October 01, 2014, 04:22:12 PM

Author Topic: 70D and Dxomark....  (Read 65471 times)

jrista

  • Canon EF 400mm f/2.8L IS II
  • *******
  • Posts: 4732
  • POTATO
    • View Profile
    • Nature Photography
Re: 70D and Dxomark....
« Reply #420 on: September 03, 2013, 11:32:33 AM »
I know everybody has invested in lenses and accessories, so jumping ship is not practical, but here is what that 3 stop of extra DR can mean (5D3 on top, D800 on the bottom):


All this really proves that even two of the best cameras currently made are each capable of producing suboptimal images when used incorrectly.

It isn't that the 5D III was used incorrectly. Fred Miranda was the one who did that review, and he is a respectable authority on such matters. The original sample images were shot to keep the whites from blowing in a high DR scene...and this, quite plain and simply, is indeed the result when you lift the shadows to compensate. There isn't any question that Canon has issues with their low ISO DR. It has been blotchy and banded for well over five years now, and it is Canon's weakest point.

Given DXO's review of the 70D, it's clear they have not put any effort into improving sensor IQ. Canon's focus is clearly different, and the weight of their focus has clearly shifted heavily towards video. I was hoping to see a better DR improvement for the 70D, and these results are rather disheartening. I'm a stills photographer, and I use my DSLR for still photographs. Quite frankly, I would really like to see Canon invest some of their huge cash flow into improving their products for their primary intended purpose, and not some secondary "sometimes convenient" purpose like video.

The Magic Lantern crew proved that Canon's sensors are capable of capturing the full 14 stops of dynamic range allowed by a 14-bit ADC...I think its high time Canon stopped their focus in video and returned their focus to EVERY aspect of IQ, including factors based on the sensor. I've argued in the past that there are other factors that affect IQ, such as AF system, but Canon has already perfected those technologies. Fred Miranda's comparison of the 5DIII and D800 is an excellent demonstration of why Canon should finally shift their focus back towards improving still photography IQ. I don't even think the problem is in the sensor, either...I think its in the ADC...and Canon pretty much whips up a new version of that every major release cycle anyway...so there shouldn't be any reason they couldn't invest some serious R&D into solving their low ISO IQ problem.

No more clever arguments from me about why Canon's products are good. They are, but that isn't the point any longer. I truly hope to see something significant on the sensor/ADC IQ front with the 7D II...otherwise, the 5D III will probably be my last Canon DSLR purchase until they stop focusing on secondary video functionality and refocus on the primary purpose of a DSLR: photography.

+1

The latest patent they have released (2013 for the newest variant, apparently they already had ones as far back as 2006 though) seem to make it clear they even have the designs to do it, or least largely so it appears, so they just need marketing to let the engineers start producing their DSLR sensor designs on Canon's more modern fabs or to build a new one instead of making the DSLR engineers have to work within the limitations of the very old 500nm Canon fab. But they have so far wanted to keep milking the old fab DSLR while all other other companies move on to new fabs for large sensors. (someone claimed they heard that Canon had shifted some of the newest P&S models to the older fab and speculated Canon had finally decided to use the 180nm fab for DSLRs now. Who knows. Just a rumor.)

I do have to say I do love the 5D3 video abilities though now that they have been unlocked by Magic Lantern. It's pretty astonishing. I don't think that really had anything to do with not getting better stills quality though (maybe costs it a touch in the MP count but then again it also got it to a very nice 6fps by staying lower, although I suppose they could've maybe fit in two digic since the 7D could fit two, but whatever)).

The new liveview AF for the 70D does sound it took quite a lot of resources up. It is very cool though. All the same if it was better sensor for IQ vs that I'd have takne the better sensor and then done this new AF later. For the 70D market perhaps the AF matters more. But in the end I think it's probably mostly just that marketing wanted them to keep milking the old 500nm line which probably meant there wasn't much the sensor engineers could really do.

I do hope Canon moves to 180nm soon. They certainly have the capability of building a 180nm fab, and they are on solid financial footing...its not like they would need to go billions into debt like Sony did in order to move to a new fab line.

If the marketing department really has that much control at Canon, then something is really wrong. Marketing shouldn't have a say in what technological developments a company invests in and deploys...marketing should simply sell what the technical departments of a company develop. So, I've never really bought the argument that marketing controls what Canon does. If there is some concrete information about who controls what at Canon that indicates their marketing department(s) have that much direct control, then maybe that is something.

For whatever reason, Canon has gone full-bore on video, and for years now. Ever since the 5D II hit the streets with video, it seems Canon thinks that is the path to financial glory. Sadly, it seems to have consumed them entirely, to the detriment of the original purpose of a "camera": to take still photographs. I think Canon has taken it about as far as it can go though, or almost. With full-sensor FPPDAF, I don't really know what else Canon can do other than maybe tie in the metering even more, include even more advanced face detection, and maybe go to quad-split pixel designs for both horizontal and vertical phase detection. After that...what else could Canon possibly invent in the video arena for their DSLRs that would actually be meaningful and competitive?

I just hope the 7D II brings some significant IQ improvements to the table. Across the board...low ISO DR, high ISO SNR/noise, increased resolution, higher frame rate, better AF. If nothing more than to show their customers that when the next generation of FF finally hits, it won't just be a tweaked 61pt AF system and the same old 12 stops of DR.
My Photography
Current Gear: Canon 5D III | Canon 7D | Canon EF 600mm f/4 L IS II | EF 100-400mm f/4.5-5.6 L IS | EF 16-35mm f/2.8 L | EF 100mm f/2.8 Macro | 50mm f/1.4
New Gear List: SBIG STT-8300M | Canon EF 300mm f/2.8 L II

canon rumors FORUM

Re: 70D and Dxomark....
« Reply #420 on: September 03, 2013, 11:32:33 AM »

hutjeflut

  • PowerShot G1 X II
  • ***
  • Posts: 55
    • View Profile
Re: 70D and Dxomark....
« Reply #421 on: September 03, 2013, 01:24:07 PM »
I know everybody has invested in lenses and accessories, so jumping ship is not practical, but here is what that 3 stop of extra DR can mean (5D3 on top, D800 on the bottom):




what does this have to do with DR it seems a iso comparison to me nothing more nothing less?

maby im missing something but you cant tell me these 2 are shot on the same iso setting and shutter speed.
« Last Edit: September 03, 2013, 01:26:34 PM by hutjeflut »

jrista

  • Canon EF 400mm f/2.8L IS II
  • *******
  • Posts: 4732
  • POTATO
    • View Profile
    • Nature Photography
Re: 70D and Dxomark....
« Reply #422 on: September 03, 2013, 02:12:26 PM »
I know everybody has invested in lenses and accessories, so jumping ship is not practical, but here is what that 3 stop of extra DR can mean (5D3 on top, D800 on the bottom):




what does this have to do with DR it seems a iso comparison to me nothing more nothing less?

maby im missing something but you cant tell me these 2 are shot on the same iso setting and shutter speed.

You should read the original review:

http://www.fredmiranda.com/5DIII-D800/index_controlled-tests.html

The noise exists because of a fairly significant shadow lift. The beams in the samples originally showed up as entirely black, and were lifted enough to bring out color and detail (at least a few stops). The rest of the photos were "correctly exposed", so the tests were not an ISO comparison...they were quite explicitly DR comparisons. Dynamic range is ultimately limited, in the shadows, by noise. Read noise, to be specific. Sony's Exmor sensor simply has less at low ISO (in exact terms, the 5D III has 33.1e-, where as the D800 has a mere 2.7e-).
My Photography
Current Gear: Canon 5D III | Canon 7D | Canon EF 600mm f/4 L IS II | EF 100-400mm f/4.5-5.6 L IS | EF 16-35mm f/2.8 L | EF 100mm f/2.8 Macro | 50mm f/1.4
New Gear List: SBIG STT-8300M | Canon EF 300mm f/2.8 L II

Aglet

  • 1D X
  • *******
  • Posts: 1016
    • View Profile
Re: 70D and Dxomark....
« Reply #423 on: September 03, 2013, 02:37:53 PM »
These are all largely pointless debates...Nikon vs. Canon color depth bits, minute 1-2% differences in SNR, etc.

I agree with pretty much everything you've stated but the SNR differences are much greater than 1-2%.  They may be that at the highlites end but SNR difference is actually huge where it matters in post, on the dark end of the curve.
there's a good 2-4 stops difference between Canon and …everybody else at 0db SNR.

jrista

  • Canon EF 400mm f/2.8L IS II
  • *******
  • Posts: 4732
  • POTATO
    • View Profile
    • Nature Photography
Re: 70D and Dxomark....
« Reply #424 on: September 03, 2013, 03:31:23 PM »
These are all largely pointless debates...Nikon vs. Canon color depth bits, minute 1-2% differences in SNR, etc.

I agree with pretty much everything you've stated but the SNR differences are much greater than 1-2%.  They may be that at the highlites end but SNR difference is actually huge where it matters in post, on the dark end of the curve.
there's a good 2-4 stops difference between Canon and …everybody else at 0db SNR.

Sorry, I wasn't clear. I was specifically referring to higher ISO settings, and the SNR 18% measure at DXO. At higher ISO settings, the differences are miniscule, if they exist at all. Canon maintains an edge thanks to a WEAKER CFA, but in the end color quality is so phenomenally good on all modern DSLRs those differences don't matter much. For high ISO users, other factors matter much more...AF system, metering, frame rate, etc.

At low ISO, improved SNR does indeed lead to more stops of dynamic range, no one disputes that.
My Photography
Current Gear: Canon 5D III | Canon 7D | Canon EF 600mm f/4 L IS II | EF 100-400mm f/4.5-5.6 L IS | EF 16-35mm f/2.8 L | EF 100mm f/2.8 Macro | 50mm f/1.4
New Gear List: SBIG STT-8300M | Canon EF 300mm f/2.8 L II

jrista

  • Canon EF 400mm f/2.8L IS II
  • *******
  • Posts: 4732
  • POTATO
    • View Profile
    • Nature Photography
Re: 70D and Dxomark....
« Reply #425 on: September 03, 2013, 07:54:25 PM »


Just to demonstrate a point. The Canon image in the samples above looks quite terrible. While you can probably never extract the same amount of shadow detail as the D800 with a 5D III, it need not look quite that bad. Here is a copy that was run through both Topaz DeNoise 5, with vertical debanding, 100% chroma noise removal, and minor luma noise removal, as well as maximum dynamic range recovery. It was then run through Nik Dfine 2 in order to remove more luma noise. DeNoise 5 does an excellent job removing chroma and banding noise, while Dfine 2 does a better job at reducing luma noise without smudging detail.

Here is the result:



Clear loss of detail on the upper beam relative to the D800. Still some noise on the door frame and window relative to the D800. Some artifacts on the no smoking sign (although more likely due to the use of a JPEG as my source...denoising from an original RAW or high quality TIFF would produce better results.) Despite those, the results are FAR more acceptable than the original 5D III RAW without any noise reduction. Dynamic range has improved considerably with some effective NR. Not the same 13.2 stops as the D800 has, but definitely more than the 10.97 stops that DXOMark measured.

So, yes, the 5D III, like most other Canon sensors, has some pretty crappy shadow noise. Doesn't mean the buck stops there, however.  ;D
My Photography
Current Gear: Canon 5D III | Canon 7D | Canon EF 600mm f/4 L IS II | EF 100-400mm f/4.5-5.6 L IS | EF 16-35mm f/2.8 L | EF 100mm f/2.8 Macro | 50mm f/1.4
New Gear List: SBIG STT-8300M | Canon EF 300mm f/2.8 L II

Pi

  • 1D Mark IV
  • ******
  • Posts: 937
    • View Profile
    • Math and Photography
Re: 70D and Dxomark....
« Reply #426 on: September 03, 2013, 08:17:31 PM »
Here is a copy that was run through both Topaz DeNoise 5, with vertical debanding, 100% chroma noise removal, and minor luma noise removal, as well as maximum dynamic range recovery. It was then run through Nik Dfine 2 in order to remove more luma noise. DeNoise 5 does an excellent job removing chroma and banding noise, while Dfine 2 does a better job at reducing luma noise without smudging detail.

Can you denoise now the crop from the Toronto's Yonge-Dundas Square shot I posted earlier without losing detail? If I like it, I may send you the whole image, and pay you for that.  ;)

canon rumors FORUM

Re: 70D and Dxomark....
« Reply #426 on: September 03, 2013, 08:17:31 PM »

jrista

  • Canon EF 400mm f/2.8L IS II
  • *******
  • Posts: 4732
  • POTATO
    • View Profile
    • Nature Photography
Re: 70D and Dxomark....
« Reply #427 on: September 03, 2013, 08:37:42 PM »
Here is a copy that was run through both Topaz DeNoise 5, with vertical debanding, 100% chroma noise removal, and minor luma noise removal, as well as maximum dynamic range recovery. It was then run through Nik Dfine 2 in order to remove more luma noise. DeNoise 5 does an excellent job removing chroma and banding noise, while Dfine 2 does a better job at reducing luma noise without smudging detail.

Can you denoise now the crop from the Toronto's Yonge-Dundas Square shot I posted earlier without losing detail? If I like it, I may send you the whole image, and pay you for that.  ;)

Hmm, I think I missed that one...I came into this thread in the middle. I can read from the beginning if its in this thread, and I'd be happy to give it a whirl. :)
My Photography
Current Gear: Canon 5D III | Canon 7D | Canon EF 600mm f/4 L IS II | EF 100-400mm f/4.5-5.6 L IS | EF 16-35mm f/2.8 L | EF 100mm f/2.8 Macro | 50mm f/1.4
New Gear List: SBIG STT-8300M | Canon EF 300mm f/2.8 L II

Pi

  • 1D Mark IV
  • ******
  • Posts: 937
    • View Profile
    • Math and Photography
Re: 70D and Dxomark....
« Reply #428 on: September 03, 2013, 08:38:58 PM »
Here is a copy that was run through both Topaz DeNoise 5, with vertical debanding, 100% chroma noise removal, and minor luma noise removal, as well as maximum dynamic range recovery. It was then run through Nik Dfine 2 in order to remove more luma noise. DeNoise 5 does an excellent job removing chroma and banding noise, while Dfine 2 does a better job at reducing luma noise without smudging detail.

Can you denoise now the crop from the Toronto's Yonge-Dundas Square shot I posted earlier without losing detail? If I like it, I may send you the whole image, and pay you for that.  ;)

Hmm, I think I missed that one...I came into this thread in the middle. I can read from the beginning if its in this thread, and I'd be happy to give it a whirl. :)

Please don't! It was a joke (the image is real though).

jrista

  • Canon EF 400mm f/2.8L IS II
  • *******
  • Posts: 4732
  • POTATO
    • View Profile
    • Nature Photography
Re: 70D and Dxomark....
« Reply #429 on: September 03, 2013, 08:44:20 PM »
where is the banding?
5D3
ISO 4000
extreme cropped

You shot at ISO 4000. Thanks to Canon's use of a Bias Offset, by that ISO, the black point has moved above the level at which banding noise exists. The only noise remainder is slight quantization noise during ADC...and photon shot noise.
My Photography
Current Gear: Canon 5D III | Canon 7D | Canon EF 600mm f/4 L IS II | EF 100-400mm f/4.5-5.6 L IS | EF 16-35mm f/2.8 L | EF 100mm f/2.8 Macro | 50mm f/1.4
New Gear List: SBIG STT-8300M | Canon EF 300mm f/2.8 L II

jrista

  • Canon EF 400mm f/2.8L IS II
  • *******
  • Posts: 4732
  • POTATO
    • View Profile
    • Nature Photography
Re: Black-Cap event
« Reply #430 on: September 03, 2013, 08:46:14 PM »
70D in stock at my local store.
Goin' in for some lens cap shots.

Case in point - this is how Aglet evaluates a new camera: 'pictures' with the lens cap on.  He'll then boost the black images 4-5 stops, and tell us all how crappy the 'pictures' are, and how Canon still hasn't made their sensors cameras any better.

Yawn.

raw files are pushed 4 stops

Do you look for bio-luminescent markers with the room lights on?

(BTW, are you a real scientist?...  cuz you sure spend a LOT more time on this forum during the work day than I'd like to see any employee of mine doing)

Lens cap shots?  Pushed 4 stops??  I've never heard of a worse way to evaluate a camera.  That's not even photography.  It shows something, but has no practical relevance.

It has practical relevance for certain kinds of photography. The most obvious being landscape photography. Another, maybe not so obvious, would be interior design photography (where you often encounter bright windows resulting in scenes that easily top 13 stops of DR.) It is not practical for EVERY kind of photography, and once you get past ISO 400, on Canon cameras, black point is usually above the banding noise floor and physics is the primary driver of noise, so it doesn't matter.
My Photography
Current Gear: Canon 5D III | Canon 7D | Canon EF 600mm f/4 L IS II | EF 100-400mm f/4.5-5.6 L IS | EF 16-35mm f/2.8 L | EF 100mm f/2.8 Macro | 50mm f/1.4
New Gear List: SBIG STT-8300M | Canon EF 300mm f/2.8 L II

jrista

  • Canon EF 400mm f/2.8L IS II
  • *******
  • Posts: 4732
  • POTATO
    • View Profile
    • Nature Photography
Re: 70D and Dxomark....
« Reply #431 on: September 03, 2013, 09:02:46 PM »
Here is a copy that was run through both Topaz DeNoise 5, with vertical debanding, 100% chroma noise removal, and minor luma noise removal, as well as maximum dynamic range recovery. It was then run through Nik Dfine 2 in order to remove more luma noise. DeNoise 5 does an excellent job removing chroma and banding noise, while Dfine 2 does a better job at reducing luma noise without smudging detail.

Can you denoise now the crop from the Toronto's Yonge-Dundas Square shot I posted earlier without losing detail? If I like it, I may send you the whole image, and pay you for that.  ;)

Hmm, I think I missed that one...I came into this thread in the middle. I can read from the beginning if its in this thread, and I'd be happy to give it a whirl. :)

Please don't! It was a joke (the image is real though).

LOL

Well, if you ever pick up DeNoise 5, it should solve the problem nicely. It is very configurable. The debanding can work in horiz, vert, or both, and each are independently adjustable. You have fine control over how denoising is applied to different color channels, the ability to recover black level (DR), etc. If you've been having trouble cleaning up shadow noise, it's pretty much the tool that will do the job. I use either Nik Dfine 3 or some other NR tool for luma noise reduction (photon shot noise), as Topaz DeNoise 5 past a certain "strength" level starts to heavily blur detail.
My Photography
Current Gear: Canon 5D III | Canon 7D | Canon EF 600mm f/4 L IS II | EF 100-400mm f/4.5-5.6 L IS | EF 16-35mm f/2.8 L | EF 100mm f/2.8 Macro | 50mm f/1.4
New Gear List: SBIG STT-8300M | Canon EF 300mm f/2.8 L II

FunkyCamera

  • Power Shot G16
  • **
  • Posts: 18
    • View Profile
Re: 70D and Dxomark....
« Reply #432 on: September 07, 2013, 03:13:08 PM »
I can't wait for Nikon to go out of business. Their sensors are worse in every way that matters, their bodies are low quality and the controls make no sense and their pathetic excuses for lenses  perform worse than coke bottles.

I have to try hard not to laugh out loud every time I see a moron who spent thousands on a "pro" Nikon body. Pathetic. The 70D wipes the floor with everything Nikon can put out at any price level.

canon rumors FORUM

Re: 70D and Dxomark....
« Reply #432 on: September 07, 2013, 03:13:08 PM »

jrista

  • Canon EF 400mm f/2.8L IS II
  • *******
  • Posts: 4732
  • POTATO
    • View Profile
    • Nature Photography
Re: 70D and Dxomark....
« Reply #433 on: September 07, 2013, 04:03:03 PM »
I can't wait for Nikon to go out of business. Their sensors are worse in every way that matters, their bodies are low quality and the controls make no sense and their pathetic excuses for lenses  perform worse than coke bottles.

I have to try hard not to laugh out loud every time I see a moron who spent thousands on a "pro" Nikon body. Pathetic. The 70D wipes the floor with everything Nikon can put out at any price level.

Well, I think we can boil this down to a single word:

T R O L L



 :P
My Photography
Current Gear: Canon 5D III | Canon 7D | Canon EF 600mm f/4 L IS II | EF 100-400mm f/4.5-5.6 L IS | EF 16-35mm f/2.8 L | EF 100mm f/2.8 Macro | 50mm f/1.4
New Gear List: SBIG STT-8300M | Canon EF 300mm f/2.8 L II

Aglet

  • 1D X
  • *******
  • Posts: 1016
    • View Profile
Re: 70D and Dxomark....
« Reply #434 on: September 07, 2013, 08:11:06 PM »
I can't wait for Nikon to go out of business. Their sensors are worse in every way that matters, their bodies are low quality and the controls make no sense and their pathetic excuses for lenses  perform worse than coke bottles.

I have to try hard not to laugh out loud every time I see a moron who spent thousands on a "pro" Nikon body. Pathetic. The 70D wipes the floor with everything Nikon can put out at any price level.

Well, I think we can boil this down to a single word:

T R O L L



 :P
That troll is a lovely single-shot HDR exposure capturing him in action and clearly showing us his dingy candle-lit domain while retaining excellent fidelity in the outdoor scene visible beyond his entrance.  No obvious banding or read noise problems so must be an image created with Exmor-based tools.

canon rumors FORUM

Re: 70D and Dxomark....
« Reply #434 on: September 07, 2013, 08:11:06 PM »