Canon has released the LI5030SA CMOS sensor for sale. It is a full frame global shutter sensor with 12 bit read out.
From Canon
LI5030SA is a CMOS type of solid-state image sensor with a 35mm full frame effective pixel array of 19 Megapixels. It uses a global shutter function instead of conventional rolling shutter. It enables simultaneous exposure timing for all 19 Mega pixels. It can output an effective 5688 x 3334 pixels of video at 57.99 fps and 12bit via 24 channels of digital signal output. LI5030SA series consists of LI5030SAC (color), LI5030SAI (RGBIR), LI5030SAM (monochrome) and LI5030SAN (Naked). LI5030SAN does not have a micro lens or color filter.
it appears as if it would be capable of 5.7k 50fps so oversampled 4k is certainly probable.
- Equipped with a global shutter function that exposes all pixels simultaneously
- Capable of recognizing the shape of objects moving at high speed with high accuracy
- All pixels 57.99fps, 5.7K3.2K 60fps
The sensor is available with or without a color filter array.
Full specifications for the sensor
| LI5030SAC | LI5030SAI | LI5030SAM | LI5030SAN | |
| Filter Type | Color | RGBIR | Monochrome | Naked |
| Recording screen size: | Full frame 35mm equivalent (42.12 mm x 32.64 mm) | |||
| Number of effective pixels: | 5688 x 3334 (Horizontal x Vertical) | |||
| Pixel size: | 6.4 μm x 6.4 μm | |||
| Output format: | LVDS output, 720Mbps@12bit | |||
| Main clock frequency: | 60 MHz (recommended) | |||
| Analog gain: | 0dB, 6dB, 12dB, 18dB | |||
| Saturation (Typ.): | 20,000e (Typ.) | |||
| Sensitivity (Typ.): | (Green) 81,000 e/lx/sec | (Green) 80,000 e/lx/sec | 170,000 e/lx/sec | 30,000 e/lx/sec |
| Dark random noise (Typ.): | 1.8 erms @Analog gain 18dB | |||
| Dark current (Typ.): | 28 e/sec @Analog gain 18dB, 60°C | |||
| Power consumption (Typ.): | 2400 mW @57.99 fps full area | |||
Now could Canon put this into a camera and call it good? Well, not quite. It's not dual pixel, so they would have to do something a bit differently than this sensor, and 12 bit usually implies that you are going to lose at least 1 stop DR. This isn't the sensor either that was at one time in the C700, as it was only a super-35 sized global shutter sensor.
By all accounts the Sony Global Shutter A9 III loses more than that in stills photography performance if you discount the amount of in camera NR that is occurring, but I doubt that is something that Canon wants to do with its imaging products.
Source: Canon USA CMOS Sensors
Many have been saying "we don't need more than 20-25MP" and yet the 45MP sensors have absolutely acceptable sensitivity, noise and dynamic range for many shooters and most shooting.
Back in the 90s, I had the idea for a filter that instead of the traditional R/G/G/B 2x2 checkerboard (with double green channel as it's most important to human vision so the biggest payoff when it's lower noise), that R/B/B/G would make more sense as it'd get you double blue sensitivity in the lowest light, where blue is usually the weakest and where lower noise would most matter.
I also thought a larger checkerboard with 2-3 IR channels would make sense. The camera could then shoot as color infrared or have a choice of channels for B&W infrared. Perhaps each pixel could be a 3x3 checkerboard with 2xR, 2xG, 2xB, then one each of the IR frequencies.
I also though an unfiltered (and thus very substantially better sensitivity) White channel that would have an out-sized role to play in setting pixel brightness. I think the color filter probably rejects 80% of the light coming to the sensor or something, so if so, a R/G/B/White 2x2 filter array would give about twice the total brightness information of a four-filter array.
Finally while on the subject, you could also expand the number of primaries significantly. An 8-primary system would be able to capture 99% of the colors humans can see vs. the current 40% or something (though to be fair the current RGB standard still captures a large majority of the colors we actually see in the world around us; the other colors are rare outside of rainbows, glints in gemstones and so on, but still, there has to be SOME market for a camera that can capture a far greater range of visible color...)
Dynamic range is an interesting thing in that there are a couple different ways it can be quantified. We often see test performed on a Xyla21 chart. Many take those test to be essentially gospel. But in truth the Xyla21 test is easy to fool with noise reduction. So the results you get may not reflect real world dynamic range.
Consider the iPhone 15 Pro Max. The website CineD lab test section reveals that the iPhone scores a seemingly remarkable 13.4 stops of dynamic range. To put this in perspective, this is the exact same score as the Alexa Mini LF shooting in ARRIRAW.
Anyone that owns the iPhone knows that the footage falls apart readily in post. There is little to no room to adjust shadows or highlights. In essence, if you mess up when shooting… your goose is cooked. The key to this poor “real world” dynamic range performance is Exposure Latitude. Examining the CineD lab results for the iPhone reveals an exposure latitude score of 5. This is not a spectacular score. For reference, the current Canon flagship Eos R3 provides an excellent exposure latitude of 8 stops; while the Canon Eos R5C manages a respectable 7 stops. The previously mentioned Alexa LF scores 10 stops. Anyone that has worked with these cameras will have no trouble confirming that these results accurately reflect real world performance.
So what did the Sony A9III score? A solid 9 stops with wiggle room to 10. This score is not only better than every Canon camera ever made… it’s better than every digital camera made with the exception of the Arri cameras. The A9III ties with the $30,000 Red V-Raptor 8K VV and V-Raptor-X… And it does this a 1/10th the power draw.
So back to my personal experience shooting with the Sony A9III. The Sony has the most dynamic range of all of the cameras that I tested. I performed numerous test… at first just to prove that my Canon was superior… then upon seeing the results I successively moved up to cameras I believed would best the Sony. One by one the test results revealed the ugly truth I had refused to see… this dam Sony had better dynamic range than all the cameras I had tested it against. It took me some time to adjust my mind to this reality… in my eagerness to refute the CineD finding, my testing had instead confirmed that the CineD test were correct.
The takeaway is simply this… don’t discount Sony. They are a force to be reckoned with. I sincerely hope that Canon did not hold back with the R1, because the competition is bringing their A-game.
I have watched Sony over the years and have observed many friends and fellow shooters swap from Nikon or Canon to Sony or newer shooters start out with Sony. I have also watched their innovation initially in sensors, then in bodies and finally in glass as they have caught up and in many ways surpassed competition.
I know this post and my reply will illicit all sorts of hate as Canon bashing, look at the statistics for sales, if you don’t like Canon just switch and be gone, etc., etc. My reply is not meant to stoke that response but to note that as you say, Sony is bringing their A game and Canon better step up. I am more of a landscape shooter with interest in wildlife and what Sony is going with the A7R series and the A9III series is spectacular. Match that with their native lenses and open lens architecture and third party options and they continue to be a compelling platform for my style of shooting. My biggest barriers to entry continue to be existing investment in platform and user interface. That barrier will be tested with what Canon does for the rest of this year as I really would like to see a high MP replacement for the 5DSR and R5 that I currently shoot with as well as a compelling competition to the A9III with better native lens options and an open lens architecture that would convince me to upgrade my aging EF lens lineup (600 F4 II, 11-24 F4, Sigma 14mm f1.8, and a long line of other EF lenses).
So, thanks for your appraisal - it's always good to hear direct experience.
And to be clear. It’s not just in dynamic range where Sony has taken the lead. The AF in the A9III is the best I’ve ever seen. Nothing currently publicly available even comes close… and I include both Canon and other Sonys in that. The new Dynamic Active Stabilization might rival the best from Panasonic. Even the ergonomics will not leave most Canon users wanting. Sony has been taking notes.
There are currently 4 new Canons on the horizon. Probably the R5 Mk2 and the R1… and two Cinema Eos cameras. I hope Canon takes Sony serious now, and gives us some impressive tech in these models. I have tons of Canon glass. When I purchased the R3, it was a stopgap for the R1, that we all knew was coming… eventually.
So I’ve really been waiting for the R1 since the announcement of the R3. I would love to see some of the amazing tech we read about on this site actually make its way into something we can buy. Maybe, just maybe Sony stepping up their game will motivate Canon to do likewise.
The takeaway is that Canon’s ‘competition’ mainly comprises their own prior products.
Canon releases LI5030 sensor
By your metric (which is a very important one), that means Canon is succeeding. Yet people insist they need to 'catch up'.
It sound by your testing and comparisons that you are testing in video. That's entirely different as camera manufacturers have a bazillion tricks up their sleeves, especially with very fast sensors to compensate for the raw sensor itself having poor performance. But you are right, i should have clarified. A very fast sensor that is recording 60fps video can take 3 total images at different exposures at 180 fps and combine them to give the appearance of better dynamic range - a really rough example of course. It's one of the toughest things reading through Canon's "high dr sensor patents" is trying to figure out .. what is actually in electronics and what is actually in post sensor software.
In the stills department - there are not so many tricks. There are some, such as applying NR that can cause the amount of noise to be subjectively lower, but not really the same amount of tricks.
I was talking about stills performance, myself - and in this case, they are even with heavy software noise reduction 1.25EV off of their competitors at low ISO.
But absolutely an extremely fast global shutter sensor can do magic for video.
When the D700 came out, I sold everything and jumped ship. It was a mini-D3 and had performance far beyond anything Canon had available at the time, or seemed likely to release at a price point (and in a form factor) that I would want. It took a bit of time to re-buy all the glass I had, but got there, and then some. I shot Nikon for a decade.
When it was time to no longer be a working pro, I got curious about mirrorless. Nikon's only options were the original Z6 and the Z7, both of which, to be blunt, sucked. Canon was dipping their toes into RF but there was no R6 or R5.
Sony was mostly a curiosity to me, until I picked up an A7iii and the AF and overall performance rocked my world. Sure, compared to what the A9iii can do today the A7iii is a turd but damn, it was leaps and bounds ahead of anything else on the market back in 2018. The lowlight performance of that 24mp BSI sensor was also very impressive. So I once again sold everything and moved to Sony.
I'm not afraid to move systems. New interfaces don't bother me. The cost is a consideration but I'll pay if it's a big enough improvement in enough ways.
I'd move back to Canon if they leaped far enough ahead of Sony, and if they opened their mount up at least as much as Sony has. Why not? Life is short.
THAT BEING SAID, they did very well catching up to the game with the R5 and I suspect it will be a similar tale when they do finally release higher-end cinema cams.