5D4 @ 30 MP X 7 fps --> enough upside from the 5DS for you?

[Sharlin]

But if you ask me why a stupid 4K camcorder that can record up to 240fps in 1080p For $899US
could do better what is technically possible for a camera that probably will cost somewhere betwen $3k-$4k I don't know.

Does that stupid camcorder, or your stupid phone, have a 30 Mpix 36x24mm sensor with deep photosites? Of course it is easier to have 240fps or whatever if you have much less data to flush and much less heat to dissipate! And nobody expects good quality from a phone so you can make shortcuts. Yes, I guess you're right that Canon is holding that back on purpose. But instead of just doing it to piss you off (as you appear to think) they're doing that for solid engineering and business reasons. If you actually were an engineer (software, hardware, whatever) or a manager in a tech business, you'd understand that.

And that is the idea of OPINION! Now I made NO claim about CANON. I strictly wrote IMHO. These are my humble believes and they do not necessarily match everybody else. And I don't expect it to be that way.

If you'd like the 5D4 to have 240fps or whatever, that's perfectly fine. That's an opinion. But you expressly made a claim about Canon - that they're holding the feature back just because.

 

[Jack Douglas]

So now I need to hear the thoughts of those who have been holding off on the 1DX II until the 5D4 was considered. This is going to be a tough decision considering price, given I don't really need the fps and I believe I really need more cropping capability, provided what there is to crop is high quality. While I like the idea of the 1DX II having impressive 4K video I shoot/share a lot more photos than video. Isn't it terrible to have to face such dilemmas in life!

It shouldn't be long before the early testers start to reveal there experiences with the camera.

Jack

 

[Diko]

It made headlines, but headlines don't necessarily mean sales.

Yeah. So I've heard as well, but everyone around me were buying it for that video.

EDIT: I've just checked, cause I recall it to be the first one. And yes - it's the first FF to have video at all and the first in DSRL with FULL HD.

Things like IBIS will never end up on a canon DSLR .. for either legal reasons or even practical.

Interesting. You mean for patents infringement?

and your phone? you're comparing a 1/1.7" sensor to a full frame sensor in terms of heat and performance? I'm curious.. why didn't the A7RII or the A7S shoot 1080p240 if it was so easy?

Honestly, I couldn't find the data nor the bitrate of h264 @240fps neither of 1080p nor of 720p. But never-the-less regardless of the size of the sensor isn't the heat problem be both 1/1.7 and FF the same? I speculate here that it should. Think about it in terms of amount of data, not pixel size.

it's CIPA 900 versus the dual processor 5Ds which is 700.

Now, when you mentioned and I think about the battery goes faster indeed. But also it takes more time to charge.

and for some reason you expected it to have the same (or better) video performance as the 1Dx Mark II?

Mostly NOT, but with some exceptions. And again. Sport photojournalists (heavy 1Dx users) would benefit the most.

.....Canon's never going to look at dominating that market .. because you know what? there's far more people that want it for stills.

True that. But do you realize the tech leap they had to make to offer 4k and how little over it is needed to offer better Slo Mo?

I have no problems about bitching about actual things that canon could have done practically. The electronic EVF hotshoe that they ALREADY have. that actually has worth just beyond the aspect of video and is a huge miss for the 5D Mark IV.

This is interesting.... Since it is known that there's gonna be touchscreen on the 5Dm4 I really really puzzled how this will be implemented in my way of shooting. Put EVF in that and that makes even more hard to imagine things for me. I guess you really well know how to put it in good use.

 

[Sharlin]

and your phone? you're comparing a 1/1.7" sensor to a full frame sensor in terms of heat and performance? I'm curious.. why didn't the A7RII or the A7S shoot 1080p240 if it was so easy?

Honestly, I couldn't find the data nor the bitrate of h264 @240fps neither of 1080p nor of 720p. But never-the-less regardless of the size of the sensor isn't the heat problem be both 1/1.7 and FF the same? I speculate here that it should. Think about it in terms of amount of data, not pixel size.

It's not feasible to just read 720p worth of data from a 30Mpix sensor. You have to drain the whole thing and then decide what to do with all the electrons. Similarly, as I implied before, pixel size matters because larger pixel area means deeper wells means more electrons means more current means more power means more heat to dissipate.

 

[Diko]

Does that stupid camcorder, or your stupid phone, have a 30 Mpix 36x24mm sensor with deep photosites? Of course it is easier to have 240fps or whatever if you have much less data to flush and much less heat to dissipate!

For Slow Mo video they could be doing hardware crop instead of merging pixels and still have drastically better results due to bigger photosites' size, ergo 30Mpix not a problem. As for data rates: Look how nicely my phone handles the following data rates: 2160p@30fps, 1080p@60fps,1080p@120fps, 720p@240fps. Of course nobody expects them in 4:2:2.

If you actually were an engineer (software, hardware, whatever) or a manager in a tech business, you'd understand that.

You'd might be surprised to know I actually was one and can assure you that I already implied my believes about their motives:

...Illogical except the case with profit motives.

However two lines later on I suggest that I believe (this can be ONLY in the realm of speculation since I haven't read their internal marketing and financial reports, based on which they supposedly make their business decisions) that they could probably cover bigger market share with so much less investment or effort. Please refer to my previous post.

Ergo in the context of how CANON will "survive" 4 years on the market with the 5D series (the first video DSLR, remember) that are expected to satisfy young amateur film-makers-wanna-bes is quite strange from my POV not to include it.

If you'd like the 5D4 to have 240fps or whatever, that's perfectly fine. That's an opinion. But you expressly made a claim about Canon - that they're holding the feature back just because.

The latter I already explained. But here the opinion is more a wish/desire for that particular feature.

Something that I could hardly ever prove and that I claim is that Canon business behavior is quite conservative one and they play safe by leaving something in their sleeve, for just "in case of" emergency route. Or to have one more feature for the next model iteration ...And so on and so forth.

 

[Diko]

It's not feasible to just read 720p worth of data from a 30Mpix sensor. You have to drain the whole thing and then decide what to do with all the electrons.

True. But not that of a huge problem to crop it by hardware, meaning using 1/4th (or even only 2 Mpixes, just throwing a number for the sake of the idea) of the CMOS and still have great results. NOT pixel sharing as you might be suggesting.

Similarly, as I implied before, pixel size matters because larger pixel area means deeper wells means more electrons means more current means more power means more heat to dissipate.

Not sure about the wells size correlation to the the power drain (ergo heat issues). Feel free to share some info. I'll be more than glad to learn something new.

My understanding is that ADC (the ones off the sensor) should take the heavy load. But since not the complete 30Mpix are used its load should be far easier to process.

PS:Where is one JRISTA when you need him :/

 

[Diko]

Similarly, as I implied before, pixel size matters because larger pixel area means deeper wells means more electrons means more current means more power means more heat to dissipate.

Not sure about the wells size correlation to the the power drain (ergo heat issues). Feel free to share some info. I'll be more than glad to learn something new.

My understanding is that ADC (the ones off the sensor) should take the heavy load. But since not the complete 30Mpix are used its load should be far easier to process. As for current usage I believe it is correlated to ISO sensitivity and fps, but still not in the case where at least half of the sensor is not used. Interesting, indeed.

PS:Where is one JRISTA when you need him :/
[/quote]

 

[GMCPhotographics]

If the latest rumor turns out to be true, 5D4 stills users will get 30 MP x 7 fps.

That would (crudely) put the 5D4 throughput here in comparison:

1DX2 = 20.2 MP x 16 fps = 323.2 MP/s*
5DS = 50 x 5 = 250 MP/s
5D4 = 30 x 7 = 210 MP/s
7D2 = 20.2 MP x 10 fps = 202 MP/s

One would think parity with the 5DS would be in order here, and as such, one might think 8-9 fps might have been the better sweet spot of 'better than the 5D3' yet 'no threat to steal pricier 1DX2 sales'.

Agree/disagree?

- A

I think that the 5D3's fps was pegged at half of the 1Dx fps. I'm guessing there's an internal Canon document that states that the 5D range will always be half the fps of the 1DX series. So I guess it's no surprise that the 5D4 is half the fps of the 1DxII.
I think the 7fps is more than enough for most photographers. I think the 7DII's fps and crop put it into a very specific market out side of what ever the 1DX or 5D range are doing. I suspect that the 5DSR range is independent of the other cameras in the range. It's fps will always be low due to the priority of megapixel density.
I think the 7DII is intended as a back up to a 1DXII, but offering more reach. The 5D series offer different overlaps, the 5D4 has 1/3rd more resolution that a 1DXII and the 5DSR is unmatched in sensor resolution. So the 5D4 and 5DSR make a perfect pair. I'm not saying that there is wisdom in mixing up these choices. But I think it's clear what Canon's thoughts are. A 1DxII and a 5Dsr make a great versatile package. But there is little in the way of redundancy if one cam fails, is lost, broken, stolen in a professional context.
If I was shooting sports / wild life professionally, I would get a 1DxII and a 7DII. No question. But if I was a professional shooting landscapes and wildlife, the choices become a little more complex because we a crossing genres. I would probably get a 5Dsr, a 1DxII and either a 7DII or a 5D4 depending on which if the two genres is more important.

I shoot a lot of weddings, a bit of wild life and a fair amount of landscapes. So the 5D4 is a logical choice for me. I can get by with a keen eye and timing to get my shots and I don't need the ultra high fps of a 1Dx series...and it's shutter is too loud for my wedding work (my primary photography consideration). Also one of my 5D3's is nearing the end of it's professional life and needs a replacement next season. My oldest 5D3 is 4 years old.

 

[jrista]

Similarly, as I implied before, pixel size matters because larger pixel area means deeper wells means more electrons means more current means more power means more heat to dissipate.

Not sure about the wells size correlation to the the power drain (ergo heat issues). Feel free to share some info. I'll be more than glad to learn something new.

My understanding is that ADC (the ones off the sensor) should take the heavy load. But since not the complete 30Mpix are used its load should be far easier to process. As for current usage I believe it is correlated to ISO sensitivity and fps, but still not in the case where at least half of the sensor is not used. Interesting, indeed.

PS:Where is one JRISTA when you need him :/

Technically speaking, sensors work in the voltage domain. Voltage implies potential, but it does not necessarily mean current flows. Additionally, effectively regardless of ISO, because of gain, after the initial pixel amplification the total voltages the rest of the sensor (and off-die electronics, if there are any analog components off die) works with the same general total charge.

If the pixel FWC at base gain is 60ke-, then you are going to be working with a range of charge up to 60ke- regardless. If you are working at 2x the base gain, then your pixel range drops to 30ke-, however a gain of 2 will scale that back into the 60ke- range before anything else happens with it. So, the amount of current involved won't be changing much. And the current used here is minuscule to start with, it's an electronic integrated circuit with micron-scale transistors.

The camera itself may draw 7-9 volts, however most of that potential would be for handling mechanical stuff like focus. The circuitry itself is probably operating either in the 5V or possibly even 3.3V domain (logic circuitry usually cannot work off of high voltages.) High quality, low noise regulators would be used to clean up the power and deliver a pure DC current to the logic circuitry as well.

Heat has to do with the amount of energy dissipated. That entirely depends on how much resistance is in the circuits. Resistance is intrinsic in every component, every wire, but the right use of the necessary materials can minimize resistance. Minimizing trace lengths can reduce wire resistance. If the circuitry is designed to operate with minimal resistance, using minimal trace path lengths (i.e. by locating most of the circuitry on the sensor die, as close to the pixels as possible), operating at the lowest frequency possible, it will dissipate minimal heat. That is a matter of sensor design. Canon has not been on the cutting edge of sensor design for many years, so who knows if they have been able to integrate all the necessary technology on-chip to minimize head during video readout. Most sensors that use fully on-die readout systems, with column-parallel ADC (Sony, Panasonic, OnSemi and I think Aptina all have this technology) are able to operate at lower readout frequencies to achieve the same readout performance, and the entire readout pipeline is very close to the pixels, which means less heat generated.

If Canon is still using off-die ADC units, my guess is they will have to operate things at a higher clock frequency, they will have to move charge around to much greater distances, and all of that will generate more heat under continuous readout.

 

[jrista]

There is actually another factor that can impact how much heat is generated by a sensor during readout. There are a lot of different ADC designs. You have your simple Flash ADC, however that is not the only design. You have Ramp ADC (I know Canon has several patents for this kind of ADC design), your SA-ADC (successive approximation), your single and dual slope ADCs, and your sigma-delta ADC.

The Flash and Ramp ADCs are not necessarily the most efficient. Flash ADC can require a large amount of transistors (2^n-1 comparators, where n is ADC bit depth) so it can be quite large physically. The Ramp ADC, which I believe Canon uses, is very clock cycle heavy. It requires 2^n-1 clock cycles to properly determine how many bits a given voltage represents. So, for a 14-bit ADC, it requires as much as 16,383 clock cycles to convert each and every pixel! That means for a 30 megapixel sensor, you could burn through as many as 500 billion clock cycles just to convert one image. THAT would create a lot of heat.

A more efficient DAC-based design (the Ramp ADC is a DAC-based counter design) would be the successive approximation ADC. This kind of ADC is more complex, and can require more die area, however it only needs n clock cycles at most to determine which bits need to be flipped to properly represent the analog voltage of a pixel in digital units. So, for a 14-bit ADC, you would need at most 14 clock cycles to convert a pixel with maximum signal, meaning for a 30 megapixel sensor you would only need at most 420 million clock cycles to convert the image. This allows the ADC unit(s) to run at much lower frequency (about 1200 times slower than with a Ramp ADC), which would produce far less heat.

There are other designs, integrators, that like a Ramp ADC require high clock cycle counts (2^n-1 cycles). I know that integrators are often used with image sensors. Until more recently with very small transistor sizes (180nm, 130nm, 90nm and even 65nm as Samsung is using), fitting a more complex ADC like the very efficient SA-ADC onto the sensor die, and particularly doing so for each and every column, was not possible. Dependong on exactly what size transistors Canon is using in their latest sensors, and with what kind of fabrication technology (I don't think aluminum wiring would be efficient enough), they may still be stuck using less efficient means to ADC. If they are using 90nm Cu BSI designs, they they should have no problem putting in at least some level of parallel SA-ADC units, which would drastically reduce cycle cost, and reduce heat for video readout.

(Rather skeptical, though...Canon seems to file sensor technology patents here and there, but it is pretty rare considering how many patents in total they file each year, they are basically a non-entity and sometimes thought of as a joke in the world of sensor innovation, and they are excessively conservative in their approach to releasing new technology. Plus, all the ADC patents I've seen from Canon in the last few years seem to be Ramp ADC, so still clock cycle heavy.)

 

[rrcphoto]

(Rather skeptical, though...Canon seems to file sensor technology patents here and there

here and there?

using a VERY narrow search. This year canon's filed 27 sensor patents, and 100 patents since jan 2015.

using a more expanded search, it's in the 100's this year alone.

 

[jrista]

(Rather skeptical, though...Canon seems to file sensor technology patents here and there

here and there?

using a VERY narrow search. This year canon's filed 27 sensor patents, and 100 patents since jan 2015.

using a more expanded search, it's in the 100's this year alone.

What was your "VERY narrow" search, and in what database? There are sensors, there are image sensors, and there are image sensors for digital cameras. Canon uses sensors in a lot of their products. They also use a lot of sensors in video equipment, particularly security video. I'd like to see what you come up with for image sensor patents that pertain to their photography business. It's "here and there"...at best.

 

[Diko]

Thanks, Jrista!

In recent years CANON re-invest flat amount of money in R&D. That said it is more like a regular anual commitment budget than flexible desire to improve anything. In long-term that is correct path business-wise. However the drawback is that groundbreaking industry changing innovations are almost scarce. Like Canon 5Dm4 its innovation derives from an improvement on their core tech from back 2013 DPAF concept, which perhaps was enabled thanks to their 120 MP CMOS sensor from 2010.

Some patent come out now and then, but from what I've seen. Canon's timing is perfect. Any patent that is core for a new game changing tech comes a month or two before the DSLR body itself. Having in mind that it has been published 2 years before that.... They are very good at that. Sometimes some old patent is risen from the ashes but is not the core idea of the new tech.

 

[rrcphoto]

(Rather skeptical, though...Canon seems to file sensor technology patents here and there

here and there?

using a VERY narrow search. This year canon's filed 27 sensor patents, and 100 patents since jan 2015.

using a more expanded search, it's in the 100's this year alone.

What was your "VERY narrow" search, and in what database? There are sensors, there are image sensors, and there are image sensors for digital cameras. Canon uses sensors in a lot of their products. They also use a lot of sensors in video equipment, particularly security video. I'd like to see what you come up with for image sensor patents that pertain to their photography business. It's "here and there"...at best.

very narrow search for "solid-state image senors" .. and us patent database?
sounds like you're pulling out straws now.

if you open up the search to image sensors and camera systems, it expands.

seriously .. canon files a crap ton of sensor based patents. CR / CW, egami catch 1/100th of them if that.

a core sensor patent is immaterial of usage, and you should know that.

the imaging department's R&D budget last year was 900 million USD. it's not as if they do nothing.

 

[jrista]

(Rather skeptical, though...Canon seems to file sensor technology patents here and there

here and there?

using a VERY narrow search. This year canon's filed 27 sensor patents, and 100 patents since jan 2015.

using a more expanded search, it's in the 100's this year alone.

What was your "VERY narrow" search, and in what database? There are sensors, there are image sensors, and there are image sensors for digital cameras. Canon uses sensors in a lot of their products. They also use a lot of sensors in video equipment, particularly security video. I'd like to see what you come up with for image sensor patents that pertain to their photography business. It's "here and there"...at best.

very narrow search for "solid-state image senors" .. and us patent database?
sounds like you're pulling out straws now.

if you open up the search to image sensors and camera systems, it expands.

seriously .. canon files a crap ton of sensor based patents. CR / CW, egami catch 1/100th of them if that.

a core sensor patent is immaterial of usage, and you should know that.

the imaging department's R&D budget last year was 900 million USD. it's not as if they do nothing.

That is an excessively broad search, if that is what your actually searching. It would search EVERY solid-state image sensor patent out there from every filer. How are you narrowing it down to Canon's patents? Often you have to use the names of the engineers who filed them, as the patents often don't always specify Canon directly (or perhaps use Kwanon instead.)

There are so many uses for solid state image sensors, and Canon has multiple divisions and subdivisions within the imaging division that use image sensors. Medical imaging and security video are two huge ones, for example. Scanners, printers, etc. also use sensors. The reason there are so few patents that show up on sites like NL and CR is because those are the ones that generally pertain to DSLR imaging...the kind of technology that might actually end up in a DSLR, not the kind of technology that might end up in a flatbed scanner.

Anyway. Same old S___ here. Always. I have high hopes for the 5D IV, but, I am still in "I gotta see it to believe it" mode. After the 1D X II, my hopes are higher than they've been in years...but I can't help but be skeptical still. I spend a lot of time reading about image sensor technology...and Canon is barely a blip in the industry at large.

 

[rrcphoto]

(Rather skeptical, though...Canon seems to file sensor technology patents here and there

here and there?

using a VERY narrow search. This year canon's filed 27 sensor patents, and 100 patents since jan 2015.

using a more expanded search, it's in the 100's this year alone.

What was your "VERY narrow" search, and in what database? There are sensors, there are image sensors, and there are image sensors for digital cameras. Canon uses sensors in a lot of their products. They also use a lot of sensors in video equipment, particularly security video. I'd like to see what you come up with for image sensor patents that pertain to their photography business. It's "here and there"...at best.

very narrow search for "solid-state image senors" .. and us patent database?
sounds like you're pulling out straws now.

if you open up the search to image sensors and camera systems, it expands.

seriously .. canon files a crap ton of sensor based patents. CR / CW, egami catch 1/100th of them if that.

a core sensor patent is immaterial of usage, and you should know that.

the imaging department's R&D budget last year was 900 million USD. it's not as if they do nothing.

That is an excessively broad search, if that is what your actually searching. It would search EVERY solid-state image sensor patent out there from every filer..

you were only talking about the what the narrow term was .. I searched for the word phrase or term and the company of canon it's really not that hard. and obviously i know enough to only include canon in the results.

and you're deflecting.

canon does a ton of R&D and files a ton of patents on core image technology and secondary (such as AF with senors, HDR, and related technologies) to try and convince others that they do not is pretty false.

and reading? I guess not.. if you though that canon barely does any sensor patents.

a few blogs don't entail all of which canon (or anyone) is really doing in the field of sensor research.

and egami doesn't catch 1/100th of canon patents.


as an example:

micro lens design for mirrorless sensors.
http://pdfaiw.uspto.gov/.aiw?PageNum=0&docid=20160233259

global shutter related patent
http://pdfaiw.uspto.gov/.aiw?Docid=20160240577

sensor dynamic range patent
http://pdfaiw.uspto.gov/.aiw?PageNum=0&docid=20160234446

dual pixel ADC patent
http://pdfaiw.uspto.gov/.aiw?PageNum=0&docid=20160198110

hybrid AF on Sensor patent
http://pdfaiw.uspto.gov/.aiw?PageNum=0&docid=20160218127

basically that was a month. there could have been more, that was a quick scan.

That also does not include another 3-4 patents that deal with sensor manufacturing.

 

[jrista]

You have to look at the filed date. Those patents span the last eight months, not the last month...and you've listed only five. Five out of eight months. You might be able to scrounge up some more, but the more you find, the broader the spread those filed dates will have. You can have a whole batch of patents published within a week's period, and some of them could have been filed years earlier. There were a number of sensor patents granted to Canon last year that came out of 2008, 2010, and 2011 by filed date.

Canon is not cranking out a hundred image sensor patents a month, and at most they might file that many solid state sensor related technology patents per year, a majority of which would never apply to their photography cameras. They file about 4000 patents in total a year, across their entire company, not a chance in hell that more than a quarter of them are image sensor patents. Maybe two percent...tops.

Anyway, irrelevant conversation. I'm not here to convince you, you are not going to convince me. Time to move on.

 

[rrcphoto]

You have to look at the filed date. Those patents span the last eight months, not the last month...and you've listed only five. Five out of eight months. You might be able to scrounge up some more, but the more you find, the broader the spread those filed dates will have. You can have a whole batch of patents published within a week's period, and some of them could have been filed years earlier. There were a number of sensor patents granted to Canon last year that came out of 2008, 2010, and 2011 by filed date.

Canon is not cranking out a hundred image sensor patents a month, and at most they might file that many solid state sensor related technology patents per year, a majority of which would never apply to their photography cameras. They file about 4000 patents in total a year, across their entire company, not a chance in hell that more than a quarter of them are image sensor patents. Maybe two percent...tops.

Anyway, irrelevant conversation. I'm not here to convince you, you are not going to convince me. Time to move on.

If your esteemed sensor analysis is as good as your reading... Since I stated 100 since Jan. :

 

[jrista]

Always the same old bullshit here. Dear god.

My reading is just fine:

as an example:

micro lens design for mirrorless sensors.
http://pdfaiw.uspto.gov/.aiw?PageNum=0&docid=20160233259

global shutter related patent
http://pdfaiw.uspto.gov/.aiw?Docid=20160240577

sensor dynamic range patent
http://pdfaiw.uspto.gov/.aiw?PageNum=0&docid=20160234446

dual pixel ADC patent
http://pdfaiw.uspto.gov/.aiw?PageNum=0&docid=20160198110

hybrid AF on Sensor patent
http://pdfaiw.uspto.gov/.aiw?PageNum=0&docid=20160218127

*** basically that was a month. *** there could have been more, that was a quick scan.

That also does not include another 3-4 patents that deal with sensor manufacturing.

You picked five patents, said they had all been from the last month. When I looked at the FILED DATES, they spanned a period of eight months.

Out of the supposed 100 you found for "this year"...how many were actually filed THIS year, vs. last year, or the year before, or four, five, six years ago? In terms of when an innovation, or as is often the case a simple evolution on top of an existing innovation, was made, it's the filing date that matters.

Anyway. Same old shit. Same old arguments. So freakin tiring. I'm out.

 

[Deleted member 7828]

It'll depend on the high ISO performance of the 5D4.

If it's stagnant, or a step back from the 5D3, if I truly decide I want more resolution, may as well just go all the way to 50 and grab the 5DS(R). The 5D3 (or 1DX II - still deciding) would remain the go-to cam.