New BSI stills sensor on the way?

L0n3Gr3yW0lf

Dad Pun Joke Master Over Nine Thousand Meme Lord
Joined
Jul 31, 2013
Messages
1,177
Location
UK
Real Name
Ovidiu
Did not read all the replies here as I wanted my brain droppings first before I lose my train of thoughts. I have seen that rumor last night.

A BSI sensor should have multiple advantages beyond just low light capabilities in terms of Image Quality ... but do take "this" \/ like listening to a sidewalk doomsday preacher on a lazy evening in L.A. :p
Backside Illuminated Sensor should clear up the photon detectors on the sensor for more light and sensitivity by moving some (not all of it) of the wiring (mainly copper) on the back side of the sensor. This would, in effect, give the photon detectors a faster gathering of light capability but NOT necessarily more quantities if there is no improvement in the area of how much photons it can hold, hence lower noise BUT not more dynamic range.
One additional effect is, potentially, better autofocus because more light getting on the sensor faster means the PDAF on sensor has more data to work with, so we could see better AF-C as 2ell, especially in low light. Since Olympus is still, currently and as far as I know, the only company that deploys crosstype PDAF on sensor more data could or should give greater accuracy and speed.
If you check the data sheet in the "leak" it says 120 frames per second at 20 MP as well as 60 and 30 FPS. IF Olympus can get stronger/faster processor that can churn that high amount of data we could see something like Capture 2.0 with possibly (this is even more speculative then an episode of Star Trek all about Faster the Light Travel) 30, 60 or 120 FPS ,electronic shutter only of course, with AF-C or AF-S, possibly no blackouts on the lower speeds, improved object detection and tracking ... something that may blow people's socks of in a E-M1 X Mark II.

I would personally love such a sensor and camera capability for wildlife and even portraits as winter can be quite dark (and full of terrors ... legs bricks on the floor at 5 AM between me and my coffee) in the Northen Hemisphere plus the English Weather and I am starting to become a believer of shaved heads :p
But even on smaller cameras, once the True Pic successor comes to handle that amount of data in a smaller body, could bring plenty of benefits for peeps to enjoy.

PS. Okayu, now that I read all the replies on the thread it's understandable that the sensor will not benefit everyone in a way that people can justify a new camera, by how they shoot or what they shoot or even why they shoot. Since Olympus is the only one using PDAF in M⁴/³ system the amount of stock they will buy for Sony is going to probably keep this sensor on the high-end high cost models for quite a while before it can trickle down, by that time there will be other alternative solutions, be it cheaper or better (rarely both at the same time). This is going to be more of a niche solution then the average "Superman" of M⁴/³ that the 16MP sensor was.
Faster data readout is even more essential that most people give credit to because it improves the users experience and daily drive of any cameras, 120 Hz screens can make the UI so much better that it can be a night and day difference but it requires 120 FPS worth of data to make it work, that's 120x20 MP=2.400 MP per second, on average of a 20 MB RAW files that makes it 2.4 GB per second of data to read, downsample to the EVF/back screen to display is no small feat ... especially if you want to stay with the competition or even up the ante. I would welcome with open heart and wallet a black-out free, 120 or even 240 FPS, 5 Million dots (or more) UI experience that could make subject tracking child's play.
And to AI, the more data you have the more accurate the prediction and recognition (more data will not give you more speed, quite the opposite), the more subjects you can recognise and become more versatile. We might actually see dedicated AI chips that can churn the sensors data separately the way Google Pixel and Apple iPhone has done. Which could free up the main CPUs in the cameras to dedicate towards other tasks and speed up the overall experience even more.
Well ... Next time on Star Trek: Mirrorless Evolution ... Picard realises it was Olympus all along :p
 
Last edited:

Hypilein

Mu-43 All-Pro
Joined
Mar 18, 2015
Messages
1,664
I'd rather a few more mp than any perceived advantages of bsi.

The thing is, with faster readout you will also get better hand held high rez, possibly to the point where it's even usable for portraits and wildlife of the slower variety.
 

BDR-529

Mu-43 Top Veteran
Joined
Jun 27, 2020
Messages
791
BSI claws back the photosensitive area you lost to having supporting electronics at the front of the sensor by moving it to the back. If the fill factor was already high on a FSI design, and gapless microlenses already do a great job at helping this, then BSI won’t gain you much at all.

BSI is not only about fill factor but it has two other benefits which come from basic geometry: light has to travel shorter path before reaching the photosensitive diode area and it can also arrive at wider angles.

This all leads to more sensitivity which has quite often been wasted in having more but smaller photosites while maintaining previous performance. Or in laymans terms having more megapixels without losing previous IQ. Only during the latest FF sensor generation all manufacturers have sort of agreed that 24MP is enought and concentrated on increasing the absolute photosite quality instead of their number.

Stacked sensor on the other hand will not increase sensitivity or IQ per se, just speed by simply integrating super-fast RAM storage in the sensor silicon itself which removed the previous bottleneck of "hey, we also need to move and store all this data somewhere before we can read more from the next photosite".

This of course does indirectly improve IQ compared to traditional designs when they try to reach same read speed. Traditional non-stacked sensors can reach high read speeds only by cropping just APS-C or 1:1 pixels area from the center of a FF sensor and/or dropping resolution to 8bits from 10/12bits. (and of course the same applies to MFT)

The bottom line is still the same. As long you only shoot relatively static and reasonably well lit scenes which dont have exteme contast and have no need for high read speed, there will be no visible difference worth mentioning. That being said, any scene beyond this confort zone will see anything from visible to dealbreaking IQ improvement with latest sensor technology.
 
Last edited:

Ross the fiddler

Mu-43 Hall of Famer
Joined
May 20, 2012
Messages
4,304
Location
Blue Mountains, NSW, Australia
Real Name
Ross
Ross, I've printed to A0(+) size using A4 size crops from my E-510 (10 MPx). Still no sign of 'noise' as routinely seen on very low resolution devices (high quality/resolution, 100% aRGB, IPS monitors ... ).

My R3880 has around 1,000 times (approx.) the resolution per square inch of my best monitor!
It's not the noise but the amount of detail that can be seen in a large print with high resolution images was what I was referring to. Ken Duncan likes to use his Phase One & still takes multiple frames to construct a panorama so when printed it is possible to walk up close to see the fine detail. He prides himself on those sorts of images (apart from the other key points he values).
 

John King

Member of SOFA
Joined
Apr 20, 2020
Messages
2,843
Location
Beaumaris, Melbourne, Australia
Real Name
John ...
It's not the noise but the amount of detail that can be seen in a large print with high resolution images was what I was referring to. Ken Duncan likes to use his Phase One & still takes multiple frames to construct a panorama so when printed it is possible to walk up close to see the fine detail. He prides himself on those sorts of images (apart from the other key points he values).
Ross, I don't claim to be in quite the same league as Ken Duncan, or Steve Parrish, FTM.

However, a 10 MPx image printed at A0 size and viewed from an appropriate distance for such a size gives a very respectable amount of detail.

I'm also not silly enough to wade neck deep in a Kakadu lagoon, either ... ;) :rofl: .
 

swifty

Mu-43 Regular
Joined
Jan 30, 2013
Messages
188
Location
Singapore
Real Name
Dave
BSI is not only about fill factor but it has two other benefits which come from basic geometry: light has to travel shorter path before reaching the photosensitive diode area and it can also arrive at wider angles.
That really depends on the pixel design. But this issue is largely mitigated by gapless microlenses anyways.
This all leads to more sensitivity which has quite often been wasted in having more but smaller photosites while maintaining previous performance. Or in laymans terms having more megapixels without losing previous IQ. Only during the latest FF sensor generation all manufacturers have sort of agreed that 24MP is enought and concentrated on increasing the absolute photosite quality instead of their number.
Not sure what you mean by more sensitivity. Sensors only really have one sensitivity. Maybe you're talking about greater efficiency?
Stacked sensor on the other hand will not increase sensitivity or IQ per se, just speed by simply integrating super-fast RAM storage in the sensor silicon itself which removed the previous bottleneck of "hey, we also need to move and store all this data somewhere before we can read more from the next photosite".

This of course does indirectly improve IQ compared to traditional designs when they try to reach same read speed. Traditional non-stacked sensors can reach high read speeds only by cropping just APS-C or 1:1 pixels area from the center of a FF sensor and/or dropping resolution to 8bits from 10/12bits. (and of course the same applies to MFT)

The bottom line is still the same. As long you only shoot relatively static and reasonably well lit scenes which dont have exteme contast and have no need for high read speed, there will be no visible difference worth mentioning. That being said, any scene beyond this confort zone will see anything from visible to dealbreaking IQ improvement with latest sensor technology.
Stacked sensors are really a natural progression of BSI. Stacking just means bonding another piece of silicon to the back of the sensor silicon with the photosensitive diodes.
It's a generic term but in the context of Sony Semiconductor, what they've chosen to stack are DRAM, among things. Nikon has recently shown a 1" sensor with a different stack.
By having the sensor support electronics on the back, now you can connect those electronics directly to other circuitry on the second stack. In Sony's case for the A1, I believe it is a triple stack with DRAM as the middle stack and readout circuitry on the third stack.
The implications of stacking can be very wide depending on what the stack you're bonding to the BSI sensor. For example the Nikon demonstration 1" stacked sensor allows for independent localised exposure zones in addition to the increase readout speed which will lead to increases in DR in certain readout modes.
But yes, since we're talking about a Sony Semi catalogue sensor, it is highly likely it'll take on the characteristics of other Sony stacked sensors in the past. So we can use those as indicators of what this sensor may be able to achieve.
 

JonSnih

Mu-43 Veteran
Joined
Mar 26, 2017
Messages
290
Location
CZE
IF Olympus can get stronger/faster processor that can churn that high amount of data we could see something like Capture 2.0 with possibly (this is even more speculative then an episode of Star Trek all about Faster the Light Travel) 30, 60 or 120 FPS ,electronic shutter only of course, with AF-C or AF-S, possibly no blackouts on the lower speeds, improved object detection and tracking ... something that may blow people's socks of in a E-M1 X Mark II.
I am not sure, Ovidiu. When the E-M1ii was introduced, ppl were responding like: "Wow, 60 FPS. In RAW! But it is stil a 4/3 sensor. No thanks." With 120 FPS it will end up the same, I guess.
Faster data readout is even more essential that most people give credit to because it improves the users experience and daily drive of any cameras, 120 Hz screens can make the UI so much better that it can be a night and day difference but it requires 120 FPS worth of data to make it work, that's 120x20 MP=2.400 MP per second, on average of a 20 MB RAW files that makes it 2.4 GB per second of data to read, downsample to the EVF/back screen to display is no small feat ...
They have to change memory + buffer(s) + data buses design. Regarding dual card slots, my bet is that we will see a combination of UHS-II + CFexpress slots to utilize all the data.
And to AI, the more data you have the more accurate the prediction and recognition (more data will not give you more speed, quite the opposite), the more subjects you can recognise and become more versatile.
I really would like to see another visible progress in AF-C/tracking/AI pattern recognition. I still think that OM could squeeze out even better AF performance out of the 1X/1iii which is about 10-15 % above the E-M1ii.
But yes, since we're talking about a Sony Semi catalogue sensor, it is highly likely it'll take on the characteristics of other Sony stacked sensors in the past. So we can use those as indicators of what this sensor may be able to achieve.
Well, both Panny and OM/Oly hinted several times in the past 12 months that with 4/3 they can get faster data read out than it is possible with larger - especially FF - sensors. The 50MP A1 sensor is capable of 1/240 sec read out. I am sure for 4/3 BSI stacked sensor the 1/1000 sec read out speed is possible (only 1ms lag!). That would unlock great potential.
 

swifty

Mu-43 Regular
Joined
Jan 30, 2013
Messages
188
Location
Singapore
Real Name
Dave
Well, both Panny and OM/Oly hinted several times in the past 12 months that with 4/3 they can get faster data read out than it is possible with larger - especially FF - sensors. The 50MP A1 sensor is capable of 1/240 sec read out. I am sure for 4/3 BSI stacked sensor the 1/1000 sec read out speed is possible (only 1ms lag!). That would unlock great potential.
Of course it could be possible that Panny or Oly have been able to commission very fast sensors. Sony Semi is the leader in this field and from my understanding, any customer who are willing to pay can have designs commissioned to their needs. But it’d likely be pricey though.
In terms of pushing the performance envelope, this tend to happen with smaller sensors first. The first Sony Semi stacked sensor arrived in smaller sizes than FF well b4 the A9.
You just need to look at some of what’s being achieved in smartphone sensors to get an idea of what scaled up larger sensors might be able to achieve in the future.
The sensor in question here had been spotted in Sony Semi’s catalogue though, rather than being a custom. Meaning it’s available to any paying customer. It should mean it’s cheaper compared to customs but unlikely to have any breakthrough tech in it but bear in mind stacked sensor’s are still going to be much more expensive compared to non-stacked.
 

Baenwort

Strictly Amateur
Joined
Aug 22, 2017
Messages
261
Location
Wisconsin, USA
I wonder if BSI sensors are heavier that the current 20MP sensor? That could impact the IBS if I remember that Olympus interview that mentioned one of the reasons their IBS is better is due to the lighter sensor enabling fast movement?
 

L0n3Gr3yW0lf

Dad Pun Joke Master Over Nine Thousand Meme Lord
Joined
Jul 31, 2013
Messages
1,177
Location
UK
Real Name
Ovidiu
I wonder if BSI sensors are heavier that the current 20MP sensor? That could impact the IBS if I remember that Olympus interview that mentioned one of the reasons their IBS is better is due to the lighter sensor enabling fast movement?
BSI sensors shouldn't be heavier then the current 20 MP sensors available because they are just the same product but with the location of the components moved. Stacked BSI is a different subject because that one has another layer in the sensor structure, which contains contains fast readout storage. It could affect the performance of the IBIS as every gram would have an impact on the IBIS springs and the force of impact when you move the sensor stack. Maybe we will see another improvement/evolution in IBIS design to counteract that, like switch from springs to magnets, the way Fujifilm did with X-T4.
Or they could try and balance the weight with lighter glass structure on the sensor stack, the ones that filter different types of light.
 

apete

Mu-43 Regular
Joined
Mar 14, 2018
Messages
101
Location
Poland
Real Name
Piotr
Rob Trek and Ben Chappell have recently had a very interesting discussion about the new sensors. They cover various aspects including BSI.
 
Last edited:

L0n3Gr3yW0lf

Dad Pun Joke Master Over Nine Thousand Meme Lord
Joined
Jul 31, 2013
Messages
1,177
Location
UK
Real Name
Ovidiu
That would mean HHHR for (slow) moving objects.
It would help get there but I doubt it will make a significant enough change. HHHR is 16 images in sequence unless the shutter speed is incredibly fast (over 1/2000 of a second) which means very bright light the total exposure is still quite slow that BSI and read-out will not change much in the shooting style at the shutter button press. Where it would change is after the shutter button is pressed, a fast readout and through-put means a lot shorter processing time, it could make shooting it on a more dependable base instead of having to wait 10 seconds or more for each shot. I use HHHR as a last resort if there is even a very slow-moving/movement subject and only after I make sure I got my single shots I wanted before, to allow myself to experiment with HHHR. Things might change if HHHR has faster processing and reaction time after the shutter action.
 

pake

Mu-43 Hall of Famer
Joined
Oct 14, 2010
Messages
2,503
Location
Finland
Real Name
Teemu
It would help get there but I doubt it will make a significant enough change. HHHR is 16 images in sequence unless the shutter speed is incredibly fast (over 1/2000 of a second) which means very bright light the total exposure is still quite slow that BSI and read-out will not change much in the shooting style at the shutter button press. Where it would change is after the shutter button is pressed, a fast readout and through-put means a lot shorter processing time, it could make shooting it on a more dependable base instead of having to wait 10 seconds or more for each shot. I use HHHR as a last resort if there is even a very slow-moving/movement subject and only after I make sure I got my single shots I wanted before, to allow myself to experiment with HHHR. Things might change if HHHR has faster processing and reaction time after the shutter action.
A new CPU is a must and that should reduce the HHHR processing time as well (hopefully significantly - to a whole new level).
 

L0n3Gr3yW0lf

Dad Pun Joke Master Over Nine Thousand Meme Lord
Joined
Jul 31, 2013
Messages
1,177
Location
UK
Real Name
Ovidiu
A new CPU is a must and that should reduce the HHHR processing time as well (hopefully significantly - to a whole new level).
I, for one, hope they will introduce a 3rd option for HR (and 2nd option for HHHR) where instead of 16 images it makes 8 and they focus the processing on moire, colour correction and accuracy and noise reduction. Pentax does this where it focuses on IQ instead of pure resolution. While it doesn't give mind-blowing improvement it would help the M4/3 sensors get more IQ that brings them closer tk the competition.

With fewer shots you get less risk (though it doesn't completely eliminate it) of motion artefacts, easier to correct artefacts from motion, less processing time, (possibly) more densely packed image files (more data within a smaller canvas size) the way Sigma files are and with some cleaver processing more dynamic range if the Light gathering is averaged the way Live Composite works.
 

greensteves

New to Mu-43
Joined
Jun 2, 2021
Messages
8
If Olympus and Panasonic can improve their animal autofocus AI to the level of Sony cameras by using a BSI sendor, that would be big bonus for wildlife photographers.
For landscapes, possibly an improved hi-res pixel shift processing that has less lag between combined images, and thus less motion blur, would be a nice plus.
I am disappointed that the megapixel count hasn't increased, as sometimes I have to crop heavily in images of distant small birds. I guess I'll just have to buy a longer zoom (or was that part of Olympus' plan 😉 ). On the other hand, if HHHR is sufficiently improved to work with non-flying birds, with maybe a 1/4 second gap between images instead of 13 seconds, this could stand in for a longer lens or more megapixels.
 
Last edited:

Jonathan F/2

Mu-43 Legend
Joined
Aug 10, 2011
Messages
5,041
Location
Los Angeles, USA
As someone who sold off his M43 small kit and switched to the Nikon 1 J5 with 20.8mp BSI sensor for my compact camera needs, the BSI sensor makes a noticeable difference in areas of dynamic range, color richness and even image clarity with smaller sensors. M43 would benefit vastly especially with all the tech embedded in M43 cameras such 5-axis IBIS, hi-res mode, etc. My Nikon 1 J5 outputs like a "baby" Sony FF BSI sensor with the same richness in colors and ability to push DR. I even prefer the image output from my J5 vs the 20mp Pen-F I used to own.

I think an M43 BSI size sensor would be even better and I think would negate most advantages of APS-C.
 

pake

Mu-43 Hall of Famer
Joined
Oct 14, 2010
Messages
2,503
Location
Finland
Real Name
Teemu
If Olympus and Panasonic can improve their animal autofocus AI to the level of Sony cameras by using a BSI sendor, that would be big bonus for wildlife photographers.
For landscapes, possibly an improved hi-res pixel shift processing that has less lag between combined images, and thus less motion blur, would be a nice plus.
I am disappointed that the megapixel count hasn't increased, as sometimes I have to crop heavily in images of distant small birds. I guess I'll just have to buy a longer zoom (or was that part of Olympus' plan 😉 ). On the other hand, if HHHR is sufficiently improved to work with non-flying birds, with maybe a 1/4 second gap between images instead of 13 seconds, this could stand in for a longer lens or more megapixels.
If you need to crop THAT much, you definitely need a longer lens - not more megapixels.
 

Latest posts

Links on this page may be to our affiliates. Sales through affiliate links may benefit this site.
Mu-43 is a fan site and not associated with Olympus, Panasonic, or other manufacturers mentioned on this site.
Forum post reactions by Twemoji: https://github.com/twitter/twemoji
Forum GIFs powered by GIPHY: https://giphy.com/
Copyright © Amin Forums, LLC
Top Bottom