After a 50MP camera what is the next breakthrough?
- Thread starter Hector1970
- Start date
Stu_bert said:
I think you are just misunderstanding what DR is, and why it's valuable. It's not about printing an image with 20 stops. It's about having the cleanest, most accurate data possible out of the camera, so that you do no have limitations when you do process the image.
Noise is just useless information. It reduces how much of the numeric space we are representing the original analog signal in that we can use. That's it. More electronic noise == less usable data.
Remember, DR is just a way of describing a ratio. FWC/RN. That's all we are talking about here. Maximum Allowed Signal over Noise. DR is different than SNR, and also different than contrast ratio. You can achieve whatever contrast ratio you want even if you have 20 stops of DR (or more)...that's all in the processing (and truly, you don't want that much DR in an image? All you need is a tight contrast curve, and your done.
). More DR simply means that you start out with better data when you begin processing...that's all. Also remember that noise contributors are additive. The noise contributors add together, then the square root is taken, to get the total noise:
Code:
DR = MaxSignal / SQRT(DarkCurrent + RN^2)In most daytime photography, dark current is negligible. It might account for an electron or two, which results in 1-1.4e- additional noise at most. That makes read noise and photon shot noise the primary contributors to noise in most daytime photography.
The total noise is added to the signal. So, if you have a maximum signal of 30,000e-, 13e- RN, and 2e- dark current:
Code:
DR = 30000/SQRT(2 + 13^2) = 30000/SQRT(171) = 30000/13.077 = 2294.16:1I am showing the raw ratio here rather than decibels or stops, as it gives us an important bit of information. Tonality. The number of discrete tonal levels that can actually be discerned in a file with 13.077e- noise is 2294. That is barely more than 11 bits worth of information, barely more than 11 stops. That tonality could be easily represented by a 12-bit file. You wouldn't lose anything with a 12-bit ADC.
Now, let's say we had a 30ke- FWC and 3e- read noise:
Code:
DR = [SIZE=2px]30000[/SIZE]/SQRT(2+3^2) = [SIZE=2px]30000[/SIZE]/SQRT(11) = [SIZE=2px]30000[/SIZE]/3.317 = 9044.317:1Just with the reduction in noise, tonality has increased by a factor of four. We now have 13.2 stops of DR, and definitely need 14 bits to represent this number of discretely discernible tones. What if we double the FWC:
Code:
DR = 60000/SQRT(2+3^2) = 60000/SQRT(11) = 60000/3.317 = 18088.634:1.Double the tonality, and we have 14.2 stops of DR. We need 15 bits to represent this number of discretely discernible tones.
If we wanted to get that same tonality without increasing FWC (which generally requires larger pixels), then the only other option is to reduce noise even further. Let's say we reduce dark current to the point where we get only 1e-, and reduce our read noise to 1.32e-, with a 30ke- FWC:
Code:
DR = 30000/SQRT(1+1.32^2) = 30000/SQRT(2.74) = 30000/1.66 = 18072.29:1We are just talking about pure, raw, unadulterated numbers here. ;P There isn't a lot that's special to dynamic range. It's a ratio that ultimately determines how many discrete tonal levels we can discern within our data. Because noise is added to the signal, it eats up the least significant bits, basically rendering them useless. It's effectively the same as using fewer bits. Because of the way gain works, increasing bit depth when the system generates a certain amount of noise does not actually increase tonality. If you lose one LSB with a 12-bit ADC, you'll lose two with a 14-bit ADC, four with a 16-bit ADC. You don't actually gain anything by increasing bit depth unless you also reduce noise. When you do reduce noise, as you can see, you make more effective use of the number space required to represent the given tonality as represented by a direct DR calculation like those above.
What would we need to get 20 stops out of a 30ke- FWC? We would need total electronic noise (read noise and dark current noise) to be 0.03e-:
Code:
DR = 30000/0.03 = 1000000:1 == 20 * log(1000000) / 6 = 120dB / 6 db/stp = 20 stopsIf we had 20 stops of DR in a camera like the 5Ds, all that really means is that it basically has no electronic noise. It would mean we get a pure signal out of the camera. That signal would have photon shot noise (which would still limit tonality by limiting SNR), but it wouldn't be limited by electronic noise.
Upvote
0
jrista said:
sorry, I was mixing formats and actually talking about the 44 by 33 millimeter sensors that you mentioned.
At wost Full Frame is 2.58x bigger than the Canon APS-C sensor, and 44×33 is at best 1.7x bigger than Full Frame.
I call that incremental.
Upvote
0
jrista said:Stu_bert said:
I think you are just misunderstanding what DR is, and why it's valuable. It's not about printing an image with 20 stops. It's about having the cleanest, most accurate data possible out of the camera, so that you do no have limitations when you do process the image.
Noise is just useless information. It reduces how much of the numeric space we are representing the original analog signal in that we can use. That's it. More electronic noise == less usable data.
Remember, DR is just a way of describing a ratio. FWC/RN. That's all we are talking about here. Maximum Allowed Signal over Noise. DR is different than SNR, and also different than contrast ratio. You can achieve whatever contrast ratio you want even if you have 20 stops of DR (or more)...that's all in the processing (and truly, you don't want that much DR in an image? All you need is a tight contrast curve, and your done.
Maybe I am indeed getting confused. The Dynamic Range of a scene is surely the difference between the lightest and darkest point. The DR of a sensor would then be it's ability to capture the scene as close to the live scene as possible. Noise in the camera's various systems as you describe limit the sensors' ability to capture it completely. Remove those limitations, and I may sometimes end up with more DR than I desire (I have to reduce post-capture).
And that's what your maths is demonstrating, how you can improve the tonal ranges you can record through the ability to capture more photons, or have less noise. All good, thank you for taking the effort.
But if I don't need 16 stops, let alone 20, and my sensor can capture say 15 - and thus the tonal range required, then I'm not sure I appreciate why I need the other 5?
Upvote
0
9VIII said:
Ok, I can understand that a little better. I still think gathering 170% more light is meaningful, but it certainly isn't as big as gathering several hundred percent more light with something more like 6x4.5cm size.
Even with a 44x33mm sensor, though, the optics are going to become significantly more expensive if you want corner to corner performance. I think that is where the major value in 35mm format is...you can get excellent optics at relatively reasonable prices. The larger the sensor gets, finding excellent optics at even relatively reasonable prices, let alone affordable prices, becomes exponentially more difficult.
If you have the funds for medium format, I say go for it. There are some excellent and "relatively affordable" options out there right now. It wouldn't be any different than Canon doing it, since Canon would need a new mount anyway.
Upvote
0
Stu_bert said:
Well, now your getting into personal need. I honestly cannot speak to that.
In a scene, dynamic range is the range of tone from brightest to darkest. In hardware, dynamic range is the range of tone from the brightest that can be captured before clipping, to the amount of read noise in an image entirely devoid of light. In a photo, dynamic range is really not a thing...I think it's better to call it contrast ratio or tonality or something like that, since an actual image has a signal to noise ratio. If you don't utilize all of the dynamic range your camera has to offer, you can still have a good S/N.
As for not needing DR...well, if you don't need it, it's really pretty much as simple as attenuating your contrast curve more. That's all it really takes to discard parts of a signal you do not need or did not use.
Upvote
0
jrista said:
For sure 35 millimeter is never going to go away (whereas APS-C might), but in the landscape and portrait range, large format lenses would not be overly expensive. The cheap ones would probably start at around $1000, but if Canon thinks people are willing to pay $3000 for a nice landscape lens, I don't think the idea is unreasonable.
I just think it makes sense to make a new mount as large as possible, because chances are they're never going to do anything like that ever again. It would be the absolute best of its type that there ever will be, so this isn't the kind of system that comes and goes, the idea is they would be making something that would be built upon for potentially many generations.
The attractiveness of the system is undeniable, the practicality of making it happen (affordably) is the question.
Upvote
0
jrista said:Stu_bert said:
Well, now your getting into personal need. I honestly cannot speak to that.
In a scene, dynamic range is the range of tone from brightest to darkest. In hardware, dynamic range is the range of tone from the brightest that can be captured before clipping, to the amount of read noise in an image entirely devoid of light. In a photo, dynamic range is really not a thing...I think it's better to call it contrast ratio or tonality or something like that, since an actual image has a signal to noise ratio. If you don't utilize all of the dynamic range your camera has to offer, you can still have a good S/N.
As for not needing DR...well, if you don't need it, it's really pretty much as simple as attenuating your contrast curve more. That's all it really takes to discard parts of a signal you do not need or did not use.
Personal need - lol. It's all about personal need. Does the tech do enough to allow you to portray the scene how you want to. But you and I hope most others know this
Thanks for taking the time to explain further. I'm with you in the most, and perhaps using DR for the scene and S/N for the sensor might be indeed simpler, but without labouring the point. If portraying a scene, a landscape from your earlier example, I dont necessarily need 16 stops let along 20. That was my point. It's good to have a clean image ie S/N ratio, but just because a scene has 24 stops of range, doesn't mean I as a photographer require a sensor capable of capturing it. And that's what I was trying to say (albeit not as elonquent or succinctly as I should), technology advancement has to have a point. Wasn't the same said about hi-fi devices in the 80s which strived for the lowest THD and ended up being "too clinical / without soul" to some people?
Appreciate that every time I use a grad I'm acknowledging the scene is greater than the sensor. I guess I'm trying to figure out how much more the sensor needs to have. Is it 15 stops (based on using a 3 stop grad on a 12 stop sensor?).
I'd also love to know how much the average person would look at a picture and dismiss it because it only had 12 stops of DR. Clearly a side by side, you'd hope they notice....
Upvote
0
Sporgon said:
Yeah, that's making some waves on astro forums. It's got significantly more Hydrogen alpha sensitivity than any other DSLR on the market...significantly more than even the 60Da. The long exposure modes are nice as well.
The price seems out of touch, unless it can also double as a landscape camera, though. An extremely good mono CCD camera with filter wheel and filters is only a couple hundred bucks more.
Upvote
0
9VIII said:
Surely until they improve materials, the bigger the sensors, the bigger the lenses? APS-C and 4/3 will exist because not everyone wants to carry around that much kit. Bigger lenses are more expensive - all other things being equal.
Mirrorless produces a smaller body, but not necessarily a smaller set of lenses.
Until they come up with "lighter" materials for the optics & and lens itself, or massive improvements in the long range zooms (the xx-xxx type), then I figure I'll still be carrying 4 or 5 lenses with a couple of bodies and a tripod, or for wildlife less lenses but even heavier. And for me, that weight in addition to the cost is probably the deal breaker.
Ironically, Michael Richman over at LuLu has been downsizing for years - lighter is better, which based on him not being a young man makes sense. And then he takes the 645Z and does a bit of a u-turn. I think Pentax has got it pretty good in terms of pricing and specs. 3 years ago, at that price, I would have been tempted. I still am. Just wrestling with the notion of heavier and therefore less. Which may be no bad thing...
Upvote
0
jrista said:
Can't you just add a filter to block out those wavelengths which it's allowing? I thought it was the same sensor, different filter in front? You just want to block some of those wavelengths for normal landscape, and allow them for astro?
Upvote
0
Stu_bert said:
I figure I'll address this question first. Personally, I think about DR entirely differently than this...however, it seems that this is more how most people think about it. To put it simply, the answer is: no one. A little more complex than that: some people (mostly other photographers.
)That said, having more DR isn't about what the user who sees your final, processed image things. At least as far as I see it, having more DR is about making your life as a photographer easier, simpler, with less work. You already know one case for how that works: you could stop using GNDs!
I use GND filters myself...but I have a hard time with them. Not because they are hard to use, but because of how they are limited in the ways they can manage contrast in a scene. For one, they have a flat gradient, and as such, they only work best when you have a flat horizon. Say, a sunset photo on a beach, with a clear sky. THAT would basically be THE epitomy of ideal use case for a GND. The moment you add clouds...or have that nice big rock out a short ways from shore, or are photographing mountains with a deep valley between them, or are photographing a forest with an uneven boundary with the sky, etc....then the use of a GND filter breaks down. You end up having to make additional compromises, you either end up lightening the use of the GND, and thus are still DR-limited, or you heighten the use of the GND, and suffer funky dark shading across the top of your image (you've seen those landscape photos where the mountain peaks fade into pitch black at the top, silhouetted against a bright orange or pink sky...ugh.)
You can always do HDR, bracket your shots, blend in post. That also has limitations. Images don't always blend well, sometimes you have ghosting artifacts that are impossible to correct properly, some have funky artifacts in the shadows. HDR is less applicable when there is movement within the scene. HDR is also still additional work. Sometimes HDR is the only option, and even when you have high DR thanks to low read noise, if you still have a limited FWC, you might still need to use HDR anyway (because photon shot noise will still dominate the shadows.)
More dynamic range, the way I see it, is all about simplifying life for the photographer. Your viewers may not know the difference in the end...but you will know the difference, because the journey from point a (taking the photo) to point b (sharing the photo online/printing the photo) was quite different...simpler, faster, more efficient, more enjoyable, even?
Stu_bert said:
There are a couple things here. First, no one would always need 16 stops of DR every time they took a photo. There are often times when I don't even need the 10.97 stops of DR my 5D III has...say when I am photographing a misty landscape under and overcast sky. Heh, I probably need a mere 6 stops tops for such scenes. Conversely, I could easily use several more stops of DR than I usually have when photographing wildlife in certain ways at certain times of day. Most of my birds and wildlife are shot head-on, light almost directly behind me, fully illuminating the subject. That leads to very flat contrast. It's easy to to get great photos up to as high as ISO 12800 that way, because DR is low...but, such photos are not really the most flattering or best. A lot of the time I would prefer to get at a more interesting angle to my subject...but doing so rapidly increases contrast between the shaded side of the subject and the illuminated side. My DR needs can jump several stops with only a few degrees change in angles between me, my subject, and the sun.
I do believe it is entirely appropriate to say that a camera has dynamic range. I was saying before that images don't really have DR...they have a contrast ratio. The camera, though, does have dynamic range, as it has a maximum allowed signal and a relatively fixed amount of electronic noise. So, scenes and cameras have dynamic range...images have contrast. One might be able to make the argument that a RAW image has dynamic range...I guess what I'm getting at is once the image signal in the sensor is digitized, you are locked into what you have, and can no longer get any more out of what the camera may have to offer (i.e. you couldn't overexpose, without clipping the highlights, more than you did to preserve more shadow detail...once you have the RAW, you can lift the shadows, recover the highlights...but they are what they are, they have the noise they have, you cannot change that.)
My question is...if you periodically DO need (or, maybe less controversially...periodically "could make full use of") 15 or 16 stops of DR, would you settle for a camera that had only 12?
Upvote
0
Ah... I'm not sure the question is as simple as you pose it
Would I like more DR? Sure. Would I like sharper lenses? Sure. Across the frame? Yes please. I care about corners as much as the centre as my subject isn't always there. Would I want Canon to invest heavily in this and less in other areas? Would I want DR and Canon to go under in 5 years?
The camera makers haven't figured out where the bottom of the market is for them and therefore what their financial position is going to be next year, let alone 5 years, which is probably how long a return on investment would be for a new sensor fab. Finance people tend to control companies more in a declining market, and Japanese tech companies are more cautious. Even Sony, but it is clearly deriving a lot of money from sensors not just in the photo industry. I would also guess they never had a 500nm fab...
Nikon would not be where they are without Sony tech. Nor would Pentax who I believe use them. That to me suggests the cost for Nikon to design and deliver a new sensor with a new fab was too much for them. And Nikon has less headroom financially than Canon.
Where am I going with this?
Canon talk most to Pro photographers. I reason that the average Pro would like to edit pictures quicker, but as long as they can produce something which sells, then DR isn't burning for them. As in sufficient for them to jump ship. That's for landscapes through to wedding photographers. And that's why I asked you about the average user noticing Jon. Not to catch you out, I also agree with you. It's funny, I would guess more Pro's jumped ship from Canon during the 1D/1Ds MK III AF debacle than over DR. And I'm not sure that's going to change significantly more with the new release.
Canon aren't about to make the change as they dont perceive it's important enough vs the risk of investing the cost. And I would guess they aren't seeing enough churn. Maybe the 5Ds / MK IV / 1x2 will change that. I just dont think it's (DR gap) is compelling enough.
As you stated, and I agree, a decent number of scenes dont need more than >12 stops. As an aside, yup, agree about filters, but most end-users wouldnt notice.
So it comes back to what you want a camera to do for you, what are you doing with the images you produce, and is that aligned enough to the other consumers of Canon gear (Pro & Am). If you sell, then DR is broadly only an efficiency thing.
You're one of the few people who explain why and what you want it for, which I respect. I'm just not sure this is the place for you to influence Canon more, which is ultimately what more people have to do. If it becomes the #1 thing for everyone, and that gets back to Canon, well they might change (if the finance is right, and they fear losing more to Sonikon).
If it's not, then that's not a reason for you and I not to want it, and request it, but it doesn't become the be-all....
Would I like more DR? Sure. Would I like sharper lenses? Sure. Across the frame? Yes please. I care about corners as much as the centre as my subject isn't always there. Would I want Canon to invest heavily in this and less in other areas? Would I want DR and Canon to go under in 5 years?
The camera makers haven't figured out where the bottom of the market is for them and therefore what their financial position is going to be next year, let alone 5 years, which is probably how long a return on investment would be for a new sensor fab. Finance people tend to control companies more in a declining market, and Japanese tech companies are more cautious. Even Sony, but it is clearly deriving a lot of money from sensors not just in the photo industry. I would also guess they never had a 500nm fab...
Nikon would not be where they are without Sony tech. Nor would Pentax who I believe use them. That to me suggests the cost for Nikon to design and deliver a new sensor with a new fab was too much for them. And Nikon has less headroom financially than Canon.
Where am I going with this?
Canon talk most to Pro photographers. I reason that the average Pro would like to edit pictures quicker, but as long as they can produce something which sells, then DR isn't burning for them. As in sufficient for them to jump ship. That's for landscapes through to wedding photographers. And that's why I asked you about the average user noticing Jon. Not to catch you out, I also agree with you. It's funny, I would guess more Pro's jumped ship from Canon during the 1D/1Ds MK III AF debacle than over DR. And I'm not sure that's going to change significantly more with the new release.
Canon aren't about to make the change as they dont perceive it's important enough vs the risk of investing the cost. And I would guess they aren't seeing enough churn. Maybe the 5Ds / MK IV / 1x2 will change that. I just dont think it's (DR gap) is compelling enough.
As you stated, and I agree, a decent number of scenes dont need more than >12 stops. As an aside, yup, agree about filters, but most end-users wouldnt notice.
So it comes back to what you want a camera to do for you, what are you doing with the images you produce, and is that aligned enough to the other consumers of Canon gear (Pro & Am). If you sell, then DR is broadly only an efficiency thing.
You're one of the few people who explain why and what you want it for, which I respect. I'm just not sure this is the place for you to influence Canon more, which is ultimately what more people have to do. If it becomes the #1 thing for everyone, and that gets back to Canon, well they might change (if the finance is right, and they fear losing more to Sonikon).
If it's not, then that's not a reason for you and I not to want it, and request it, but it doesn't become the be-all....
Upvote
0
I totally understand what you are saying. Here are some of the things I do know.
1. Canon themselves have blamed part of the decline in DSLR sales on lagging technological improvements
2. A couple years ago financials from Nikon indicated that they had gained a significant amount of DSLR growth thanks to their improved technology (can't say what technology specifically, the documents don't say), in significant part due to the D600 and D800 as well as some of their mirrorless cameras (the latter seems to have fallen off, honestly not sure about the others as there have not been more recent financial reports that state similar information)
3. Canon already has a 180nm fab, so they would not need to invest in a new one
Given the above, particularly given #3, why would Canon "go under" in five years by working on improving their dynamic range? It's even simpler than that. Canon already has patents for many of the same kinds of technology that Sony used to improve DR in their Exmor sensors. The big one was going to CP-ADC, but Sony also upsells their digital CDS as well. At the very least, since Canon already has a newer and more advanced fab (this very low cost to just start using it), and since they already have patents for the kinds of technology that would improve dynamic range, the cost barrier seems very low, for the gains it could provide.
As you can see with the math, reductions in read noise are the best way to rapidly gain more DR. Increases in FWC can help as well, but that is a lot more physics-bound...small pixels have small capacities unless you do something much more drastic (like embed CCDs into each pixel that can expand the charge capacity, called multibucketing...but that requires smaller transistors, too!)
When the D800 came out, Nikon benefited from it, according to their financial papers from the following fiscal cycle. Canon openly acknowledged that lagging technological advancements (generally speaking, they did not call out sensors specifically, but sensors are a key technology area in digital imaging) was a part of their growth problem. If Nikon, a much smaller company with more schizophrenic marketing and model naming (well...I donno about the marketing...Canon has had some STRANGE marketing lately....
), could make gains of 45% in a single fiscal cycle thanks in significant part to the D600 and D800...why couldn't Canon, a company that has a history of excellent execution?
Something else I know:
4. Sony is making big waves in many markets.
I used to be snugly engulfed in bird and wildlife photography. Lot of photographers there are long-time Canon users, and there is solid brand loyalty. That said, I have been seeing an increasing number of bird photographers using the D800. For certain kinds of birds...say chickadees, or common loons...birds with extremely high color contrast (deep black and bright white all on the same bird, as the case with these two), the D800 up through ISO 400 performs significantly better than the 5D III or 1D X. I was amazed at the quality of some ISO 200 Loon images from one particular professional bird photographer...very fine detail in both the bright white feathers as well as the deep black feathers on the shaded side of the bird. The lighting on the bird was more "non-standard" in that it wasn't lit broadside like usual...the light came more from the side, and the bird was angled towards the photographer...very difficult situation which usually blows the DR limit of a 5D III or 1D X.
I am now also engulfed in the astrophotography world. The D810, D810a, A7s and A7II are all making big waves in the astrophotography world lately. I'm friends with a modder, a guy who opens up these cameras (well, Sony and Canon cameras) and mods them for higher hydrogen alpha sensitivity. He has been modding A7s and A7IIs lately, and he routinely remarks about the higher build quality of the Sony cameras vs. Canon cameras, their tighter construction. Feature-wise that doesn't say anything...there are still some functional limitations on the A7 series of cameras that don't compete with Canon cameras...but it is very interesting to me the comments he makes about the superior build construction and quality of the innards of Sony cameras vs. Canon cameras (and he's taking those things apart, modding them, and rebuilding them all the time, it's one of his businesses, how he makes part of his living.)
The D800 and D810 have been regularly demonstrated to produce much lower noise across the board, even at higher ISO, than the 5D III and 6D. This is largely thanks to lower dark current, but many users are shooting at wider fields, and using lower ISOs like 200, where they also have lower read noise. Nikon cameras have much better color discernment than Canon cameras, and also have higher natural Ha sensitivity, and tend to produce more richly colored subs right out of the camera.
Old, used Canon Rebels used to be THE go-to camera for beginner astrophotographers. That paradigm, which has been in place for near a decade, is rapidly shifting. The D5100, 5200, and particularly the D5300 have become the new darlings of the novice. They can be picked up for similar used prices as Canon Rebels, but they have significantly lower dark current (for all except the 7D II), have very low read noise (the D5200 in particular, as it uses the Toshiba sensor and has some of the lowest high ISO dark current I've ever seen), and have absolutely no banding or blotchiness of any kind.
I know this is all anecdotes...but, it's things I've observed. I've been observing little things like this for years now. The trend keeps spreading as the years go on. I used to recommend Canon Rebels or DSLRs in general for new astrophotographers, as Canon used to take the cake on signal linearity. Today, Nikon and Sony cameras are significantly more linear. Nikon has even stopped using black point clipping in their newest models, and are now using a small bias offset of 600 natively. The Exmor sensors had very good linearity to start, but both Nikon and Sony were black point clipping. Removal of the black point clipping results in even better linearity, and consistent linearity across ISO settings (something that Canon is actually fairly poor at...there are one or two thresholds where increasing Canon ISO results in key changes in signal linearity in Canon cameras, probably due to the secondary amplifier kicking in.)
Bill Claff recently generated new plots on his charts from some Samsung NX1 data. The NX1 has some of the best linearity I've ever seen, besting even the D810. The NX1 has one of the most advanced sensors on the market, and is the first APS-C camera to get a BSI sensor design. It's got an extremely compelling featureset that directly competes with the 7D II (and the key features have nothing to do with DR)...Canon may well be losing a 7D II sale to me, as I am pretty sure I'll go with an NX1 instead.
I notice these things as I continue expanding my knowledge of photography and cameras. I also notice that I'm not the only one noticing them. I hear a lot more in recent months people questioning why Canon does the things they do. Those questions are even more significant when they come from modders who are actually in the guts of these cameras on a regular basis, seeing the differences in design and construction.
I think Canon is right in their assessment that lax technological improvement has, in part, hurt sales. It's certainly making me hold back, consider other options. It's also certainly made me recommend cameras from different brands.
Has any of this impacted Canon's bottom line? Well, tough to say. I do know that I've recommended the D5100 or D5300 to many new astrophotographers who took the advice. I would say those are at least some sales that at the very least did not go to Canon (not sure that's a "lost" sale to Canon...but Canon did not get the sale.) I know a number of skilled astrophotographers who rave about the D800/810 for anyone looking to buy a 6D...bit of a price dichotomy, but the D600 is just as good, and priced much more like the 6D (and that's often my recommendation, for them to get the D600 instead.) Not everyone takes that advice, the 6D is one of the best astro cameras Canon currently offers, but some people do indeed take it. More sales that Canon did not get.
How long does it take for such small changes in purchasing decisions to become a big trend? I dunno, but Canon has seen falling DSLR sales. Too bad we can't get information about WHY each manufacturer is seeing falling sales... ;P What about Canon's mirrorless offerings? They repeatedly pale in comparison to the big players in that segment...such as Panasonic and Sony. Sony's A7 line is quite new to that market, but is still more highly regarded by far than any EOS-M option. The A7 series even made digital-converts out of many die hard large format film landscape photographers, the ones who said they would never, ever give up their Velvia 50 4x5. That's really saying something...
I can't say that Canon is "in fact" losing sales to the competition because they don't have more DR. What I can say is that I see people questioning what Canon is doing with their cameras, and I see many of them (a growing majority, even?) ultimately buying non-Canon. It isn't just a handful of pain in the ass nutcases here on CR bitching about Canon...people are actually making purchasing decisions based on information from people like myself, like my camera-modding friend Mike, etc. and most of that information is based on the core, fundamental technology. In the astro world, basically the only piece of technology that matters is the sensor, and how it's read out...which makes sensor technology the key deciding factor for many discerning astrophotographers (which, BTW, is a growth market.
)
1. Canon themselves have blamed part of the decline in DSLR sales on lagging technological improvements
2. A couple years ago financials from Nikon indicated that they had gained a significant amount of DSLR growth thanks to their improved technology (can't say what technology specifically, the documents don't say), in significant part due to the D600 and D800 as well as some of their mirrorless cameras (the latter seems to have fallen off, honestly not sure about the others as there have not been more recent financial reports that state similar information)
3. Canon already has a 180nm fab, so they would not need to invest in a new one
Given the above, particularly given #3, why would Canon "go under" in five years by working on improving their dynamic range? It's even simpler than that. Canon already has patents for many of the same kinds of technology that Sony used to improve DR in their Exmor sensors. The big one was going to CP-ADC, but Sony also upsells their digital CDS as well. At the very least, since Canon already has a newer and more advanced fab (this very low cost to just start using it), and since they already have patents for the kinds of technology that would improve dynamic range, the cost barrier seems very low, for the gains it could provide.
As you can see with the math, reductions in read noise are the best way to rapidly gain more DR. Increases in FWC can help as well, but that is a lot more physics-bound...small pixels have small capacities unless you do something much more drastic (like embed CCDs into each pixel that can expand the charge capacity, called multibucketing...but that requires smaller transistors, too!
)When the D800 came out, Nikon benefited from it, according to their financial papers from the following fiscal cycle. Canon openly acknowledged that lagging technological advancements (generally speaking, they did not call out sensors specifically, but sensors are a key technology area in digital imaging) was a part of their growth problem. If Nikon, a much smaller company with more schizophrenic marketing and model naming (well...I donno about the marketing...Canon has had some STRANGE marketing lately....
), could make gains of 45% in a single fiscal cycle thanks in significant part to the D600 and D800...why couldn't Canon, a company that has a history of excellent execution?Something else I know:
4. Sony is making big waves in many markets.
I used to be snugly engulfed in bird and wildlife photography. Lot of photographers there are long-time Canon users, and there is solid brand loyalty. That said, I have been seeing an increasing number of bird photographers using the D800. For certain kinds of birds...say chickadees, or common loons...birds with extremely high color contrast (deep black and bright white all on the same bird, as the case with these two), the D800 up through ISO 400 performs significantly better than the 5D III or 1D X. I was amazed at the quality of some ISO 200 Loon images from one particular professional bird photographer...very fine detail in both the bright white feathers as well as the deep black feathers on the shaded side of the bird. The lighting on the bird was more "non-standard" in that it wasn't lit broadside like usual...the light came more from the side, and the bird was angled towards the photographer...very difficult situation which usually blows the DR limit of a 5D III or 1D X.
I am now also engulfed in the astrophotography world. The D810, D810a, A7s and A7II are all making big waves in the astrophotography world lately. I'm friends with a modder, a guy who opens up these cameras (well, Sony and Canon cameras) and mods them for higher hydrogen alpha sensitivity. He has been modding A7s and A7IIs lately, and he routinely remarks about the higher build quality of the Sony cameras vs. Canon cameras, their tighter construction. Feature-wise that doesn't say anything...there are still some functional limitations on the A7 series of cameras that don't compete with Canon cameras...but it is very interesting to me the comments he makes about the superior build construction and quality of the innards of Sony cameras vs. Canon cameras (and he's taking those things apart, modding them, and rebuilding them all the time, it's one of his businesses, how he makes part of his living.)
The D800 and D810 have been regularly demonstrated to produce much lower noise across the board, even at higher ISO, than the 5D III and 6D. This is largely thanks to lower dark current, but many users are shooting at wider fields, and using lower ISOs like 200, where they also have lower read noise. Nikon cameras have much better color discernment than Canon cameras, and also have higher natural Ha sensitivity, and tend to produce more richly colored subs right out of the camera.
Old, used Canon Rebels used to be THE go-to camera for beginner astrophotographers. That paradigm, which has been in place for near a decade, is rapidly shifting. The D5100, 5200, and particularly the D5300 have become the new darlings of the novice. They can be picked up for similar used prices as Canon Rebels, but they have significantly lower dark current (for all except the 7D II), have very low read noise (the D5200 in particular, as it uses the Toshiba sensor and has some of the lowest high ISO dark current I've ever seen), and have absolutely no banding or blotchiness of any kind.
I know this is all anecdotes...but, it's things I've observed. I've been observing little things like this for years now. The trend keeps spreading as the years go on. I used to recommend Canon Rebels or DSLRs in general for new astrophotographers, as Canon used to take the cake on signal linearity. Today, Nikon and Sony cameras are significantly more linear. Nikon has even stopped using black point clipping in their newest models, and are now using a small bias offset of 600 natively. The Exmor sensors had very good linearity to start, but both Nikon and Sony were black point clipping. Removal of the black point clipping results in even better linearity, and consistent linearity across ISO settings (something that Canon is actually fairly poor at...there are one or two thresholds where increasing Canon ISO results in key changes in signal linearity in Canon cameras, probably due to the secondary amplifier kicking in.)
Bill Claff recently generated new plots on his charts from some Samsung NX1 data. The NX1 has some of the best linearity I've ever seen, besting even the D810. The NX1 has one of the most advanced sensors on the market, and is the first APS-C camera to get a BSI sensor design. It's got an extremely compelling featureset that directly competes with the 7D II (and the key features have nothing to do with DR)...Canon may well be losing a 7D II sale to me, as I am pretty sure I'll go with an NX1 instead.
I notice these things as I continue expanding my knowledge of photography and cameras. I also notice that I'm not the only one noticing them. I hear a lot more in recent months people questioning why Canon does the things they do. Those questions are even more significant when they come from modders who are actually in the guts of these cameras on a regular basis, seeing the differences in design and construction.
I think Canon is right in their assessment that lax technological improvement has, in part, hurt sales. It's certainly making me hold back, consider other options. It's also certainly made me recommend cameras from different brands.
Has any of this impacted Canon's bottom line? Well, tough to say. I do know that I've recommended the D5100 or D5300 to many new astrophotographers who took the advice. I would say those are at least some sales that at the very least did not go to Canon (not sure that's a "lost" sale to Canon...but Canon did not get the sale.) I know a number of skilled astrophotographers who rave about the D800/810 for anyone looking to buy a 6D...bit of a price dichotomy, but the D600 is just as good, and priced much more like the 6D (and that's often my recommendation, for them to get the D600 instead.) Not everyone takes that advice, the 6D is one of the best astro cameras Canon currently offers, but some people do indeed take it. More sales that Canon did not get.
How long does it take for such small changes in purchasing decisions to become a big trend? I dunno, but Canon has seen falling DSLR sales. Too bad we can't get information about WHY each manufacturer is seeing falling sales... ;P What about Canon's mirrorless offerings? They repeatedly pale in comparison to the big players in that segment...such as Panasonic and Sony. Sony's A7 line is quite new to that market, but is still more highly regarded by far than any EOS-M option. The A7 series even made digital-converts out of many die hard large format film landscape photographers, the ones who said they would never, ever give up their Velvia 50 4x5. That's really saying something...
I can't say that Canon is "in fact" losing sales to the competition because they don't have more DR. What I can say is that I see people questioning what Canon is doing with their cameras, and I see many of them (a growing majority, even?) ultimately buying non-Canon. It isn't just a handful of pain in the ass nutcases here on CR bitching about Canon...people are actually making purchasing decisions based on information from people like myself, like my camera-modding friend Mike, etc. and most of that information is based on the core, fundamental technology. In the astro world, basically the only piece of technology that matters is the sensor, and how it's read out...which makes sensor technology the key deciding factor for many discerning astrophotographers (which, BTW, is a growth market.
)
Upvote
0
You'd need to ask: What they can and can't improve.
I mean, you do have the 1DX, 5D3 and 6D to replace. ¿How are they going to do that and leave enough space to consider one from the other? (And not kill the 5ds, 5dr, 7D2 and current dslr).
On the 1DX you do have 10/12/14 fps with dual cf card, high iso performance and a body that would endure hell and heaven.
So, to improve the 1DX you'd have to: Get a better frame rate (which i don't believe this global shutter will feature 30 fps, it's too much and cf/sd cards won't handle it, unless you get), better raw/jpeg buffer, a crop mode (x1.3, x1.6), and those small features people have been asking on the 1dx.
The 5D4 (which i think it'll be announced on 2k16, since the 5DS is around) needs to make a jump good enough to justify it over a used 1DX, a used 5Ds, a cheapo 6D and cheapo 5D3.
So, better iso performance, better DR, not that much frame rate because it'll kill the 7D2, ¿2k/4k video?, small software features (intervalometer, hdr, etc).
And then the 6D, the one i believe will get replaced this year (¿October perhaps?), the cheapest full frame ever.
I do believe they will add 1/8000 shutter, dual sd cards, outstanding iso performance (perhaps a tiny bit worst than the 5D4 this time), and more focus points.
I mean, you do have the 1DX, 5D3 and 6D to replace. ¿How are they going to do that and leave enough space to consider one from the other? (And not kill the 5ds, 5dr, 7D2 and current dslr).
On the 1DX you do have 10/12/14 fps with dual cf card, high iso performance and a body that would endure hell and heaven.
So, to improve the 1DX you'd have to: Get a better frame rate (which i don't believe this global shutter will feature 30 fps, it's too much and cf/sd cards won't handle it, unless you get), better raw/jpeg buffer, a crop mode (x1.3, x1.6), and those small features people have been asking on the 1dx.
The 5D4 (which i think it'll be announced on 2k16, since the 5DS is around) needs to make a jump good enough to justify it over a used 1DX, a used 5Ds, a cheapo 6D and cheapo 5D3.
So, better iso performance, better DR, not that much frame rate because it'll kill the 7D2, ¿2k/4k video?, small software features (intervalometer, hdr, etc).
And then the 6D, the one i believe will get replaced this year (¿October perhaps?), the cheapest full frame ever.
I do believe they will add 1/8000 shutter, dual sd cards, outstanding iso performance (perhaps a tiny bit worst than the 5D4 this time), and more focus points.
Upvote
0
There is something I've said a lot, and somehow it doesn't seem to sink in. That's ok, I'm probably just not explaining it well enough. But here goes again:
Noise affects the entire signal.
When people hear read noise, they first think: shadows. Sure, read noise has the most significant impact on the shadows, but the shadows are not the only thing it impacts. The midtones are impacted as well. The tones halfway between an 18% gray midtone and the shadows are affected. The highlights are affected. Every single tonal level in the image is affected. It isn't just the shadows. I'll see if I can create a graphic to demonstrate, maybe that will explain better than words can (they do say a picture is the embodiment of 1000 words...)
So, for a camera like the 1D X, with over 38e- read noise at ISO 100...could it's image quality be improved? Are we assuming that just because the 1D X has big pixels and a big FWC that it is already delivering the best IQ possible?
The 1D X has an FWC around 91ke-. With 38e- read noise, it's got around 11.26 stops of dynamic range. What happens if Canon could reduce that read noise to 3e-? Dynamic range jumps to 14.94 stops. FOURTEEN POINT NINE FOUR STOPS.
Ever looked at the OOF background blur of a bird photo taken with the 1D X? A wildlife image? Ever looked at the boke in a macro photo taken with the 1D X? Ever compared that background with the background of similar photos taken with the D800? Big difference....especially on a normalized basis. HUGE difference on a normalized basis. Creamy smooth, clean, practically noise-free OOF backgrounds from the D800, D810, D600, D750, A7r, A7s, etc. But even the mighty 1D X has visible noise in OOF backgrounds...
I don't deny that there are a lot of other things Canon could improve. I look at it this way, though. Canon has no problem with their AF system. Their AF systems are awesome...well designed, highly capable, top of the line in terms of frame coverage, very fast, and now meter-linked which allows even more advanced features. Canon has no problem with their metering. It's now (in the models that matter) a full color RGB meter with high resolution sensor, and advanced algorithms are hooked into the metering subsystem to enhance it's effectiveness. Canon has no problem with their ergonomics, nor their button placements (they are sublime, if I do say so myself!) Canon has few problems with their lenses (wide angles suffered a bit in the past, but Canon certainly seems to be remedying that problem.) Canon now currently holds the title for highest resolution camera...certainly no problems there.
Could canon make improvements in those areas? Probably...they have, even. The 7D II increased the resolution of the iTR meter sensor. It also expanded the AF point count a little bit above and beyond even the great 1D X. But, how much do those improvements actually matter? Outside of a tiny bit of jitter in a couple cases, I've never even had problems with my 5D III AF/meter system, and it doesn't even have the iTR system. The biggest focus issue I have is that the lens hunts forward first, rather than outward first, when I try to photograph BIF (would love to have a little switch on the lens for that, but that's a small thing.)
The single biggest gain for Canon, given that they already excel in every other area, would be to address the one area they don't excel at: read noise. Change that one thing, and the 1D X could skyrocket from delivering a mediocre 11.26 stops of DR, to delivering 14.94 stops of DR. Bam. Suddenly they actually have a NEED for a 16-bit ADC, because they can actually use the bit depth offered by it. Suddenly they rocket forward as the premier camera model with the highest dynamic range...but not only that, it's also paired with all the other excellent technology I just mentioned above. Who could possibly complain about that!
I see Canon as having one single weak area. As far as I am concerned, they don't NEED to "fix" anything else, because nothing else is broken. Canon needs to improve their read noise issues, though...because, well at least in my own personal opinion, it's the one single thing that Canon really sucks at. ;D
Noise affects the entire signal.
When people hear read noise, they first think: shadows. Sure, read noise has the most significant impact on the shadows, but the shadows are not the only thing it impacts. The midtones are impacted as well. The tones halfway between an 18% gray midtone and the shadows are affected. The highlights are affected. Every single tonal level in the image is affected. It isn't just the shadows. I'll see if I can create a graphic to demonstrate, maybe that will explain better than words can (they do say a picture is the embodiment of 1000 words...)
So, for a camera like the 1D X, with over 38e- read noise at ISO 100...could it's image quality be improved? Are we assuming that just because the 1D X has big pixels and a big FWC that it is already delivering the best IQ possible?
The 1D X has an FWC around 91ke-. With 38e- read noise, it's got around 11.26 stops of dynamic range. What happens if Canon could reduce that read noise to 3e-? Dynamic range jumps to 14.94 stops. FOURTEEN POINT NINE FOUR STOPS.
Ever looked at the OOF background blur of a bird photo taken with the 1D X? A wildlife image? Ever looked at the boke in a macro photo taken with the 1D X? Ever compared that background with the background of similar photos taken with the D800? Big difference....especially on a normalized basis. HUGE difference on a normalized basis. Creamy smooth, clean, practically noise-free OOF backgrounds from the D800, D810, D600, D750, A7r, A7s, etc. But even the mighty 1D X has visible noise in OOF backgrounds...
I don't deny that there are a lot of other things Canon could improve. I look at it this way, though. Canon has no problem with their AF system. Their AF systems are awesome...well designed, highly capable, top of the line in terms of frame coverage, very fast, and now meter-linked which allows even more advanced features. Canon has no problem with their metering. It's now (in the models that matter) a full color RGB meter with high resolution sensor, and advanced algorithms are hooked into the metering subsystem to enhance it's effectiveness. Canon has no problem with their ergonomics, nor their button placements (they are sublime, if I do say so myself!) Canon has few problems with their lenses (wide angles suffered a bit in the past, but Canon certainly seems to be remedying that problem.) Canon now currently holds the title for highest resolution camera...certainly no problems there.
Could canon make improvements in those areas? Probably...they have, even. The 7D II increased the resolution of the iTR meter sensor. It also expanded the AF point count a little bit above and beyond even the great 1D X. But, how much do those improvements actually matter? Outside of a tiny bit of jitter in a couple cases, I've never even had problems with my 5D III AF/meter system, and it doesn't even have the iTR system. The biggest focus issue I have is that the lens hunts forward first, rather than outward first, when I try to photograph BIF (would love to have a little switch on the lens for that, but that's a small thing.)
The single biggest gain for Canon, given that they already excel in every other area, would be to address the one area they don't excel at: read noise. Change that one thing, and the 1D X could skyrocket from delivering a mediocre 11.26 stops of DR, to delivering 14.94 stops of DR. Bam. Suddenly they actually have a NEED for a 16-bit ADC, because they can actually use the bit depth offered by it. Suddenly they rocket forward as the premier camera model with the highest dynamic range...but not only that, it's also paired with all the other excellent technology I just mentioned above. Who could possibly complain about that!
I see Canon as having one single weak area. As far as I am concerned, they don't NEED to "fix" anything else, because nothing else is broken. Canon needs to improve their read noise issues, though...because, well at least in my own personal opinion, it's the one single thing that Canon really sucks at. ;D
Upvote
0
50MP is an evolution not really a breakthrough. It's 8K vs 7K lines. If you had say, gone 80-100MP then maybe that could be called a breakthrough. As it stands, it is just a scaled up APS-C pixel pitch density from years ago.
you want a breakthrough, here are some ideas:
-15 stops of DR, or 16 or maybe even higher DR at higher ISOs to avoid the DR nosedive at high ISOs.
-single exposure HDR by taking a high MP sensor and reading it at variable ISO per line similar to what magic lantern was playing with
-a non bayer sensor of the likes of foveon
-full readout high MP 4K video sensor. do away with pixel bins, line skipping nonsense
-a real bump in pixel pitch density. APS-C should logically go to ~40 and full frame to ~70. THAT's a breakthrough
-really, really low ISO base. I'm talking 16 or so enabling high aperture lenses with sync speeds of 1/250 under the full sun.
-while we talk sync speed, how about a new shutter, or electronic shutter that is as good as current shutters but has a sync speed of 1/8000, THAT'S a breakthrough.
-Medium format
so on. Sorry but 50MP is just keep up. It's the next step and it was predicted long ago that it was going to happen. A breakthrough changes everything. 50MP won't. In fact I'm going to predict that a consumer friendly 80-100MP camera will show up in less than a year just because the technology is there already and higher pixel densities have been around for a long long time.
you want a breakthrough, here are some ideas:
-15 stops of DR, or 16 or maybe even higher DR at higher ISOs to avoid the DR nosedive at high ISOs.
-single exposure HDR by taking a high MP sensor and reading it at variable ISO per line similar to what magic lantern was playing with
-a non bayer sensor of the likes of foveon
-full readout high MP 4K video sensor. do away with pixel bins, line skipping nonsense
-a real bump in pixel pitch density. APS-C should logically go to ~40 and full frame to ~70. THAT's a breakthrough
-really, really low ISO base. I'm talking 16 or so enabling high aperture lenses with sync speeds of 1/250 under the full sun.
-while we talk sync speed, how about a new shutter, or electronic shutter that is as good as current shutters but has a sync speed of 1/8000, THAT'S a breakthrough.
-Medium format
so on. Sorry but 50MP is just keep up. It's the next step and it was predicted long ago that it was going to happen. A breakthrough changes everything. 50MP won't. In fact I'm going to predict that a consumer friendly 80-100MP camera will show up in less than a year just because the technology is there already and higher pixel densities have been around for a long long time.
Upvote
0
Similar threads
