2336 replies to this topic

[ccs_hello]

BSI's full implementation is

back-thinned, back-side illumination.

 

The back side of the Silicon wafer is first thinned (means polished and thinned), then flipped around and Bayer-filter applied/Microlenses applied.

 

BSI in civilian applications is usually only used in tiny pitched pixel sensors.

Large pitched is not used due to ND filter would have to be conditionally used for day-light well-lighted shooting.

 

Remember, these camera firms are selling their imaging devices to general public.

 

P.S. Sammy, just like other "want to get in" late comers, has nothing to loose thus willing to be all-in.

 

Clear Skies!

 

ccs_hello

Edited by ccs_hello, 23 January 2015 - 08:06 AM.

[mmalik]

Large pitched is not used due to ND filter would have to be conditionally used for day-light well-lighted shooting.

 

 

...P.S. Sammy, just like other "want to get in" late comers, has nothing to loose thus willing to be all-in.

 

So it would make sense why a7S didn't go with BSI given large pitch? If yes, why didn't a7/A7II or a7R go with BSI? Curious to know... is it because they are full-frame after all? I wonder why NX1 and not Sony, among the late comers, jumped on BSI?

 

 

While this 'late coming' might be good for innovation, is it killing old rivals? Wouldn't this make old rivals join the competition and innovate? This market shakeup might be good in the end; your thoughts?

 

 

For those wondering what is an ND filter, it is neutral density filter that reduces or modifies the intensity of all wavelengths or colors of light equally, giving no changes in hue of color rendition. It can be a colorless (clear) or grey filter. The purpose of a standard photographic neutral-density filter is to reduce the amount of light entering the lens.

Edited by mmalik, 23 January 2015 - 01:47 PM.

[alan.dang]

 I wonder why NX1 and not Sony, among the late comers, jumped on BSI?

 

 

Cost.  R&D for the BSI isn't straightforward.  Sony is using BSI on their RX100-line.  

 

67 points

http://www.dxomark.c...t-DSC-RX100-III

 

vs. 

 

65 points for a T3i

http://www.dxomark.c.../Canon/EOS-600D

[mmalik]

Cost... high or low? Is BSI expansive or cheaper to make? Sony for example has BSI on the lower end (RX100 III), than high-end (a7II/a7S). Regards

[mpgxsvcd]

Canon knows that everyone here will still bash anyone who says Canon isn’t the greatest thing since sliced bread. If you guys keep buying it they will still keep making the same product with a different model number.

[fetoma]

Sony seems to like the technology, let's see if this chip makes it somewhere useful for us imagers. QHY or SX might bite if they make a larger version.

 

http://www.sony.net/...ring_Exmorr.pdf

[mpgxsvcd]

 Of course, if you want mirrorless, Sony is the only choice.

Or Olympus. Or Panasonic. Both have used the great Sony sensors in the past and the present.

[Jon Rista]

Cost... high or low? Is BSI expansive or cheaper to make? Sony for example has BSI on the lower end (RX100 III), than high-end (a7II/a7S). Regards

 

BSI has been more expensive...but costs are coming down. There are more risks to BSI. Your using both sides of the wafer, so that means there is twice the surface area within which defects can cost you a die. As ccs stated, the wafer is first "thinned", then more stuff is layered down on top of a thinner substrate. FSI (front-side illuminated) sensors tend to have a fairly thick silicon substrate which gives the sensor stability and rigidity. BSI sensors are more fragile, so even if you don't lose a die to a wafer defect, they can break during transport or integration. 

 

Those risks, and the increased die losses per wafer, tend to increase the cost of BSI devices. 

 

When sensors are small, like the 1/3" or smaller sized sensors (particularly the ultra tiny ones found in smartphones, tablets, other mobile consumer-grade devices), you can generate hundreds if not thousands per wafer, and their small size reduces the chance that the thinner substrate will result in fragility. A relatively large (gargantuan, compared to consumer device sensors) APS-C sized BSI sensor like the one in the Samsung NX1 carries with it a fairly high risk of either loss to wafer defects or loss to breaking. Samsung is a pretty advanced company, though, and they have a number of cutting edge patents for large-sensor BSI designs. They have techniques to strengthen the thinned substrate, they have ISOCELL technology (not used in the NX1, as far as I know, but it practically eliminates color crosstalk in Bayer type BSI sensors, which has been a problem in the past), etc. My guess is Samsung employed their substrate strengthening patents in their NX1 sensor, which is why they were able to create an APS-C sized one before everyone else. 

 

Canon actually has patents for FF layered (not bayer) BSI sensors. They have for a couple of years. They haven't employed them yet, and my guess is they haven't found a way to make sensors that large (2.6x greater area than an APS-C sensor) stable enough to be integrated into a consumer grade DSLR. 

 

Sony's BSI sensors are on the smaller side...a small fraction of the area of an APS-C sensor, for example. 

[mmalik]

A relatively large (gargantuan, compared to consumer device sensors) APS-C sized BSI sensor like the one in the Samsung NX1 carries with it a fairly high risk of either loss to wafer defects or loss to breaking.

 

Thanks Jon for the elaboration; I wonder how this BSI fragility plays into a mod (occasional/minor physical stress as filters are removed from the sensor)? Regards

[mmalik]

What do you make of these sensor overall scores (which are based on following criteria)?

 

• Sensor quality in terms of noise
• Ability to render high contrast
• Formation of colored noise
• Ability to shoot in low light

 

Note: What these scores do NOT include is resolution, i.e., its ability to render fine details

 

 

Data source: DXOMark

No such data available for NX1 yet...

Attached Thumbnails

• 

Edited by mmalik, 24 January 2015 - 07:30 PM.

[GJJim]

What do you make of these sensor overall scores (which are based on following criteria)?

 

• Sensor quality in terms of noise
• Ability to render high contrast
• Formation of colored noise
• Ability to shoot in low light

 

Note: What these scores do NOT include is resolution, i.e., its ability to render fine details

 

No such data available for NX1 yet...

 

Data source: DXOMark

I'd say canon needs to put the 6D sensor in a EOS-M body ASAP. 

[mmalik]

 

I'd say canon needs to put the 6D sensor in a EOS-M body ASAP. 

 

What that really means Canon needs to put out new and creative "performance" packages like Sony alpha 7 series; hope they do it sooner than later. Sony seems to have already cornered the market on that "6D sensor in an EOS-M body" concept. Regards

Edited by mmalik, 24 January 2015 - 07:02 PM.

[mmalik]

Going by the pixel size (and sensor score above...), one would think a7II and 6D performance should be comparable; your thoughts? Any other comparisons comparing a7II with 6D performance?

 

Attached Thumbnails

• 

Edited by mmalik, 24 January 2015 - 07:16 PM.

[mmalik]

A composite view...

 

 

Data source: DXOMark for sensor score

Attached Thumbnails

• 

Edited by mmalik, 26 January 2015 - 01:21 AM.

[GJJim]

^{On a value basis, the 6D is the clear winner for use in astrophotography. While Sony offers slightly better sensor specs and a slimmer, lighter body that requires less back focus, Canon counters with a much lower price and unmatched software support. As I said earlier, if Canon offered the features and sensor of the current 6D in a EOS-M size mirrorless body, they would remain very competitive with these Sony cameras.}

[bclaff]

Going by the pixel size (and sensor score above...), one would think a7II and 6D performance should be comparable; your thoughts? Any other comparisons comparing a7II with 6D performance?

 

For Photographic Dynamic Range (PDR, similar to DxOMark landscape print DR) they are close:

Although the A7II is much higher at low ISO that's overkill and not too relevant to AP AFAIK.

The unity ISO's are similar I have 488 for th 6D and 362 for the A7II.

The A7II goes "isoless" much earlier, around ISO 400, whereas for the 6D it's closer to ISO 3200.

 

This is based on files that I have analyzed.

I have temporarily fallen behind in processing new DxOMark data due to non-photographic priorities.

[Jon Rista]

Bill, have you had a change to generate your own results with an NX1 yet?

[mmalik]

The A7II goes "isoless" much earlier, around ISO 400, whereas for the 6D it's closer to ISO 3200.

 

Didn't make much of isoless comment, so I asked. Bill's response below, hope he wouldn't mind. Regards

 

 

 

"Most believers in "isoless"-ness feel that earlier is better than later.

If it's below the lowest ISO then the camera is deemed entirely "isoless" and some people look for that.

 

But IMO the whole "isoless" concept is false thinking.

 

For example, an "isoless" camera could have higher read noise than one that is not "isoless" at all; so blindly prizing a camera that is "isoless" does not make sense."

Edited by mmalik, 25 January 2015 - 01:22 PM.

[mmalik]

Point I am trying to make is this; a7S is great but pricey for most folks. Why not look at a7II as a viable alternative to a7S and/or 6D? Your thoughts?

[alan.dang]

Bill, have you had a change to generate your own results with an NX1 yet?

 

Bill and I are going to work together to get some numbers w/the NX1.

[mmalik]

Alan, not sure what to make of your... Read Noise (e/pixel) numbers when compared against sensorgen... data for your control (a7R), especially at lower ISO; your thoughts?

 

 

Note: Alan's data in [ ]

 

 

a7R ISO100: 5.3 [2.76]

a7R ISO1600: 2.3 [2.15]

[mmalik]

^{On a value basis, the 6D is the clear winner for use in astrophotography. While Sony offers slightly better sensor specs and a slimmer, lighter body that requires less back focus, Canon counters with a much lower price and unmatched software support.}

 

Note: a7II specific data is not readily available yet, so I am using a7 sensorgen... numbers for a7II in the following chart since they share the same sensor.


a7/a7II data is quite flat, which would make 'em 'isoless...' in terms of read noise; how do you read this comparison? Does it make the case for giving a7II an astro mod/test chance and/or a viable alternative for 6D?


Except for not having 5-axis stabilization (NOT needed for astro), a7 at $1,298... makes a strong case for itself (compared against 6D at $1,499...). Note: Pixel sizes of a7/a7II and 6D are quite close. Your thoughts?

 

 

Data source: sensorgen.info...

Attached Thumbnails

• 

Edited by mmalik, 26 January 2015 - 10:49 PM.

[GJJim]

That is an interesting comparison. I didn't know that Sony was offering such a large discount, it makes the A7 a much better value. The lack of software support in astro imaging apps is still a weakness they need to address. I have the original 5D where Canon never offered a decent SDK (likely due to the Digic II). Its level of software support is probably similar to the current situation vis à vis Sony. Long exposures on the 5D require the three-pin timer cable connection and since it lacks live view, focusing on night sky objects is difficult compared to cameras like the 6D. I can connect to the 6D via USB from many applications (TheSkyX, BYEOS, Maxim, etc.) and use the sophisticated tools they offer. With that in mind and until Sony can generate support from software developers, I'd still say the 6D is a better value for astrophotography.

[mmalik]

You'll be amazed what I value... a flip open LCD, being able to use a timer remote in bulb mode, being able to process RAWs in known packages like ImagesPlus, PixInsight, etc.

 

 

SDK argument is a crutch we'll need to throw out for little while; I am happy in a way Sony doesn't have or care much about the strings that might hold back developing a revolutionary DSLR. Canon needs to think the same way.

 

 

I for one don't use any of the software listed, for that matter I don't use PC for image capture at all. Guiding is a necessary evil for the time being. Regards

Edited by mmalik, 26 January 2015 - 10:44 PM.

[alan.dang]

Alan, not sure what to make of your... Read Noise (e/pixel) numbers when compared against sensorgen... data for your control (a7R), especially at lower ISO; your thoughts?

 

 

Note: Alan's data in [ ]

 

 

a7R ISO100: 5.3 [2.76]

a7R ISO1600: 2.3 [2.15]

 

1. Temperature may play a role

2. My read noise was calculated using ImagesPlus.  Two flats, Two bias shots.  14-bit computation.