hi, just stumbled upon this...
This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.
You say 2 whole stops of noise at high ISO?The technology of image sensors is ripe, and we should expect small improvements. Something like 1/3 stop better seems quite realistic.If Canon does not change their overall architecture, I agree...1/3rd stop at best. The A7s is an indication of what's possible, and it's more than two stops better than the 1D X at ISO 51200. If Canon improves their overall architecture, I really hope we see at least a 2-stop improvement across the board, low and high ISO.
Are you sure you were not talking about 2 stops of DR?
DR, lower noise. Same thing. We're primarily talking photon shot noise here. Assuming you have a perfect system...one that does not introduce any read noise, then more dynamic range or lower photon shot noise, there is no difference. It's the same thing, because dynamic range is the ratio between the maximum signal power and noise level. Read noise is just additive with photon shot noise, so it's the same thing in the end...lower noise, higher DR...same thing.
Keep in mind...noise is a random, undesired deviation from the proper, correct sample value. It's a range itself, +/- X, around the correct signal value. At higher signal power, that +/- X becomes less significant, so at higher ISO, increasing your maximum signal power becomes increasingly important. This has nothing to do with shadow pushing anymore...it has to do with overall signal quality. This is why pixel size becomes increasingly important at very high ISO settings, where as it is not really important at all at low ISO settings...at low ISO settings, there is (usually, assuming you expose properly) plenty of light, and it's read noise that becomes the most dominant source of noise. Bigger pixels gather more light, it's basically real-time averaging, like downsampling an image, only at the time the image is created.
At higher ISO settings, read noise drops to effectively imperceptible levels. Once your at ~3e- or lower, read noise is pretty trivial (so if you start out at 3e- at base ISO, like the D810, your doing really freaking good!) Photon shot noise completely dominates, and the weaker the signal, the more dominant noise will be. That is where the A7s pulls ahead of everything else right now for overall signal performance across all ISO settings. It has great low ISO DR (not the best, but excellent, at 12.9 stops), but it has the signal strength at high ISO to produce better images. It's got a maximum saturation (signal strength) of 323e- at ISO 51200, paired with a phenomenal read noise level of 0.7e-. In contrast, the 1D X has a max. sat. of 163e- at ISO 51200, paired with a still great, but still more DR-diminishing than the A7s, read noise level of 1.7e-. The A7s has twice the signal power and less than half the read noise. Ignoring the read noise entirely...you still have twice the signal strength, which in and of itself is a full stop improvement.
Also keep in mind, noise levels are additive. The 0.7e-/1.7e- RN levels are primarily additive with photon shot noise, and with the signal itself...so again, those read noise levels affect the whole signal, not just the shadows.
So, it doesn't matter what you call it, really. Lower noise, higher DR...same difference. It all means a better image in the end.
Can't really speak too much about modular components, but I'd love to see a video-focused cinema DSLR camera with something in the ballpark of:
• $3k - $4k price range
• Full frame sensor ~12mp
• Tilt touch screen
• 1080p 60fps (4K would be nice, but probably not going to happen)
• MPEG2/AVCHD codec options
• Clean 4:2:2 out (again, 10 bit would be nice but probably not happening at sub $4k)
• No time recording limit
• Focus peaking/zebras
• 19 point all cross type auto focus
• ISO range 100-51,200 (expandable to 50-204,800)
• Smooth ISO ramping auto-exposure like the Nikon D810
• Mic input with dedicated gain knob
• Headphone output
• Built-in WiFi
In my dreams... I know
What would you guys like to see in a DSLR cinema camera that would win you over more than the 1Dc?
$5k is my max limit for any 35mm format camera body.very wise move rick. mine is a 5D. Paid the premium 3.5k in August 2012, but I wanted to make those photographs at ISO 6400 which my trusty 30D couldn't do ;-) 3.5K to 4K max. And next time: i'll bite the bullet the year after release... As my only interest is better high ISOs. Regards, Peter
Announcement reminds me of the sensor they announced in development last year that could shoot in 0.01 lux... They had a video of firefly's shot in near pitch black.
So perhaps a new camera great in low light?
High ISO NR is not applied to RAW files, nor is ALO. Long exposure NR is applied to RAW, with that setting the camera takes a dark frame after the exposure (a 40 s exposure means a 40 s dark frame), then subtracts the dark frame before writing the RAW file. That effectively removes hot pixels, but it can actually add random noise.
I'd turn IS off to save battery power during long exposure shooting.
I wonder if the anti-flicker thing is a video or stills technology. If it's video then, meh. If it's meant for stills to help handle low frequency florescents then that's a bit of a game changer for indoor sports.
The 6D is much more capable when it comes to dynamic range than the Crop canons I have used.
The croppers are better than they're being given credit for.
This is a random 160 ISO 70D CR2 I picked up off the web somewhere.
And this is how it looks (white balance not addressed) with the shadows slider pushed to 100 and a stop of +ve EC added in Capture One - about 5 stops in total on the shadows. Default NR, no other changes made.
Looks OK, doesn't it? All the detail that was hidden in shadow is there.
Here is a 100% before crop; and here's after.
Nothing much wrong with this. Yes, there's a little bit of "texture" where the shadows were darkest; but the detail is all there and there's no pattern noise.
Pointedly, it's not at all far from what the "magic" Sony sensors are capable of...
The thing your missing is that detail buried in "shadow" isn't the problem. It's detail buried in READ NOISE that's the problem. "Shadows" extend for hundreds to thousands of levels...read noise usually only intrudes a dozen or so levels into the deepest of shadows. It's those very deep shadows that mark the difference between a Canon sensor and an Exmor.
The image you edited is more along the lines of this:
Being lifted to this with a +3 stop pull:
The "shadows" are not totally buried in the read noise...so they are nearly fully "recoverable". Which is a very far cry from this:
Being lifted to this with a +5 stop pull:
The detail here is not just "in shadow"...the detail is completely buried well into the noise floor on my 5D III sensor. It's near impossible to recover...it's riddled with red banding, and the noise level in the deeper shadows is two or three times what it is anywhere photon shot noise limited, as it'c compounded with dark current and various forms of noise caused by readout. It's these extra two stops...which, when you get right down to it, don't actually represent a ton of tonal levels...maybe 10-30 at most, if that...it's these deep shadow levels where the read noise exists that mark the difference between a Canon sensor and an Exmor.
The inevitable question that comes after I try to make this clarification (either when defending Canon, or when defending Exmor...these days it doesn't seem to matter), is: Why would you ever need those extra two stops? or How often do you actually have detail buried that deeply?
Sure, this isn't an every-day case. Some types of photography NEVER encounter a situation like this because they are always shot at higher ISO where DR is limited by physics. That said...even if the situation is a niche situation, it doesn't change the facts. The facts are...Canon sensors currently suffer about a two stop disadvantage or handicap compared to Exmor, and a one to two stop disadvantage compared to many other sensors on the market, such as the Toshiba sensors used in Nikon's D5000 line.
The next statement that is usually made is, the detail buried way down near the read noise floor in a Nikon RAW file can't possibly be clean, low noise, high color fidelity detail. Here are two images I downloaded way back around the time the D800 was released (shortly after the 5D III was released...about the time Fred Miranda wrote his review of both...I was looking for some real evidence to prove, either way, whether Exmor had better low ISO DR):
These two images, which are large (click on them for full size) show the exceptional quality of detail you can recover out of deep, black shadows.
The images were originally downloaded from here (although the link seems to be dead now):
The difference between current Canon sensors, and an Exmor, is not "you can lift shadows". We've been lifting shadows for years. The difference between the two is that you can lift ultra deep shadows that would otherwise be buried in red-banded read noise on a Canon, by 5-6 stops on an Exmor...and STILL have decent color fidelity and clean, random noise.
If you need that kind of shadow recovery capability...and I stress IF, I still believe a majority of photographers tend to shoot some kind of action at higher ISO, in which case Canon sensors have a very small edge over Exmor sensors at higher ISO...then don't hold yourself back. Canon may release something that can do what a D800 could do over two years ago at some point...but why wait until that unknown future time, when you could solve your problem now with a Nikon, Sony, or Pentax (or any number of other cameras that use Exmor sensors)?