I remember when they claimed that the 1Ds Mark IV was all but confirmed and annoncements would be coming soon...
Upvote
0
DR from 5Ds will be 2 stop better then 7D mk II
- Thread starter etto72
- Start date
bmwzimmer said:
I would have to think it is. Step back a moment and look at the big picture. 5D3 has 14 stops but only 11-12 usable by reasonable measure due to read noise. 5DS has "equivalent DR of 5D3" (14 stops). Brunhill adds to this "more ability to pull up shadows with lower noise floor" (paraphrasing). We know Canon "tuned this for Low ISO" which I'm not sure what else that could really mean other than lower noise and more DR along with this better CFA. In order to get usable images at higher ISO, I imagine you have to sacrifice SOME level of capability at the lower end, so now they have eliminated that trade off by foregoing higher ISOs for the intended audience of this camera. Here's what else makes sense. All this talk about Canon having to remake their fab lines or greatly alter the process to get the AD on the Sensor like Sony.... Canon (if this is all correct, which I have no reason to suspect not at this point) has found a way to keep making sensors just as they have been with separate AD blocks (minimizing new capital expenditures to re tool manufacturing) but making the sensor itself operate at much lower temps (as per Northlight sources) because it's not worried about 3200 and beyond sensitivity...thus creating far less potential for noise creation. This to me appears to be a perfect trade off between making a new product and keeping the company financials happy. I think my logic there is sound, but I'm happy to hear debate.
Upvote
0
bmwzimmer said:
5D4 I suspect will be more of a change in features with moderate tech steps. Adding DPAF, adding 4k capability, small step up in resolution, better 7D2 type pixels with less noise. A great "all-around" camera that general and wedding photography demands. Don't think we will see anything earth shattering there. Now the 1DX2 however.... that should be very interesting...
Upvote
0
PureClassA said:
I still think people are mincing words here. The 5D III does not have 14 stops of DR. The 5D III has 11 stops of DR. You cannot ignore the read noise, and people seem to be trying to do that with these funky explanations of why the 5Ds "has the same DR as the 5D III, but also has more DR than the 5D III". None of it makes sense.
Dynamic range is a very simple concept. It is the ratio between the FWC and the noise floor. Many things can affect the noise floor...including shot noise in the signal itself. I haven't actually heard anyone from Canon officially state that this camera has more dynamic range...they just keep talking about the "noise floor". The 7D II, which nearly the same pixel size, has lower read noise. It has lower read noise because it has small pixels (something about how Canon designs sensors leads to this scaling of read noise with pixel size...not all sensors are that way). The 5Ds has the same small pixels...so, it stands to reason that the sensor will have lower read noise.
The KEY here is that...for the 5Ds to have more dynamic range, it MUST ALSO have a higher FWC than the 7D II does at ISO 100. If it has, say, 13e- RN per pixel at ISO 100, and still has a 29455e- FWC...well, it'll actually have slightly less DR than the 7D II (which has 12.9e- RN per pixel with ever so slightly smaller pixels.) The 5Ds, at 13e- RN, would certainly have less read noise than the 5D III, the 6D, the 1D X, etc. But with a 30ke- FWC, it's got less than half the FWC of a 6D!!! It wouldn't have more DR...it would have the same DR as all Canon cameras have...around 11 stops and change.
If the 5Ds has 13e- RN and say a 45ke- FWC, alright, now we're talking. That is 11.8 stops of DR...half a stop to a stop better DR than other Canon cameras. If it has a 60ke- FWC (double the 7D II...eh, I don't think that's possible with 4.14 micron pixels, especially not on a 500nm process, but let's go with it), then you have 12.2 stops of DR. Hmm...a solid stop better than any other Canon camera...but still not comparable to the 13.8 stops of a D810. So...let's just go all out and say the thing has 75ke- FWC (same as the 6D). It'll have 12.6 stops of DR...eh. For this camera to get 13.8 stops of DR at 13e- RN, it would need an FWC of 180ke-. Not gonna happen...not with 4.14 micron pixels and a 500nm process.
Let's go the other way here. Let's reduce read noise. Let's say it has 6e- RN and a 30ke- FWC. That's 12.3 stops right there. What about at 3e- RN? Bam, 13.3 stops of DR. Let's say it has a higher FWC than the 7D II...say 40ke-, and that same 3e- RN. Voila! 13.75 stops of DR.
Canon needs to reduce their read noise to Exmor levels. There is no way they are going to get enough well capacity in a 4.14 micron pixel to support more than MAYBE a stop more DR at similar RN levels to the 7D II. There are certainly things Canon could have done to reduce read noise to 3e- at ISO 100...however it would be THE unprecedented move they have needed to make for oh so, so long. To actually achieve it, Canon would have had to have moved to a newer, better process (180nm at least, smaller better), and moved all the core image processing logic (amp, CDS, ADC) onto the sensor die, and go CP-ADC (thus reducing ADC frequency considerably). Just doing that wasn't even enough for Sony to achieve the ultra low level of 3e- RN...they had to use additional innovations. They added a digital CDS stage to the readout pipeline. They also moved the clock generator to a special remote area of the sensor die to prevent it from adding uneven noise to whatever ADC units were closest to it. Finally, they designed ADC/DCDS units which adapt themselves to each pixel column, eliminating vertical banding.
I dunno. Maybe it's just me...but that sounds like much too big a leap. Especially given the rumors that the 5Ds is STILL manufactured on a 500nm process. Even more especially given that it's already been stated so many times that it has the DR of the 5D III. Yeah....just doesn't sound plausible to me. I don't think the 5Ds will have Exmor-level 13+ stops of DR. At best...I'd say 12.2 stops...with the faint hope that Canon will totally surprise me and prove me totally wrong with some truly amazing technology in the 5Ds.
Upvote
0
Jrista, why do you even shoot with Canon? You seem to bash them so much. The d810a was just announced. Maybe you should just make the switch. Sounds like you'd be a lot happier than you are here.
Upvote
0
jrista said:
It's possible (although not likely) they're discussing about two different things. The dynamic range of each pixel is related to its capacity versus noise, and the dynamic range of the output is limited by the bit-depth of the ADC and file structure.
If I took a sensor like a Newton 920 with 94dB DR but ran it through an 12-bit ADC and store export 8-bit JPEGs, which would be the appropriate DR to quote?
The 5D line uses 14-bit ADCs and 14-bit files, and thus the output is bounded at 14-stops, for all 5D cameras.
The 5Ds sensors may have lower noise (and similar FWCs) to previous 5D cameras, and thus have greater flexibility to lift shadows. Again, I don't think it's likely, but whatever, 12-ish stops is generally plenty.
benperrin said:
I think you're confusing jrista with dilbert. The former talks plainly about technology, the latter bashes.
Upvote
0
benperrin said:
First, I don't "bash" Canon. I am annoyed with Canon for ignoring the sole issue they have for so long, but I don't bash them. Not anywhere remotely close to how some past members have, not by a long shot. I think Canon can do better. I KNOW Canon has patented technology that would allow them to do better. It's annoying that they just sit on the technology, rather than employing it...and I call them out for that. But that's different than raw, unmitigated Canon hate (of which there are plenty of members here who do just that...hate Canon regardless.)
I'm simply delivering simple facts here. This is simple math. It's objective truths. This has nothing to do with bashing Canon...it's just the facts. DR is DR...it's not complicated, it's not quirky...it's just a ratio. Unless Canon has magically pulled an Exmor out of their conservative hat...I believe the evidence is strongly stacked against the notion that Canon is suddenly going to get 13+ stops of DR on a 500nm process with their high read noise and tiny pixels.
BTW, I don't shoot with Canon for landscapes. I use them for my high ISO stuff, where they perform perfectly well. I also own a $13,000 600mm f/4 Canon lens (and it's the best of it's class, across brands, so I'm not complaining), but because of that lens I am kind of stuck with Canon over the long term for high ISO regardless. High ISO is more physics bound, and Canon doesn't have much in the way of problems there, so it's not an issue.
However, I don't use Canon for landscapes. I gave up on shooting landscapes, at least the kinds of landscapes I really like, once I realized how bad Canon's read noise is in my 5D III, and having seen little improvement in the 6D, 70D, 7D II. I don't have much time for landscapes...I certainly can't go driving all about the Colorado Rockies all the time to find amazing vistas, and it's even more difficult to find them with good light. On top of that, I'm putting all my resources into astrophotography these days. With astro, you have no option but to dig way deep into your signal data...and once you do, there is simply no denying the fundamental and radical differences in data quality between Canon...and the rest.
Canon has stuck with their old technology. From a business standpoint, I understand...but it's still put them well behind the curve when it comes to core image quality. Objective image quality, measurable image quality...I'm not talking about my own personal subjective perceptions of how I "feel" when I see a Canon image or a Nikon image. Again, this is simple, objective, mathematical stuff here. Cold, hard facts...not feelings. I'm not proclaiming from the walls that everyone should "jump ship or die" or anything like that...I'm just delivering some simple facts. If you choose to get emotional over that...that's your deal.
Otherwise...take it for what it is...realism. Unless Canon is truly hiding some phenomenal technological improvements (and if they are...wtf, why?!?!), the cold, hard facts indicate that your very likely not going to be getting 14 stops of DR out of your 5Ds. (Not without Magic Lantern, anyway.
)
Upvote
0
3kramd5 said:
Aye, I think that's mincing. The output is all we have. We can't get the raw sensor signal pre-ADC. According to Roger Clark, the 5D III sensor itself is capable of over 15 stops of DR (ratio of ISO 100 FWC and lowest recorded read noise.) But, that's entirely moot and meaningless, because you cannot get the signal out of the camera without running it through the ADC. The dynamic range of a camera IS the dynamic range of the output.
3kramd5 said:
That would depend on the context. If your just selling the sensor, then 94dB (15.67 stops) is certainly the only number you could quote. However, once you stick that sensor in a camera with a 12-bit ADC, the CAMERA can only deliver, at absolute best, 12 stops of DR. It'll likely deliver less, possibly a couple of bits less depending on the quality of the ADC unit, because of the added read noise. In the case of the CAMERA (vs. the SENSOR), the appropriate DR number to quote would be the one derived from FWC/RN ratio for that particular camera.
I honestly don't see how you could quote anything else. What can you actually USE when you bring that RAW file up in an editor? Can you use the 15.67 stops the sensor itself is capable of? Of course not...so why would you use that number when marketing your camera? You wouldn't, simple as that. (Well, maybe you would...guess it would depend on how honest a business man you were.
)[/size]
3kramd5 said:
Personally, I think whether 12-ish stops is "plenty" is entirely subjective and contextual. There are those of us who could use 16 stops, 20 stops, if we had it. In some contexts, 12 stops is woefully inadequate.
Subjectivity and feelings vs. objectivity and facts. I'm trying to stick to the latter.
Upvote
0
We have some VERY expensive test equipment that we use to verify manufacturers figures on dynamic range. The most honest reading we have had in Cinematography are the ones Arri claim for the Alexa XT which is around 14 stops. It uses large pixels compared to Sony, Canon and Red but equally gives the lowest resolution of these cameras. The C300 / C500 give around 11.7 stops this is a figure similar to that found in the 5D MKIII & 6D. Sony sensors as seen in the A7R and Nikon D800 series are also close to 14 stops averaging around 13.8 a figure similar in the F65 and F55 they also show virtually no visable banding unlike Canon.
Upvote
0
Well fair enough. I'm sure we've all been annoyed at Canon at one point or another for holding back a feature we desperately want that would really help. They certainly seem to be slower to implement features than other companies.
Upvote
0
PureClassA said:
I'm hoping I'm wrong...it would be nice to see Canon blow my mind with a truly radical IQ improvement for a change.
Upvote
0
This is making me think of Horsepower and Torque measurements on cars. In advertising we are quoted the power to the wheels but behind that is a pure power reading straight from the engine before it hits the transmission and drive shafts where some degree of HP is lost. Same true here. Sensor (engine) delivers A given HP before some of it is lost through the signal path and ADC (transmission and drive shaft). If i want to boost my output to the wheels I can either get more out of the engine or make my drive train a more efficient machine.... Or both. canon perhaps did one or the other somewhere in the mix to boost the final horsepower to the wheels.
Upvote
0
