Jump to content

Why are some sensors a perfect 3x2 and others not?


Stef63

Recommended Posts

Advertisement (gone after registration)

I noticed that the M10 sensor has a resolution that is a perfect 3 by 2.  As well as the SL2-S sensor.  That is not the case for the M9, M240, M10-R, M11 and SL2 both DNG and JPG sizes. Furthermore some pixels always get lost in the translation from DNG to JPG.  For some camera models this reduction is in a 3x2 ratio, sometimes it's a different ratio vertical versus horizontal.

So some questions for more knowledgeable forums members :

1) why are not all sensors perfect 3 by 2 or why is the resolution not always a perfect 3 by 2? It seems to me that the physical sensor could be modified by disabling pixels at the edges to obtain that 3x2 size. 

2) what could be the side effect opening these files at 100% in a software (like Lightroom) that might "think" in the 3 by 2 ratio?

2) why are DNG files always some pixels larger on each side than the high res JPG's?

Thanks to enlighten me 🤔

Link to post
Share on other sites

1) If it bothers you, you can always crop your pictures to the  3:2 ratio or, indeed, any ratio you desire.

2) There is no software which 'thinks' in any particular ratio. Every software known to me (possibly known to man, who knows) merely consults the information contained in the image file and uses the image as given.

2) A DNG file of any Leica camera does not actually contain pixels. They call them 'pixels' just for convenience. A JPEG file, on the other hand, consists of pixels. When calculating the contents of each pixel for the JPEG file, a number of image elements surrounding each pixel is taken into consideration. Since the pixels at the very border of the image do not have surrounding elements on all sides, their values can not be calculated this way. Hence, they are simply discarded.

  • Like 2
  • Thanks 1
Link to post
Share on other sites

On 1/8/2022 at 4:20 AM, Stef63 said:

I noticed that the M10 sensor has a resolution that is a perfect 3 by 2.  As well as the SL2-S sensor.  That is not the case for the M9, M240, M10-R, M11 and SL2 both DNG and JPG sizes. Furthermore some pixels always get lost in the translation from DNG to JPG.  For some camera models this reduction is in a 3x2 ratio, sometimes it's a different ratio vertical versus horizontal.

So some questions for more knowledgeable forums members :

1) why are not all sensors perfect 3 by 2 or why is the resolution not always a perfect 3 by 2? It seems to me that the physical sensor could be modified by disabling pixels at the edges to obtain that 3x2 size. 

2) what could be the side effect opening these files at 100% in a software (like Lightroom) that might "think" in the 3 by 2 ratio?

2) why are DNG files always some pixels larger on each side than the high res JPG's?

Thanks to enlighten me 🤔

I may not fully understand your questions, but i believe that ALL full frame sensors are a Natural 3:2 Aspect Ratio being 36mm x 24mm.
Jpeg image settings not withstanding, all DNG files coming out of the camera would thereby be in a 3:2 ratio and would open in that format
in Editing software as far as i know.
Some Leica cameras like the SL2 i have and Q2 allow images to be formatted (cropped) in other Aspect Ratios "in-camera," but these settings ONLY affect the JPEG files.


 

 

Link to post
Share on other sites

6 hours ago, roverover said:

I may not fully understand your questions, but i believe that ALL full frame sensors are a Natural 3:2 Aspect Ratio being 36mm x 24mm.
Jpeg image settings not withstanding, all DNG files coming out of the camera would thereby be in a 3:2 ratio and would open in that format
in Editing software as far as i know.
Some Leica cameras like the SL2 i have and Q2 allow images to be formatted (cropped) in other Aspect Ratios "in-camera," but these settings ONLY affect the JPEG files.


 

 

  • M10 : 5976 x 3984 => 3 x 2
  • SL2-S 6000 x 4000 => 3 x 2
  • SL2 : 8368 x 5584 => 3 x 2,0021 (one less pixel horizontal and 6 pixels less vertical would be a perfect 3 x 2)
  • M9 : 5212 x 3468 => 3 x 1,9961 (one less pixel horizontal and 6 extra pixels vertical would be a perfect 3 x 2)

So some sensors are a Natural 3 x 2 Aspect Ration some are not.  And this data is not what Leica tells in their "technical data sheet" about the sensor, this is what I see in Adobe Lightroom as the size of a DNG coming out of the camera. 

As another example about the confusion of tech specs and real world:

  • the "M10 Technical Data" from the Leica website states the sensor produces DNG's
    • 5976 x 3992.  
  • DPreview specs the same sensor :
    • 5952 x 3992.  
  • I see a DNG coming out of my M10 :
    • 5976 x 3984.

My other question was why JPG (full res and not in camera cropped) seem to always be some pixels smaller than the corresponding DNG

  • M10 DNG 5976 x 3984 (according to Leica)
  • M10 JPEG 5952 x 3968 (according to Leica)

 

Link to post
Share on other sites

14 minutes ago, Stef63 said:
  • M10 : 5976 x 3984 => 3 x 2
  • SL2-S 6000 x 4000 => 3 x 2
  • SL2 : 8368 x 5584 => 3 x 2,0021 (one less pixel horizontal and 6 pixels less vertical would be a perfect 3 x 2)
  • M9 : 5212 x 3468 => 3 x 1,9961 (one less pixel horizontal and 6 extra pixels vertical would be a perfect 3 x 2)

So some sensors are a Natural 3 x 2 Aspect Ration some are not.  And this data is not what Leica tells in their "technical data sheet" about the sensor, this is what I see in Adobe Lightroom as the size of a DNG coming out of the camera. 

As another example about the confusion of tech specs and real world:

  • the "M10 Technical Data" from the Leica website states the sensor produces DNG's
    • 5976 x 3992.  
  • DPreview specs the same sensor :
    • 5952 x 3992.  
  • I see a DNG coming out of my M10 :
    • 5976 x 3984.

My other question was why JPG (full res and not in camera cropped) seem to always be some pixels smaller than the corresponding DNG

  • M10 DNG 5976 x 3984 (according to Leica)
  • M10 JPEG 5952 x 3968 (according to Leica)

 

The dimension that you lists are all crops as applied automatically by post-processing software (according to manufacturers spec in EXIF).

E.g., the real sensor dimension of SL2-S is 6048 x 4048 (also stored in EXIF).

Edited by SrMi
  • Like 1
Link to post
Share on other sites

Advertisement (gone after registration)

1 hour ago, Stef63 said:
  • M10 : 5976 x 3984 => 3 x 2
  • SL2-S 6000 x 4000 => 3 x 2
  • SL2 : 8368 x 5584 => 3 x 2,0021 (one less pixel horizontal and 6 pixels less vertical would be a perfect 3 x 2)
  • M9 : 5212 x 3468 => 3 x 1,9961 (one less pixel horizontal and 6 extra pixels vertical would be a perfect 3 x 2)

So some sensors are a Natural 3 x 2 Aspect Ration some are not.  And this data is not what Leica tells in their "technical data sheet" about the sensor, this is what I see in Adobe Lightroom as the size of a DNG coming out of the camera. 

As another example about the confusion of tech specs and real world:

  • the "M10 Technical Data" from the Leica website states the sensor produces DNG's
    • 5976 x 3992.  
  • DPreview specs the same sensor :
    • 5952 x 3992.  
  • I see a DNG coming out of my M10 :
    • 5976 x 3984.

My other question was why JPG (full res and not in camera cropped) seem to always be some pixels smaller than the corresponding DNG

  • M10 DNG 5976 x 3984 (according to Leica)
  • M10 JPEG 5952 x 3968 (according to Leica)

 

It's my understanding that when a Jpegs file is created (no matter where), the outer row of image pixels are removed. I believe the outer edge is not utilized and thrown away
because of how JPEGS are created. It's a Compression format and there is always some data loss AFAIK.
I can't tell you the technical reason that there are a "few" pixels differences in your examples, but I believe the image pixel counts you are stating are still considered a general 3 2 aspect
ratio even though a few pixels difference exists. I assume that Image editing software interprets the data when importing using it's own algorithms and is changing the pixel counts SLIGHTLY.
When I import a file into PS for example, the file size grows to about "3 times" it's out of camera MP size. For example, an out of camera 80 MP file grows to about 240 MP when
imported in PS as the software is now reading all 3 color channels separately and seeing much more data from the RAW out of camera DNG file than my computer sees.

I am not as knowledgable about the "science" behind all this as some here on the forum so I can't provide you with a "scientific answer, which I think is what you are looking for, 
but all "large format" cameras have a 3 : 2 image sensor in them and I believe the examples you stated are all considered a 3 : 2 format even though there are "minute" discrepancies in the pixel count.
 

 

Edited by roverover
Link to post
Share on other sites

One key point is that all sensors have a "border" of pixels that are not image-forming, but perform other duties. Such as masked pixels that never get exposed, and thus provide a "reference black point." Or are simply a safety  margin.

(The M8 had a problem with this. It used the same sensor layout as the DMR digital back for the R system, but tried to "cheat" around the edges to get the crop factor up from 1.3x to 1.333x (by making the mask window a tiny bit bigger). This lead to an erratic black reference point, if really bright light (e.g. the sun's image ) fell right on the edge of the mask, and leaked onto pixels it wasn't supposed to. The result would be a green band or streak across the picture, several pixels wide. )

Another point is that color Bayer patterns with only red, green and blue pixels create full-color images by combining data from all the surrounding pixels to figure out the "true total color" any one pixel in the image should have. But at the edge of the picture, half the neighboring pixels would be missing if there was a hard edge, and one would get a weird color line right at the edge. (Pop mentions this above in regard to jpgs vs. raw dimensions).

Put another way, there is not a strict 1:1 relationship between the pixels/photosites on the sensor, and the pixels in the image.

https://www.the-digital-picture.com/News/News-Post.aspx?News=20753

If one reads a sophisticated review of a camera and sensor, the specs will list total pixels (all those on the sensor silicon) and also effective pixels (those that actually end up in the photograph).

As to why some cameras are slightly out of the 2:3 ratio, another part of this is that there is no such thing as a fraction of a pixel. A 2x3 digital image 5000 pixels in the long dimension should be 3333.3333 pixels in the short dimension. But since one cannot have 0.3333 of a pixel, the image must be either 5000 x 3333, or 5000 x 3334.

Edited by adan
  • Like 3
Link to post
Share on other sites

Thank you Adan.  As I have by nature a curiosity about technology I always enjoy reading scientific or technical explanations why and how things are working, what is the reason behind a decision that a manufacturer takes etc. 

Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...