Do Canon still develop Full Frame Cameras?

[NotABunny]

Actually, a low resolution sensor is a wet dream of people who keep dreaming that the laws of physics are bendable, and who belive there is this magic recipe for low noise levels. There is none!

Tell me you're joking :-)

I am not joking. I am simply stating that you're using the wrong comparison criteria: individual pixels instead of ENTIRE photos.

Is that why camera makers are increasing pixel count at the expense of clarity? Because the market doesn't realise that a clean 10 meg image trumps a noisy 20 meg image every single time?

There was some post here to a list of images of the same subject taken with cameras from (I think) 10D to 5D2. Those showed that you claim about sharpness is false. It was the opposite. The newer the technology (which coincidently also meant more pixels), the sharper the photos were.

Noise is the variation between how each photo site reads the light coming in. The smaller the site, more variation between what one pixel thinks the colour is, and what the next one thinks it is.

I don't care about pixels. I care about the ENTIRE image. (See below.)

You mean the pixels that make up the image itself has no effect on noise level in the photo?

Correct.

I give up, I truly do :-)

That's your choice, but you'll not understand the difference between what noise means for individual pixels and what it means for ENTIRE photos.

[YoukY63]

You only have to look at the xxxD series to see that, without even looking at the other glaring example, 20D, 30D, 40D, 50D, 60D.

So people willing to spend $1000 or less have had 6 cameras in the same timeframe as the pros who spend $3-6000 have had 2.

Am I the only one that doesn't feel pros are a priority anymore?

That is a very good example, thank you very much. So people that bought all the XXXD or the 20D, 30D, 40D, 50D, 60D spent 6000+$. Finally, wat did they get? During this 6 years, they just get a camera which do not provide yet a better picture than the 3000$ 5DmI that a pro would have bought 6 years ago.
So finally, upgrading XXD bodies every year during 6 years did not even catch up the antediluvian pro body IQ, not even talking about the "old" 5DmII.

Maybe that's why real professional don't care about getting a "new" body every year.
Then, if you feel better pro with a 600D because it is more recent, please, go ahead. 

[plam_1980]

I could not agree more with jakeymate. I have been willing to move to FF for an year, but have been put down by the constant rumours and expectations for 5d Mk3. I am willing to do this only as a hobby, not for professional work, so I am not pressed by time and am not willing to pay that money for 5d mk2, when its replacement is (over)due

[Kit.]

In current camera sensors, the size of the pixel (or resolution) has no VISIBLE effect on the noise level of the ENTIRE photo (= it only affects the noise level of the individual pixels). The only thing that matters is the size of the sensor (well, obviously, besides technology).

In current camera sensors, yes. Because it's way too expensive and way too slow to have a global shutter and to average several readouts of the same pixel on 18 million pixel sensors.

But for the hypothetical 2 megapixel full-frame video sensor it's feasible. And such sensor would also not really need microlenses.

[NotABunny]

The photo is pixels, if they are noisy, then the photo is noisy.

This is not how noise works in images. When you average noisier but MORE pixels, you get the same noise level. The noise per AREA (like square millimeter) is maintained.

All you have to do is look at actual images of the same subject shot in the same conditions.

In current camera sensors, the size of the pixel (or resolution) has no VISIBLE effect on the noise level of the ENTIRE photo (= it only affects the noise level of the individual pixels). The only thing that matters is the size of the sensor (well, obviously, besides technology).

In current camera sensors, yes. Because it's way too expensive and way too slow to have a global shutter and to average several readouts of the same pixel on 18 million pixel sensors.

But for the hypothetical 2 megapixel full-frame video sensor it's feasible. And such sensor would also not really need microlenses.

I'm not sure I understand you. By "several readouts" do you mean several "exposures"? Obviously having more light per photo gives you lower noise.

Or are you saying that averaging several reads gives a more accurate (/ average) read? (But this has nothing to do with resolution, at least not in theory, though in practice in may be the only workable way.)

[Peter Canon]

It's so funny to see people get so emotional over cameras.

[NotABunny]

That said, skin in shadows looks a mess on the 5D. It doesn't fall neatly to black, but blotches horribly in the blue and red channels. They all have to be smoothed and/or crushed.

There you have it. For the shadows, the image taken at ISO 1600 has a lower noise level (than the image taken at ISO 100). The guy combined two shots (ISO 100 and 1600) to get a better image across all brightnesses. Noise works in mysterious ways.


Here is the guy's explanation:

The sensor doesn't know what ISO gain is going to be used, it just records whatever photons arrive, leaving it to circuitry off the sensor to amplify the signal and digitize it. This means that the sensor sees the *entire* range of the figure -- the upper bound or "envelope" of all the different curves. The sensor has about 14 stops of DR, but the limitations of the rest of the circuits allow the final raw data to see less than twelve stops of DR, and the user is forced to choose a "window" of EV within that 14 stop range by selecting the ISO gain.

[Kit.]

I'm not sure I understand you. By "several readouts" do you mean several "exposures"? Obviously having more light per photo gives you lower noise.

By "several readouts" I mean several readouts. I.e. several attempts to amplify and digitize the same charge accumulated by the photo cell.

Light (shot noise of the arriving photons, to be exact) is not the only source of the noise in the pictures. Another source is the read noise (thermal noise in particular) of the sensor+amplifier. And it is relatively larger for smaller cells. You can average it between several cells of a higher-resolution sensor, but you can average it between several reads of the same cell of a lower-resolution CMOS sensor as well.

[NotABunny]

I'm not sure I understand you. By "several readouts" do you mean several "exposures"? Obviously having more light per photo gives you lower noise.

By "several readouts" I mean several readouts. I.e. several attempts to amplify and digitize the same charge accumulated by the photo cell.

Light (shot noise of the arriving photons, to be exact) is not the only source of the noise in the pictures. Another source is the read noise (thermal noise in particular) of the sensor+amplifier. And it is relatively larger for smaller cells. You can average it between several cells of a higher-resolution sensor, but you can average it between several reads of the same cell of a lower-resolution CMOS sensor as well.

That's a neat trick, but my instinct says that can't work. (Sure, my instinct said that a photo at ISO 1600 can't have a lower noise in the shadows than one at ISO 100.)

I mean, if the electrical charge sits there for longer, isn't it more affected by the thermal noise? (I mean, doesn't it degrade in time?)

Do you know of anyone who has such a sensor?

[kirillica]

jakeymate, I'm fan of you!  post more, it's pleasure to read you

[kubelik]

Yes there's a complaint, it took 3 years, IF it arrives this year. That is my whole complaint.

...

On a 3 year cycle, you're happy but I am not. Clearly :-)

you nailed it.  I did not get that you were complaining about release cycles as a whole, big picture thing.

honestly, it makes me happy to hear that there are photographers doing well out there that have the money and inclination to purchase new gear or an annual or 18-month basis.  when all we hear about in the industry is doom and gloom and slowdown, frankly it's encouraging to see people doing well by doing good work.

I really like the work on your site; I definitely also believe we have dramatically different noise tolerances.  for instance, even at web-view sizes, you dirt biking photos show clear signs of noise reduction.  everyone's taste is different, and I'm not going to second-guess yours, since clearly you're doing strong business with the work you're putting out there.

I don't think anyone is going to deny the red-channel issues, or the banding on the 5DII.  I like the grain from the 1DIV far better and I think even that can be improved upon.

overall, I still give canon the benefit of the doubt.  I'm willing to believe that the reason they don't do annual or bi-annual updates is because they'd rather dedicate resources towards large advances instead of incremental advances that defer resources away from major breakthroughs.  I have no idea if in fact they have enough operating budget to do both.  believe me, I'd love to work in Canon's financial department for a month and see how they allocate capital.

[kirillica]

jakeymate, I'm fan of you!  post more, it's pleasure to read you

Really? I think I've posted more than enough for one lifetime :-)

Really. Some people have already mentioned here, that you should write poems or so 
BTW, checked your site. It looks awesome, but IMO too much skin glow/toning 

And (or moderator ban me for offtopic): I was waiting for 5dmIII and, disappointed, bought 5dmII approx half a year ago. Yes, I'm happy with the camera and it's fantastic. But I still ready invest some more and upgrade to 5dmIII: please deliver it until I'm not "overburned"

[NotABunny]

I'm not sure I understand you. By "several readouts" do you mean several "exposures"? Obviously having more light per photo gives you lower noise.

By "several readouts" I mean several readouts. I.e. several attempts to amplify and digitize the same charge accumulated by the photo cell.

Light (shot noise of the arriving photons, to be exact) is not the only source of the noise in the pictures. Another source is the read noise (thermal noise in particular) of the sensor+amplifier. And it is relatively larger for smaller cells. You can average it between several cells of a higher-resolution sensor, but you can average it between several reads of the same cell of a lower-resolution CMOS sensor as well.

That's a neat trick, but my instinct says that can't work. (Sure, my instinct said that a photo at ISO 1600 can't have a lower noise in the shadows than one at ISO 100.)

I mean, if the electrical charge sits there for longer, isn't it more affected by the thermal noise? (I mean, doesn't it degrade in time?)

Do you know of anyone who has such a sensor?

I realized that this goes hand in hand with the possibility to have hand-held HDR by having multiple ISO readings for the same exposure.

I don't know how the reading is actually done, so I initially thought that this can be done only by splitting the sensor in a grid 2 * 2, where each grid pixel is read with a different ISO (by a different AD converter). Of course, it might be possible to do a parallel reading (with multiple AD converters) for EVERY pixel (though I wonder if the electron charge can be properly split among all converters).

This would mean that it may be possible to have both hand-held HDR using different ISOs, and better (= averaged) reading for the same ISO.

[japhoto]

I guess I'm a bit late since the OP has removed his post, but I'll throw something in anyway.

I switched from Olympus to Canon last year and man was I torn between choices back then.

I didn't care so much about going with a mainstream brand, but still Canon and Nikon were the two choices. I actually wanted a good supply of used equipment since with Oly, you had to buy everything new.

I looked at both brands long and hard back then. And even though I went with Canon, I do think that Nikon makes better cameras (even with that said, I'm not switching camps any time soon though).

The gripe I had with Nikon was that damn, they don't get anything done. Canons line seemed brand spanking new at that point when compared to Nikon (at least on paper).

I actually wanted an APS-C - camera since I do mostly nature photography and the extra reach is always welcome there.

So Nikon has the D300s (no D7000 at that time), which is based on technology from the original D300 (2007) and was released in 2009. And it still hasn't been replaced. That's 4 years which needless to say is a damn long time.

My Olympus E-3 was also from 2007, so no chance I'd trade in my camera for a model which is as old as the old one.

The D700 is from 2008 and would not make me want to switch to that either. Same goes actually with the 5D MkII, so even-steven there.

If they had released the D7000 back then, I might be with Nikon now, but somehow I think I would still have wanted the D400 or what ever it may be once it's released.

It's actually funny, I stated that Nikon makes better cameras and the people who use them probably know this for a fact (otherwise nobody would buy their cameras since they are so old), but when you come from an outside brand, they just don't have an appeal.

In the end I went with the Canon 7D, quite happy (except with the low ISO noise) but that was the only APS-C - camera at that time which interested me and that hasn't changed.

After all this time, I think and at least hope that there's improvements to be put in new cameras, now all we need to do is stop the manufacturers playing the "I'll show you if you'll show me yours first" - game and get something out.

I don't think it's going to be a revolution, not anymore, but evolution is also good. On the other hand, I don't want constant updates for every little evolutionary step that comes along.

Even though the prosumer to pro market isn't that big, it's still relevant since it keeps the line-up appealing and when you get someone to buy a dslr, they are buying into a system which is important for the company.

A bit of rambling there, but that's just me