Is Canon now two generations behind Nikon?

[LetTheRightLensIn]

Again jrista, I don't see it. You've have no reason at this point not to adopt the a7r, it's a fantastic camera with basically identical IQ that doesn't require a full adoption. I like end results and the a7r gives a better end result than 5d3 for your landscapes.

Call it in, and hang up the phone on the DR posts. I mean you probably could have wrote war & peace with a stop wedge by now.

I think all jrista wants is the DR of the Sony sensors in Canon bodies, because the Sony bodies have a wealth of disadvantages that far outweigh the sensor (IMO at least). You can't take a picture with just a sensor, and Sony is inferior with nearly all of those other aspects of the camera/lens. If Canon incorporates an improved sensor in a future EOS camera, it would be the best of both worlds.

Best of both worlds is what I'm after. In the long run, if Canon doesn't drop a new sensor into their models next year, then an A7r might just be the solution. I agree that Sony bodies have a wealth of disadvantages...but, ultimately, I only really need it for landscapes...so the majority of those disadvantages (except the crappy RAW format) would really be a problem.

Same here, I'll wait for the 5D4, but if that doesn't deliver I'll have to just nab an A7R and live with the dual-brand body mess for now and think about an ultimate shift to Nikon.

 

[Skulker]

You keep being very negative about the 5D3. While you may be expressing a sincerely held opinion you are going to struggle to convince many that your opinion is well founded. There are so many excellent photographs taken with the 5D3. I have seen images taken with Nikons, some of them outstanding, and I have seen and taken images with Canon. What I have never seen is anything that convinces me that either has a fatal fault or weakness.

I wouldn't call it a fatal fault. It's just an excessively annoying fault. My 450D had it. I skipped the 5D II because of it. The 7D has it, although not as bad. The 5D III has it, about as bad as the 5D II (despite the decent number of years difference between them.) So, no, not a "fatal" fault. An ugly, nasty, frustratingly annoying fault, yes.

Don't get me wrong. Some of the excellent images I have seen were yours. (I think your astro photography is excellent) But your logic is not too hot. To say you have been frustrated by Canon for over 6 years but its only the 5D3 just isn't logical. I know that you can say you only said you were dissatisfied with the 5D3 and frustrated for years. But hey we both know that's just wriggling.


Its a shame you have been subjected to quite so much agro. But I think you have rather asked for it by the way you have expressed yourself so strongly without taking much notice of many valid points that have been raised.

Eh, I knew what this community was like when I voiced my opinion. To be frank, I never really expected anything else. We crucify anything DR related, PARTICULARLY DXO (although, I still think DXO is asking for it...they need to stop being so obscure about their methodologies and weighting, and stop posting ludicrous lens test comparisons.) Anyway, people are what they are...and here they like to crush any mention of DR differences between Canon and the competition.

As for the rest...I was frustrated with the 450D read noise, that was my first DSLR. I did not like the 7D noise either, and was frustrated with it for over a year until I got the whole Topaz filter collection for about $130. DeNoise 5's debanding works extremely well on the 7D, since it has a very regular 8-column repetition. I can just set the band width to 8, and DeNoise 5 completely eliminates it. The low read noise then means the remaining random noise cleans up well. So, I was frustrated with the 7D until DeNoise came along.

Part of my frustration with the 5D III is it seems to have largely random banding. Some bands are very thin, some are quite fat. DeNoise 5 cleans up some bands, and not others. I can run multiple passes, but then I'm eating away at detail. So, the tools I used to use to deal with banding don't work nearly as well or at all with the 5D III. The color noise is also quite bad...however last night I found a new "smoothness" slider in LR 5's color NR that seems to deal with the larger-scale blotchiness...so that may help with the issue. (Crosses fingers.)

Anyway, I had frustrations with Canon read noise a long time ago. I skipped the 5D II because of it's read noise (at the time, the rumors here were that the 5D III would hit around 28mp and have improved DR...so I waited.) I'm harping on the 5D III because it's one of Canon's newest high resolution full frame sensors. It's their current technology. The 6D performs remarkably better at high ISO...and statistically given it's read noise levels, it should perform similarly at low ISO (I don't know if there are any low ISO comparisons between those two cameras...everyone focuses on the high ISO differences.) So, my reasoning is logical. I've been WAITING a long time for Canon to fix their banding issues...and the camera I have in hand right now is the 5D III.

J - I don't think you are listening. You're posts about the 5D3 just seem so un-balanced its a shame.


Have you ever heard the saying "when you're in a hole stop digging"

 

[jrista]

J - I don't think you are listening. You're posts about the 5D3 just seem so un-balanced its a shame.


Have you ever heard the saying "when you're in a hole stop digging"

Aside from the fact that I'm saying the 5D III's low ISO is hideous and unacceptable to me, just as unacceptable as every Canon sensor that came before it (I understand people disagree with the notion that is an issue)...what is unbalanced about my posts?

I can like the camera at high ISO, and not like it at low ISO. That isn't unbalanced, given the read noise levels within those two ranges of ISO are considerably different (<6e-, and generally <3e- at high ISO, as much as 33e- at low ISO...up to a 1000% difference between the two).

 

[psolberg]

D800 -> D810

The 5dm3 is looking really old.

I can't say the 5DIII looks old. From my perspective there was never an proper upgrade to the MKII which targeted studio and landscape photographers with state of the art resolution and not mind about fps or AF that much. That's why I've shoot D800/e now and for a number of years to great results. No regrets. What was a wining formula for canon is what Nikon followed. Nikon just had the common sense of designating it as a different product, unlike canon. So the 5DMKIII is really just AHEAD of its time but has a weird name. Maybe it should have been called 3D? feel better now? What's really old is the 5DmkII which to date has no true follow up following the tradition of leader/leading in detail and image quality over FPS and AF. That's the rumored 3D, but whatever.

The only area where I agree the 5DMKIII is severely lacking is video. With basically everybody going crazy over 4K, a 1080p camera is just lacking. Then again, nobody really considers 5D's to be the state of the art in video with many manufacturers offering much more powerful options aimed at the videographer instead of the photographer wanna-be-video guy. That's the right approach IMHO. so who really cares that the 5D isn't going to be the video guy's sweetie. It's a photo tool first and foremost.

 

[jrista]

Best of both worlds is what I'm after. In the long run, if Canon doesn't drop a new sensor into their models next year, then an A7r might just be the solution. I agree that Sony bodies have a wealth of disadvantages...but, ultimately, I only really need it for landscapes...so the majority of those disadvantages (except the crappy RAW format) would really be a problem.

Same here, I'll wait for the 5D4, but if that doesn't deliver I'll have to just nab an A7R and live with the dual-brand body mess for now and think about an ultimate shift to Nikon.

Personally, I don't contemplate any kind of shift to Nikon. There are many reasons not to...manufacturing issues, customer support issues, Canon lenses in large part (particularly longer lenses) being better, many unique lens options for Canon.

My sole complaint is with Canon's low ISO performance. I guess I could live with an A7r until Canon gets that sorted out. Not ideal, and an extra cost...but it's at least a viable interim option. But moving to Nikon overall? That's never been in my playbook. Canon cameras still perform excellently at higher ISO, once their data is all above the read noise floor, and Canon's RN floor at high ISO tends to be lower than Nikon & Sony's RN floor at high ISO, so you have more high ISO DR.

I just hope Canon DOES sort it out...

 

[3kramd5]

Getting really good landscapes is actually a lot of work. You have to nail the day and time...

Indeed. I suspect that if you do eventually pick up a D8xx or canon reduces shadow noise to roughly equivalent levels, you'll find it merely marginally better since you are an afternoon landscape guy. I know I did.

 

[psolberg]

I guess I could live with an A7r until Canon gets that sorted out

why? that camera's raw files are severely crippled as shown by Lloyd chambers and others. basically you're shooting 11 bit crap. All your reasoning (lenses and options) applies to sony. Don't be foolish. I've made the switch and have used both the 800 and A7r. no real equivalence. It's just gear, not a religion. And if you already took the sony, you may as well really get true 14 bit raws. Plus if you only need the high MP for some situations you're using the A7r, just get the matching Nikon set and sell it later if you have to. It's not like you're cheating on a wife or something lol.

Getting really good landscapes is actually a lot of work. You have to nail the day and time...

Indeed. I suspect that if you do eventually pick up a D8xx or canon reduces shadow noise to roughly equivalent levels, you'll find it merely marginally better since you are an afternoon landscape guy. I know I did.

having switched I can tell you the difference will be there and it is quite obvious specially after you're used to a huge latitude in the RAW with LR and its excellent algorithms. This is no longer up for debate after two years. The question is that unless you have a burning desire, you can still take nice images with the 5DMKII. that didn't go away and realistically you're unlikely to say "oh crap, if only my gear was better!". The detail is amazing and IMHO, 50+MP can't come soon enough to oversample some of the issues around the bayer pattern and color graduations which still plague low MP bodies. But ultimately you have to balance it. Yes more detail is nice. Yes super deep shadow and highlight range is nice. But are you crippled otherwise? nah. I only switched because it is trivial to do so for me as both systems are basically the same in my lens selection and cost is similar.

I'm following closely the medium format scene. Looks like it will heat up as manufactures respond to the D810 price to performance ratio. This means cheaper or more capable medium format gear to justify the cost. And maybe now it will start to make sense to jump to it.

 

[dtaylor]

It's interesting that you keep going to IR as being the gods of testing and yet always fail to quote this part of the final remark: "Like all recent Canon SLRs, the higher quality scores are somewhat below average for a modern sensor. For example, the Nikon D7100 managed 10.1 f-stops at the highest quality level, almost 2 stops better. "

It's interesting that you still don't understand the difference between total dynamic range and dynamic range to arbitrarily selected noise thresholds.

And as for IR being sooo much more reliable and well defined than DxO and how can IR get so many stops more DR and the others are obviously wrong, etc. you always fail to quote : "As always, it's worth noting here is that ACR's default noise reduction settings reduced overall noise somewhat " I.E. they do a potentially randomly manipulated by ACR test that involves all sorts of NR.

It's also interesting that you choose to lie and misrepresent to try and make your case.

A) NR does not affect total DR, though it does impact latitude. Likewise it would impact DR measurements which used an arbitrarily high noise threshold or "quality."

B) ACR's default NR settings are mild and fixed. They are not random nor "all sorts of."

C) ALL tested cameras have default ACR settings.

D) DxO is not testing system DR, but sensel SNR. There is no simple or direct conversion of sensel SNR to system DR. If DxO measured film "sensels" or grain they would report a DR of <1 stop. Yet a piece of Portra film held 12-13 stops and with proper development some B&W films can hit 16-18.

You always use extravagant talk about minor high ISO gains, but then use the most radically minimizing talk about low ISO differences.

No idea what you're even talking about.

 

[dtaylor]

...
For tripod landscapes I can't see any disadvantage to an A7 or A7R. AF speed doesn't matter. EVF lets you judge exposure/histogram before shooting. Tilt screen is useful if your tripod is down low. What's the problem?
...

Hanging a big zoom lens on the front whilst tripod mounted is what the problem is.

A large lens will typically have its own tripod ring and foot. The EF adapters I've looked at also have a tripod foot so that the adapter bears the lens weight (for lenses too small to warrant a tripod ring).

How could you not know this? Hmmm...This is hilarious and reads like someone who has never actually taken a photograph. Do you actually go out and take any landscape photographs or are you just an Internet arm chair expert? :

 

[psolberg]

In the D810 Nikon also added a small RAW option. I really cannot understand Nikon's thinking in not offering a smaller raw in the D800, especially when those upgrading would be moving from 12 to 36 mp.

because there is no benefit to this. read this:
http://www.dslrbodies.com/accessories/software-for-nikon-dslrs/software-news/the-sraw-myth.html

basically there is no such thing as sraw, not even canon's as interpolation is used therefore technically not raw. In fact, if you see a D810 guy using sraw, please smack him in the head as he's being an idiot.

 

[jrista]

It's interesting that you keep going to IR as being the gods of testing and yet always fail to quote this part of the final remark: "Like all recent Canon SLRs, the higher quality scores are somewhat below average for a modern sensor. For example, the Nikon D7100 managed 10.1 f-stops at the highest quality level, almost 2 stops better. "

It's interesting that you still don't understand the difference between total dynamic range and dynamic range to arbitrarily selected noise thresholds.

You keep saying "arbitrarily" selected noise thresholds. It isn't arbitrary. It's very specific. It is the RMS of read noise, a figure that can usually be taken right out of a manufacturers specs for the sensor, and which is often tested for by reviewers, and which can be derived using tools like Imatest.

The noise threshold that is used to determine dynamic range is the farthest thing from arbitrary...it is very specific. The actual RMS value often changes from sensor to sensor...but that is exactly why it is used. Because not every sensor has the same DR. The DR is the range from the RMS of read noise, through the maximum signal strength. At base ISO, the maximum signal strength is the FWC. At ISOs higher than base, the maximum signal strength is the white point (before the signal is amplified.)

Very much not arbitrary.

On the other hand...using ACR/LR to perform conversions, when the underlying algorithms used in that product have changed, often considerably, over the years and countless minor version releases, not to mention the fact that the algorithms are black box, leads to highly suspect, and highly arbitrary, results.

The means of NR from one brand to the next may not even be the same. Adobe does not publish any of that information, so who knows if the way they NR a NEF is the same way they NR a CR2. It is also actually well known that the default camera calibration profiles applied by ACR are quite different than the manufacturer's default curves, and they differ from camera to camera.

That all increases the chance that IR results are entirely arbitrary, and inconsistent even within the same brand over time.

And as for IR being sooo much more reliable and well defined than DxO and how can IR get so many stops more DR and the others are obviously wrong, etc. you always fail to quote : "As always, it's worth noting here is that ACR's default noise reduction settings reduced overall noise somewhat " I.E. they do a potentially randomly manipulated by ACR test that involves all sorts of NR.

It's also interesting that you choose to lie and misrepresent to try and make your case.

A) NR does not affect total DR, though it does impact latitude. Likewise it would impact DR measurements which used an arbitrarily high noise threshold or "quality."

B) ACR's default NR settings are mild and fixed. They are not random nor "all sorts of."

C) ALL tested cameras have default ACR settings.

D) DxO is not testing system DR, but sensel SNR. There is no simple or direct conversion of sensel SNR to system DR. If DxO measured film "sensels" or grain they would report a DR of <1 stop. Yet a piece of Portra film held 12-13 stops and with proper development some B&W films can hit 16-18.

DxO tests the same RAW output as anyone else. I'm not a fan of most DXO results, however I do believe their Screen DR measures are accurate. How is a DXO Screen DR measure any different than a "system" measurement done by IR? Or, for that matter, anyone elses DR tests also performed on RAW file data? Aren't they all doing "system" DR tests?

 

[dtaylor]

You keep saying "arbitrarily" selected noise thresholds. It isn't arbitrary. It's very specific.

The Imatest "quality" DR measurements (low/med/medhigh/high) have arbitrarily selected noise floors that someone thought represented "quality" points. That's why there are 4 reported values in addition to total measured DR.

It is the RMS of read noise,

Not what we're talking about. (And for the millionth time, cannot be simply converted into system DR.)

On the other hand...using ACR/LR to perform conversions, when the underlying algorithms used in that product have changed, often considerably, over the years and countless minor version releases, not to mention the fact that the algorithms are black box, leads to highly suspect, and highly arbitrary, results.

Have you shot a step wedge and tried to use NR to make black and white steps gray yet?

No?

Please go try ;D

How is a DXO Screen DR measure any different than a "system" measurement done by IR?

http://www.stouffer.net/TransPage.htm

T4110

Buy. Shoot. Learn.

 

[3kramd5]

having switched I can tell you the difference will be there and it is quite obvious specially after you're used to a huge latitude in the RAW with LR and its excellent algorithms.

Of course the difference is there, I just find it to be marginal. Perhaps it has to do with what, or how I shoot as opposed to you. I don't bracket nor merge, I dodge, burn, mask, etc. I don't push shadows up to the point that they're obviously ugly. If you DO generally do that, you will find some more latitude in the exmor than the canon, however scenes wherein it's strictly necessary to stack exposures with the latter will likely still need it with the former.

 

[3kramd5]

A large lens will typically have its own tripod ring and foot. The EF adapters I've looked at also have a tripod foot so that the adapter bears the lens weight (for lenses too small to warrant a tripod ring).

It's problematic with the A7 due to camera shake from the shutter. It lacks an electronic first curtain, and long focal lengths exacerbate the problem for obvious reasons.

 

[deleteme]

29 pages now...

Any body actually making prints that can demonstrate the "obsolescence" of Canon?
I can agree that on screen various bodies will show some differences but I have not seen prints from anyone where I could say "Ahhh, clearly this fellow used a (insert favorite hot camera here).

 

[jrista]

It is the RMS of read noise,

Not what we're talking about. (And for the millionth time, cannot be simply converted into system DR.)

On the other hand...using ACR/LR to perform conversions, when the underlying algorithms used in that product have changed, often considerably, over the years and countless minor version releases, not to mention the fact that the algorithms are black box, leads to highly suspect, and highly arbitrary, results.

Have you shot a step wedge and tried to use NR to make black and white steps gray yet?

No?

Please go try ;D

I wouldn't consider that valid. Applying noise reduction means your injecting an arbitrary SOFTWARE factor into the process. If you measure after NR, then your not measuring the camera. Your measuring the camera plus what the software does to the camera's output.

There are countless noise reduction algorithms out there. Shooting a step wedge and denoising it only tells you how the "system", which includes the camera and whatever denoising algorithm you are using, performs as a whole. But it's specific to that denoising algorithm. And, therefor...arbitrary.

How is a DXO Screen DR measure any different than a "system" measurement done by IR?

http://www.stouffer.net/TransPage.htm

T4110

Buy. Shoot. Learn.

You mind sharing one of your unmodified RAW 41-step wedge shots? I'd like to experiment with it before I buy one, see what I come up with.

 

[jrista]

From this page: http://www.imatest.com/docs/q13/#dynamic

The following is stated:

The total range. Stepchart is extremely sensitive at detecting a camera’s total dynamic range, even when dark areas are extremely noisy or boundaries between chart zones become indistinct. This number is usually not a good measure of camera performance.

To repeat: This number (total dynamic range) is usually NOT a good measure of CAMERA performance. This page then goes on to state the following:

A range of tones over which the scene-referenced RMS noise, measured in f-stops (the inverse of the signal-to-noise ratio, SNR), remains under a specified maximum value. The lower this value (the higher the minimum SNR), the better the image quality but the smaller the dynamic range. SNR tends to be worst in the darkest regions. Imatest calculates the dynamic range for several maximum RMS noise levels, from 0.1 f-stop (high image quality; SNR = 10) to 1 f-stop (low quality; SNR = 1).
The dynamic range corresponding SNR = 1 (1 f-stop of noise) corresponds to the intent of the definition of ISO Dynamic range in section 6.3 of the ISO noise measurement standard: ISO 15739: Photography — Electronic still-picture imaging — Noise measurements. The Imatest measurement differs in several details from ISO 15739; hence the results cannot be expected to be identical. Imatest may well produce more accurate results because it measures DR directly from a transmission chart, rather than extrapolating results for a reflective chart with maximum density = 2.0.

This second definition, dynamic range measured from the RMS of read noise, or the point where SNR = 1, to the saturation point (in the case of Imatest, the saturation point as defined by ISO standards), is the definition that, as far as I know, everyone uses to define the dynamic range performance of CAMERAS.

We aren't talking about systems here. Systems can be...anything, arbitrarily complex, and this arbitrary period. We are only talking about CAMERAS, an when it comes to describing camera DR, "total dynamic range", according to Imatest themselves, is NOT a good measure of camera DR.



From this page: http://www.imatest.com/docs/q13/

The following is stated:

The detected patches have a density range of 11.2 f-stops (this should not be interpreted as the camera’s Dynamic Range).

To repeat and emphasize: (this should not be interpreted as the camera’s Dynamic Range)



I'm sure this doesn't settle the issue, but with this, explicit information strait from the source, I'm done debating "photographic DR" vs. dynamic range as it is commonly understood and used in the sensor and camera industry.

 

[LetTheRightLensIn]

It's interesting that you still don't understand the difference between total dynamic range and dynamic range to arbitrarily selected noise thresholds.

I do. How does that change when they directly said that the Canon was a good 2 stops behind and not up to the standards of other current sensors?

It's also interesting that you choose to lie and misrepresent to try and make your case.
A) NR does not affect total DR, though it does impact latitude. Likewise it would impact DR measurements which used an arbitrarily high noise threshold or "quality."

B) ACR's default NR settings are mild and fixed. They are not random nor "all sorts of."

C) ALL tested cameras have default ACR settings.

And of course ACR hasn't changed from release to release or how it treats camera to camera, nope, nobody ahs ever noticed any changes there....

D) DxO is not testing system DR, but sensel SNR. There is no simple or direct conversion of sensel SNR to system DR. If DxO measured film "sensels" or grain they would report a DR of <1 stop. Yet a piece of Portra film held 12-13 stops and with proper development some B&W films can hit 16-18.

Umm, no, they are measuring system DR, they don't measure the best the photosites can do and de-couple the sensor from the camera's downstream ADC and such, the Canon sensor sensor itself has plenty fine DR, but the downstream messes it up.

IR is measuring Camera PLUS RAW converter while DxO measures Camera.

 

[Hillsilly]

29 pages now...

My hobby is speaker building and on some audio websites, it's not unusual for threads to be 1000+ pages long. And there we're often debating sound quality characteristics differences that are highly subjective, can't be measured, and most people don't even believe exist. At least sensor performance can be measured.

 

[jrista]

29 pages now...

My hobby is speaker building and on some audio websites, it's not unusual for threads to be 1000+ pages long. And there we're often debating sound quality characteristics differences that are highly subjective, can't be measured, and most people don't even believe exist. At least sensor performance can be measured.

Hmm...anyone wanna aim for 1000 pages in this thread? ;D