Also you said you know the 120mp sensor performed great because of a press release. A press release written by who, toward what public and for what purpose?
Upvote
0
msm said:jrista said:msm said:jrista said:... Guess that's the most concerning thing about Canon. They have some amazing technology...but they aren't using it...so it isn't making money. ...
I just want to comment this. You assume that they have some amazing technology. How do you know it is amazing? If it really was amazing and they could make money on it, it should make it's way out into real products. If it doesn't we can only assume it doesn't perform competitively. Maybe they patent ideas in case they may be used in the future if they manage to overcome some hurdles which make them infeasible today.
Like the demoed 120mp sensor we read about earlier, does anyone outside Canon actually know how it performed? Seems like many just assume it was great, for all I know it could have been terrible.
I'm not assuming. I know for a fact. How? Because I've READ the patents. Canon DOES have some really amazing technology. Many of Canon's patents are similar (but not identical to) patents from Sony and Aptina. The only real difference, as far as I can tell, is Sony and Aptina are actually turning their patents into actual products. Canon...well, so far at least, they seem to just sit on them. I'm hoping that changes with the 7D II.
One of the ones I hope they actually implement is their Dual-Scale CP-ADC patent, as based on the patent it sounds like the closest thing to the Sony Exmor design I've found. If Canon can bring Exmor-like technology to their own cameras, even if it isn't quite as good, it will still be better than what they have.
I also believe that patent is the same technology that Canon used in the 120mp APS-H prototype sensor. We actually know how that performed as well, because Canon published a press release describing it's performance. They described the architecture of the sensor, which clearly stated some kind of hyperparallel on-die processing (i.e. CDS, ADC, etc.) That is exactly what CP-ADC is. The thing operated at 9.5fps, and it really doesn't matter if it had small pixels, because fundamental IQ is related to total sensor area and Q.E., not pixel area. It would have been at least as good as the 1D IV at the time, and any APS-H sensor will have better IQ than an APS-C sensor in identical framing situations. At 120mp, the thing cranked out more resolution than any larger format sensor on the planet...until and since.
Do these patent descriptions quantify the real performance?
msm said:Also you said you know the 120mp sensor performed great because of a press release. A press release written by who, toward what public and for what purpose?
Canon successfully develops world's first APS-H-size CMOS image sensor to realize record-high resolution of 120 megapixels
TOKYO, August 24, 2010—Canon Inc. announced today that it has successfully developed an APS-H-size*1 CMOS image sensor that delivers an image resolution of approximately 120 megapixels (13,280 x 9,184 pixels), the world's highest level*2 of resolution for its size.
Compared with Canon's highest-resolution commercial CMOS sensor of the same size, comprising approximately 16.1 million pixels, the newly developed sensor features a pixel count that, at approximately 120 million pixels, is nearly 7.5 times larger and offers a 2.4-fold improvement in resolution.*3
With CMOS sensors, while high-speed readout for high pixel counts is achieved through parallel processing, an increase in parallel-processing signal counts can result in such problems as signal delays and minor deviations in timing. By modifying the method employed to control the readout circuit timing, Canon successfully achieved the high-speed readout of sensor signals. As a result, the new CMOS sensor makes possible a maximum output speed of approximately 9.5 frames per second, supporting the continuous shooting of ultra-high-resolution images.
Canon's newly developed CMOS sensor also incorporates a Full HD (1,920 x 1,080 pixels) video output capability. The sensor can output Full HD video from any approximately one-sixtieth-sized section of its total surface area.
Images captured with Canon's newly developed approximately 120-megapixel CMOS image sensor, even when cropped or digitally magnified, maintain higher levels of definition and clarity than ever before. Additionally, the sensor enables image confirmation across a wide image area, with Full HD video viewing of a select portion of the overall frame.
Through the further development of CMOS image sensors, Canon will break new ground in the world of image expression, targeting new still images that largely surpass those made possible with film, and video movies that capitalize on the unique merits of SLR cameras, namely their high mobility and the expressive power offered through interchangeable lenses.
*1 The imaging area of the newly developed sensor measures approx. 29.2 x 20.2 mm.
*2 As of August 20, 2010. Based on a Canon study.
*3 Canon's highest-resolution commercial CMOS sensor, employed in the company's EOS-1Ds Mark III and EOS 5D Mark II digital SLR cameras, is equivalent to the full-frame size of the 35 mm film format and incorporates approximately 21.1 million pixels. In 2007, the company successfully developed an APS-H-size sensor with approximately 50 million pixels.
With CMOS sensors, while high-speed readout for high pixel counts is achieved through parallel processing, an increase in parallel-processing signal counts can result in such problems as signal delays and minor deviations in timing. By modifying the method employed to control the readout circuit timing, Canon successfully achieved the high-speed readout of sensor signals. As a result, the new CMOS sensor makes possible a maximum output speed of approximately 9.5 frames per second, supporting the continuous shooting of ultra-high-resolution images.
Images captured with Canon's newly developed approximately 120-megapixel CMOS image sensor, even when cropped or digitally magnified, maintain higher levels of definition and clarity than ever before.
dilbert said:jrista said:...
The thing operated at 9.5fps, and it really doesn't matter if it had small pixels, because fundamental IQ is related to total sensor area and Q.E., not pixel area.
Except that the more rows and columns that are present, the more space is lost to the barriers in-between. If they can keep the area covered by pixels constant whilst reducing the pixel size to provide more pixels then yes, you're right. This comes down to manufacturing process where Canon have been using a larger process. Canon is an using old .5 µm process, while Sony and Toshiba have advanced to .25 µm and .18 µm processes. See the chipsworks site for more info.
LetTheRightLensIn said:CarlMillerPhoto said:LetTheRightLensIn said:My fear is that the 4k it delivers will be marginally better than 5D3 ML RAW and that ML RAW for 1080p won't be possible on the new 5D4.
I hope they notice that as soon as ML RAW came out the 5D3 prices, which had been sinking a lot, shoot right back up to MSRP at most stores for some time. And stop crippling everything to pieces. It's a joke how much better ML gets out of the 5D3 than the Canon firmware alone.
I partitulary hate how Canon has pushed DIGIC processing more and more to the DNR everything to mush that a few studios sadly use on blu-rays at times. As soon as an area doesn't have brightness and extreme contrast DIGIC just turns things to 100% mush. But even in the crisp areas I don't know what they are doing. Sometimes I swear Canon marketing literally had them add a minor Gaussian blur filter in the video output stage. But it might just well be that DIGIC video processing is THAT bad. Maybe it is, there must be some reason they suck with old Canon video processing chips for the C100 and all instead of using any of the much newer DIGICs to read the sensor and do basic processing.
Anyway one would hope they are smart and give the 5D4 1080p RAW out natively and 4k compressed, but a good quality 4k and not low color and not mush and hopefully 10bits. Giving it all the hardware can do is the way to take the world by storm again. Video world moves fast and the film guys don't get into fanboy nonsense and unless Canon pushes max fast they won't ever make a big splash in low to mid-end again.
+10. ML really did save the Mark III for video shooters.
However, I doubt Canon will put Raw functionality into the Mark IV, and in my opinion they don't have to. If they give it clean and detailed 4k internal recording, 10bit 4:2:2 output, and non-mushy 1080p with at least 60p that's all it will take. They could leave ISO performance the same (Mark III blows the Gh4 out of the water in that area) and leave the stills features untouched (although they'll obviously have to update something to make it viable). Heck, if they don't make things impossible to hack they wouldn't even have to bother adding focus peaking/zebras/etc. and could just assume ML would do that work for them (and maybe that could be the way they "protect" their cinema line). Raw is awesome, but the workflow and storage requirements make it less appealing, especially if you have 4k and 10 bit 4:2:2 output as an option.
Whatever they do, I hope they keep the LP-E6 battery and at least one CF (dual CF preferred, though).
I just can't see Canon being 'daring' or smart enough to 'dare' put 10bit internal recording though or fixing up DIGIC so that it doesn't go all plasticky on everything. But what you suggest would be VERY nice, 10bits 4:2:2 4k that is crisp and keeps the fine detail and noise and doesn't do the nasty stuff that digic does even at ISO100.
Lee Jay said:Even 180nm is, what, 8 generations ago?
msm said:jrista, I don't see how this press release quantify performance. It says nothing about SNR or DR etc, it doesn't even try to claim that the sensor is competitive in those regards. It may also require hardware around it which is not feasible nor practical in todays cameras. For instance to read out and process all that data would require a lot more readout channels and processing power than what you see in a 1DX today.
Press releases may also typically be written by PR or marketing personnel written for other purposes than to scientifically describe their findings.
As for patents, I don't read them but they don't actually give any data about how well actual their actual implementations perform do they? Without that data we can not tell if its awesome or not. What seems great on paper might be bad in practice.
msm said:For instance Foveon sensors seem like a great technology on paper does it not? No CFA wasting away 2/3rds of the light and no demosaic algorithm interpolating data and making images soft in 100% view. Yet in real life Foveon is outperformed by standard CFA sensors, it gives the resolution but does not perform well in other aspects. Real life performance is what counts and Foveon sensors don't have it (yet, would like to see that change).