Overclock.net › Forums › Consumer Electronics › Photography › Megapixel count vs. noise, binning, and ISO
New Posts  All Forums:Forum Nav:

Megapixel count vs. noise, binning, and ISO

post #1 of 17
Thread Starter 
So I was talking to my buddy, and a question came up: When you select a lower megapixel count for captured frames, what is it that the camera is doing? I have read that most detectors don't support on-chip/hardware binning, so my assumption would then be that the camera is employing software binning post-capture. Does anyone have info on this?

And the question is, could one reduce the megapixel count (which, if binning is actually employed, would increase signal-to-noise--since binning reduces noise), and could one improve ISO performance by megapixel reduction? I would think that this depends on how the megapixel value is being changed.

Anyone have ideas?
    
CPUMotherboardGraphicsRAM
i5-2500K Biostar TP67B+ XFX HD5750 1GB 2x4GB DDR3 Corsair 1600 
Hard DriveOSMonitorPower
60GB OCZ SSD, 2x160GB HDD RAID0, 500GB+500GB+1.5TB Windows 7 Ultimate 64-bit Samsung SyncMaster 930B Antec SmartPower 450w 
Case
Antec 900 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
i5-2500K Biostar TP67B+ XFX HD5750 1GB 2x4GB DDR3 Corsair 1600 
Hard DriveOSMonitorPower
60GB OCZ SSD, 2x160GB HDD RAID0, 500GB+500GB+1.5TB Windows 7 Ultimate 64-bit Samsung SyncMaster 930B Antec SmartPower 450w 
Case
Antec 900 
  hide details  
Reply
post #2 of 17
noise isnt from more megapixels, its from the physical size of the photo-sites. simply put the larger the photo-sites the lower the noise.

so by reducing MP count but not physically increasing the size of each photo-site, noise will be the same (possibly worst considering how the image is more "cropped")
Dell XPS15
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7-2630QM XPS L502 Geforce GT525M 1GB 6gb 
Hard DriveOS
640GB 7200RPM Windows 7 
  hide details  
Reply
Dell XPS15
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7-2630QM XPS L502 Geforce GT525M 1GB 6gb 
Hard DriveOS
640GB 7200RPM Windows 7 
  hide details  
Reply
post #3 of 17
Thread Starter 
Quote:
Originally Posted by mz-n10 View Post
noise isnt from more megapixels, its from the physical size of the photo-sites. simply put the larger the photo-sites the lower the noise.

so by reducing MP count but not physically increasing the size of each photo-site, noise will be the same (possibly worst considering how the image is more "cropped")
Oh, sorry I wasn't clear--noise originates from the detector (of course), and a reduction in noise will occur if megapixel count is reduced by the method of binning (summing then averaging adjacent pixels to form one new pixel, thereby reducing the overall pixel count). We were curious if (a) binning was used in reducing the overall pixel count when choosing a lower-res frame, and (b) would this manifest strongly enough (if at all) to give one an advantage in ISO (with lower noise, you could choose a higher ISO at a lower megapixel to achieve some sort of desired effect).

The biggest question is--how is the megapixel count actually reduced from the native pixel count of the sensor when you select a smaller picture resolution?

Thanks for the response!
    
CPUMotherboardGraphicsRAM
i5-2500K Biostar TP67B+ XFX HD5750 1GB 2x4GB DDR3 Corsair 1600 
Hard DriveOSMonitorPower
60GB OCZ SSD, 2x160GB HDD RAID0, 500GB+500GB+1.5TB Windows 7 Ultimate 64-bit Samsung SyncMaster 930B Antec SmartPower 450w 
Case
Antec 900 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
i5-2500K Biostar TP67B+ XFX HD5750 1GB 2x4GB DDR3 Corsair 1600 
Hard DriveOSMonitorPower
60GB OCZ SSD, 2x160GB HDD RAID0, 500GB+500GB+1.5TB Windows 7 Ultimate 64-bit Samsung SyncMaster 930B Antec SmartPower 450w 
Case
Antec 900 
  hide details  
Reply
post #4 of 17
I believe you answer might be here...

http://www.velocityreviews.com/forum...-pictures.html
2010 MBP 15"
(13 items)
 
  
CPUGraphicsRAMHard Drive
INTEL CORE i5 520M nVIDIA GEFORCE GT 330M 8GB DDR3 @ 1066MHz SEAGATE MOMENTUS XT 500GB 
Optical DriveMonitor
SUPERDRIVE 15.4" @ 1440x900 
  hide details  
Reply
2010 MBP 15"
(13 items)
 
  
CPUGraphicsRAMHard Drive
INTEL CORE i5 520M nVIDIA GEFORCE GT 330M 8GB DDR3 @ 1066MHz SEAGATE MOMENTUS XT 500GB 
Optical DriveMonitor
SUPERDRIVE 15.4" @ 1440x900 
  hide details  
Reply
post #5 of 17
Thread Starter 
Quote:
Originally Posted by BlankThis View Post
I believe you answer might be here...

http://www.velocityreviews.com/forum...-pictures.html
Thanks.

I think we came across that at one point--couldn't figure out if discussions of binning were accurate to cameras vs. just a discussion of concepts. It's confusing because we came across another source that said canon dslr camera detectors specifically don't support hardware binning. Maybe only software down-sampling is done, not specifically binning? I guess in that case, the pixel count reduction would have no effect on the noise.

My guess is, based on the various resolutions available, that binning isn't used, because the resolutions aren't being reduced by factors (or multiples) of 2 (and you can't bin a non-integer number of pixels). Probably just uses some quick, crappy downsampling algorithm that doesn't have much of an effect on noise.
    
CPUMotherboardGraphicsRAM
i5-2500K Biostar TP67B+ XFX HD5750 1GB 2x4GB DDR3 Corsair 1600 
Hard DriveOSMonitorPower
60GB OCZ SSD, 2x160GB HDD RAID0, 500GB+500GB+1.5TB Windows 7 Ultimate 64-bit Samsung SyncMaster 930B Antec SmartPower 450w 
Case
Antec 900 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
i5-2500K Biostar TP67B+ XFX HD5750 1GB 2x4GB DDR3 Corsair 1600 
Hard DriveOSMonitorPower
60GB OCZ SSD, 2x160GB HDD RAID0, 500GB+500GB+1.5TB Windows 7 Ultimate 64-bit Samsung SyncMaster 930B Antec SmartPower 450w 
Case
Antec 900 
  hide details  
Reply
post #6 of 17
I still don't know anyone who ever needs to shoot sRAW. Storage is cheap, buy more.
post #7 of 17
Quote:
Originally Posted by guyladouche View Post
Oh, sorry I wasn't clear--noise originates from the detector (of course), and a reduction in noise will occur if megapixel count is reduced by the method of binning (summing then averaging adjacent pixels to form one new pixel, thereby reducing the overall pixel count). We were curious if (a) binning was used in reducing the overall pixel count when choosing a lower-res frame, and (b) would this manifest strongly enough (if at all) to give one an advantage in ISO (with lower noise, you could choose a higher ISO at a lower megapixel to achieve some sort of desired effect).

The biggest question is--how is the megapixel count actually reduced from the native pixel count of the sensor when you select a smaller picture resolution?

Thanks for the response!
That still wouldn't reduce noise. As mz said, it's about the size of the photosites, comparable to buckets catching apples falling from a tree. The smaller the photosite, the more "light spillage" there will be into adjacent pixels at higher ISO's.
Brick Sh*thouse
(26 items)
 
   
CPUMotherboardGraphicsRAM
Intel i5 2520M Lenovo ThinkPad X220 Intel i5 Motherboard 04W2109 Intel HD3000 Crucial 4x2 GB DDR3 PC1333 
Hard DriveOptical DriveCoolingOS
Samsung 840 120GB SSD Nope Itty-Bitty Fans and Heatsinks Windows 10 Professional 64-bit 
MonitorKeyboardPowerCase
LG 12.5" 1366x768 IPS Lenovo Keyboard Lenovo 90W AC Adapter Lenovo X Series ThinkPad Chassis 
MouseMouse PadAudioOther
Logitech M510 Desk Conexant 1 Watt Stereo Speakers Lenovo TrackPoint 
Other
Lenovo TouchPad 
  hide details  
Reply
Brick Sh*thouse
(26 items)
 
   
CPUMotherboardGraphicsRAM
Intel i5 2520M Lenovo ThinkPad X220 Intel i5 Motherboard 04W2109 Intel HD3000 Crucial 4x2 GB DDR3 PC1333 
Hard DriveOptical DriveCoolingOS
Samsung 840 120GB SSD Nope Itty-Bitty Fans and Heatsinks Windows 10 Professional 64-bit 
MonitorKeyboardPowerCase
LG 12.5" 1366x768 IPS Lenovo Keyboard Lenovo 90W AC Adapter Lenovo X Series ThinkPad Chassis 
MouseMouse PadAudioOther
Logitech M510 Desk Conexant 1 Watt Stereo Speakers Lenovo TrackPoint 
Other
Lenovo TouchPad 
  hide details  
Reply
post #8 of 17
Thread Starter 
Quote:
Originally Posted by GoneTomorrow View Post
That still wouldn't reduce noise. As mz said, it's about the size of the photosites, comparable to buckets catching apples falling from a tree. The smaller the photosite, the more "light spillage" there will be into adjacent pixels at higher ISO's.
I'm not following why it wouldn't be the case. Let's simplify it--keep the ISO locked at, say 100. You have a detector that has some signal 's' and some noise 'n'. Keep the ISO the same. Pixel-bin (either hardware or software), you increase your signal additively, and your detector noise (read noise/light leakage, etc.), which is random in nature should be reduced through the binning (essentially averaging). The size of the buckets haven't changed, so you don't have to worry about extra light spillage--I'm not following why you wouldn't be able to achieve higher signal-to-noise?
    
CPUMotherboardGraphicsRAM
i5-2500K Biostar TP67B+ XFX HD5750 1GB 2x4GB DDR3 Corsair 1600 
Hard DriveOSMonitorPower
60GB OCZ SSD, 2x160GB HDD RAID0, 500GB+500GB+1.5TB Windows 7 Ultimate 64-bit Samsung SyncMaster 930B Antec SmartPower 450w 
Case
Antec 900 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
i5-2500K Biostar TP67B+ XFX HD5750 1GB 2x4GB DDR3 Corsair 1600 
Hard DriveOSMonitorPower
60GB OCZ SSD, 2x160GB HDD RAID0, 500GB+500GB+1.5TB Windows 7 Ultimate 64-bit Samsung SyncMaster 930B Antec SmartPower 450w 
Case
Antec 900 
  hide details  
Reply
post #9 of 17
Quote:
Originally Posted by guyladouche View Post
I'm not following why it wouldn't be the case. Let's simplify it--keep the ISO locked at, say 100. You have a detector that has some signal 's' and some noise 'n'. Keep the ISO the same. Pixel-bin (either hardware or software), you increase your signal additively, and your detector noise (read noise/light leakage, etc.), which is random in nature should be reduced through the binning (essentially averaging). The size of the buckets haven't changed, so you don't have to worry about extra light spillage--I'm not following why you wouldn't be able to achieve higher signal-to-noise?
http://lclevy.free.fr/cr2/#sraw
post #10 of 17
Quote:
Originally Posted by guyladouche View Post
I'm not following why it wouldn't be the case. Let's simplify it--keep the ISO locked at, say 100. You have a detector that has some signal 's' and some noise 'n'. Keep the ISO the same. Pixel-bin (either hardware or software), you increase your signal additively, and your detector noise (read noise/light leakage, etc.), which is random in nature should be reduced through the binning (essentially averaging). The size of the buckets haven't changed, so you don't have to worry about extra light spillage--I'm not following why you wouldn't be able to achieve higher signal-to-noise?
keep in mind that over every pixel, there is a filter (bayer filter).
Dell XPS15
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7-2630QM XPS L502 Geforce GT525M 1GB 6gb 
Hard DriveOS
640GB 7200RPM Windows 7 
  hide details  
Reply
Dell XPS15
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7-2630QM XPS L502 Geforce GT525M 1GB 6gb 
Hard DriveOS
640GB 7200RPM Windows 7 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Photography
Overclock.net › Forums › Consumer Electronics › Photography › Megapixel count vs. noise, binning, and ISO