like codec, bit rate, etc etc etc…. But at the end of the day blown highlights and blocked up shadows look the same whatever setup is used, right?
So does this mean the only way to figure out which one delivers the best dynamic range is to try everything?
-
Sure but how about 3 small sensors (3CCD, 3MOS) compared to one bigger one (micro 4/3s)? After all Gulliver didn't last long when he was attacked by all those Lilliputians :o
Size trumps all.
Tony Jay
-
Erik, how about if the same recording format is used?
Tony, do you mean the size of each particular sensor or the total size of all three sensors in a 3CCD or 3MOS sensor array vs 1 micro 4/3 sensor?
Maybe in reality there is no very significant difference between the dynamic range of these cameras anyway unless you start comparing them to much more expensive ones?
-
Why does the D800 have more dynamic range than 5D MIII? Both cameras have same sensor size AND the D800 has smaller pixels. Looks like the people at Nikon are able to collect or use the information from the sensor allot better.
I think the trick is to know your equipment to ensure you are getting the best results. The best way is to set the image profiles up with the least amount of contrast, then grading the video, setting up a unique profile most suitable for each clip.
A friend of mine was editing footage shot with C300, he told me it could have been shot with a 550D. Not taking anything from the C300, but the profiles were poorly setup on this shoot and everything was shot @f/11. Just an example of how you can downgrade the performance of your tool by not knowing it.
-
Just an example of how you can downgrade the performance of your tool by not knowing it.
good point though in this case am just trying to figure out how the different tools can perform before selecting one of them...
Looks like the people at Nikon are able to collect or use the information from the sensor a lot better.
you mean they were born more intelligent or use a different approach that must be explained somewhere in the realms of technical papers they produce to convince sitting ducks like me to buy their stuff. Maybe other factors have such a significant influence on dynamic range that considering sensor type is quacking up the wrong tree?
perhaps the true explanation has something to do with critical flicker-fusion frequency (CFF) that I read about in The Economist this morning: “It measures.. how fast an animal’s eyes can refresh an image and thus process information.
For people, the average CFF is 60 hertz (ie, 60 times a second). This is why the refresh rate on a television screen is usually set at that value. Dogs have a CFF of 80HZ, which is probably why they do not seem to like watching television. To a dog a TV programme looks like a series of rapidly changing stills.”
Maybe this is why some prefer Micro 4/3 and others prefer 3CCD? :)
-
Just an example of how you can downgrade the performance of your tool by not knowing it.
good point though in this case am just trying to figure out how the different tools can perform before selecting one of them...
Looks like the people at Nikon are able to collect or use the information from the sensor a lot better.
you mean they were born more intelligent or use a different approach that must be explained somewhere in the realms of technical papers they produce to convince sitting ducks like me to buy their stuff. Maybe other factors have such a significant influence on dynamic range that considering sensor type is quacking up the wrong tree?
perhaps the true explanation has something to do with critical flicker-fusion frequency (CFF) that I read about in The Economist this morning: “It measures.. how fast an animal’s eyes can refresh an image and thus process information.
For people, the average CFF is 60 hertz (ie, 60 times a second). This is why the refresh rate on a television screen is usually set at that value. Dogs have a CFF of 80HZ, which is probably why they do not seem to like watching television. To a dog a TV programme looks like a series of rapidly changing stills.”
Maybe this is why some prefer Micro 4/3 and others prefer 3CCD? :)
Sorry if my post rubbed you the wrong way, not my intention.
Nikon has in the past used Sony sensors, they still are if I am not mistaken. Sony has not yet managed to get as much dynamic range out of their own line of cameras. A sensor is an essential part of the equation, if you prefer to focus on the technical engineer side, I understand that. I simply pointed out the other side of the story, easy things you can do your self with settings and in field usage.
-
on the contrary i appreciate what you had to say... it is me who ought to apologise for not containing my enthusiasm
-
perhaps the true explanation has something to do with critical flicker-fusion frequency (CFF) that I read about in The Economist this morning: “It measures.. how fast an animal’s eyes can refresh an image and thus process information.
For people, the average CFF is 60 hertz (ie, 60 times a second). This is why the refresh rate on a television screen is usually set at that value. Dogs have a CFF of 80HZ, which is probably why they do not seem to like watching television. To a dog a TV programme looks like a series of rapidly changing stills.”
Maybe this is why some prefer Micro 4/3 and others prefer 3CCD? :)
Most tvs where I live are LCD with no flicker. They do have jumpy movement, though.
Oh, and the dog loves watching tv.
-h
-
Still wasting my breaks from editing trying to figure this out came across this (http://hddv.net/archive/index.php/t-20329.html) old post by hazydave in the archives of HDVDV forum that I found more relevant but perhaps not as interesting as what was in The Economist:
The potential advantage to a 3-chipper in the old days was simple: three actual color samples per pixel rather than one. But our eyes really don't see color well enough to worry all THAT much about interpolation in HD. So the real answer is light collecting ... three 1/4.1" sensors versues one 1/2.6" sensor. Both aggregate an effective three sensor pixels per video pixel, so the question: how big is a pixel. Back of the napkin math here says a single 8.5MPixel 1/2.6" sensor has about 21.7mm^2 of imaging area, which gives you 2.5nm^2 per pixel. Each 3MPixel 1/4.1" sensor has about 7.3mm^2 of imaging area, which gives you 2.4nm^2 per pixel.
So physically speaking, they're basically equivalent... any differences ought to be in actual implementation -- how good is the CMOS sensor technology relative to noise, how much of that maximum possible pixel area is actually collecting light (eg, size of photodiode, size of microlens, etc).
So looks like all of you are right :-)
-
Hi,
It is pretty much about the number of pixels utilized. Are you using all pixels in the image and downsampling using a decent algorithm or are you just using each third pixel on a high resolution sensor. The first option is preferable but the second is easy to do.
Best regards
Erik
Erik, how about if the same recording format is used?
Tony, do you mean the size of each particular sensor or the total size of all three sensors in a 3CCD or 3MOS sensor array vs 1 micro 4/3 sensor?
Maybe in reality there is no very significant difference between the dynamic range of these cameras anyway unless you start comparing them to much more expensive ones?
-
are you just using each third pixel on a high resolution sensor
now why would anybody want to do that - unless there were so many pixels on the sensor for other reasons (like shooting stills with the D800)?
-
Comparing the Canon 5D III and the Nikon D800 is comparing apples and oranges as far as I am concerned.
You will note that I used a phrase "comparable sensor technology."
The Nikon and the Canon are not directly comparable because they use different sensor technology.
The fact that the overall sensor size is the same is neither here nor there.
If one wants to figure out the effect of sensor size on dynamic range in particular then compare sensors that are the same apart from their overall size.
It is s simple fact of physics that in that circumstance the larger sensor will capture a larger dynamic range.
With regard to the sensor in the Nikon D800 it is made by Sony and its abilities are extraordinary and represent a big advance over sensors currently found in Canon cameras.
BTW I use Canon equipment and my first choice camera currently is a 5D III.
Tony Jay
-
Looks like the people at Nikon are able to collect or use the information from the sensor a lot better.
With regard to the sensor in the Nikon D800 it is made by Sony and its abilities are extraordinary and represent a big advance over sensors currently found in Canon cameras.
It is probably quite complicated but I am curious anyway about what is the technological explanation for the extraordinary abilities and big advance over sensors in other cameras?
-
Looks like the people at Nikon are able to collect or use the information from the sensor a lot better.
With regard to the sensor in the Nikon D800 it is made by Sony and its abilities are extraordinary and represent a big advance over sensors currently found in Canon cameras.
It is probably quite complicated but I am curious anyway about what is the technological explanation for the extraordinary abilities and big advance over sensors in other cameras?
The big difference here is not "Nikon vs the rest"; it is that the Nikon-Sony sensors use the newer approach of doing analog-to-digital conversion on the sensor, with an ADC unit at the bottom of each column of pixels (as does almost every other sensor maker these days) but Canon is still doing things the older way, transporting the analog signal all the way off the sensor and then doing ADC with just a few ADC units. The newer "column parallel ADC" approach means that the analog signal travels a far shorter distance before digital conversion [helping to reduce electronic noise] and with thousands of ADC units rather than just a few, each unit can operate at a far lower frequency, which also seems to help reduce noise introduced in the ADC process.
Note that this is not at all unique to Nikon, or to Sony: Panasonic has been using column-parallel ADC for several years starting with the GH1, Samsung had this even earlier at least in video sensors, Aptina (the company designing the sensors for Nikon One) also uses it, as does CMOSIS in the sensor for the new Leica M, and Olympus now uses such sensors too, though who makes them is still in some dispute. So the sole major outlier still using off-board ADC is Canon (along with the larger-than-35mm camera sensors which all still use the even older CCD technology).
-
Hi,
There may be other differences:
D800 sensor by Sony
(http://www.chipworks.com/blog/technologyblog/files/2012/10/Figure_1a.jpg)
D4 Sensor:
(http://www.chipworks.com/blog/technologyblog/files/2012/10/Figure_2a-300x223.jpg)
1DX sensor:
(http://www.chipworks.com/blog/technologyblog/files/2012/10/dlsr-2-fig1a-300x223.jpg)
Images from these articles by Chipworks:
http://www.chipworks.com/en/technical-competitive-analysis/resources/blog/full-frame-dslr-cameras-part-1-nikon-vs-sony/
http://www.chipworks.com/en/technical-competitive-analysis/resources/blog/full-frame-dslr-cameras-canon-stays-the-course/
Best regards
Erik
The big difference here is not "Nikon vs the rest"; it is that the Nikon-Sony sensors use the newer approach of doing analog-to-digital conversion on the sensor, with an ADC unit at the bottom of each column of pixels (as does almost every other sensor maker these days) but Canon is still doing things the older way, transporting the analog signal all the way off the sensor and then doing ADC with just a few ADC units. The newer "column parallel ADC" approach means that the analog signal travels a far shorter distance before digital conversion [helping to reduce electronic noise] and with thousands of ADC units rather than just a few, each unit can operate at a far lower frequency, which also seems to help reduce noise introduced in the ADC process.
Note that this is not at all unique to Nikon, or to Sony: Panasonic has been using column-parallel ADC for several years starting with the GH1, Samsung had this even earlier at least in video sensors, Aptina (the company designing the sensors for Nikon One) also uses it, as does CMOSIS in the sensor for the new Leica M, and Olympus now uses such sensors too, though who makes them is still in some dispute. So the sole major outlier still using off-board ADC is Canon (along with the larger-than-35mm camera sensors which all still use the even older CCD technology).
-
Very interesting BJL and Erik, I was not remotely familiar with the technical details behind the results!
I really hope Canon adopts this clear step forward. Today I am heavily invested in Canon, if the next generation does not improve and catch up, I may need to seriously reconsider my gear investments.
-
Very interesting BJL and Erik, I was not remotely familiar with the technical details behind the results!
+1! :)
-
Hi,
According to the Chipworks article all Canon CMOS sensors are built to 0.5 my rules, but Canons is said to build a 0.18 micron fab. With 0.18 micron they could make features smaller, thus increase fill ratio and still increase gate count. That combination my make it attractive to integrate more functionality on chip.
The advantages of the Sony design are mainly better readout noise. It helps with DR at minimum ISO. Canon's main focus seems not to be maximum DR at minimum ISO. Another way to put it. If you have problems of noisy shadows on you Canon, go look somewhere else, but if you don't have problems just be happy. Canons DR may not be the best but it may be good enough. If you shoot at higher ISOs, the Canons are not second to anyone.
Best regards
Erik
Very interesting BJL and Erik, I was not remotely familiar with the technical details behind the results!
I really hope Canon adopts this clear step forward. Today I am heavily invested in Canon, if the next generation does not improve and catch up, I may need to seriously reconsider my gear investments.