5D3 same max dynamic range as the 5D2???

[Abraxx]

Oh, what I feared, no real DR improvement.
If further tests confirm this, I'm not going to buy the 5D Mark III, especially not at the extreme price point Canon has chosen here in Germany.
DR improvement was crucial for me. I'm really disappointed if this turns out to be true. :'(

I had a chance to touch the new 5D & the D800.
I have to slightly adjust my statement above, I still have an issue with the pricepoint in Germany,
BUT
There were some things which I liked a lot... For me haptics and grip are important too and
the 5D Mark iii just melted into my hand. ;D
The touch dial is really great and the control with the, let me call it "JOYstick", was very intuitive.
I just don't like were Nikon placed the ISO button.
I like to fiddle around with settings : and Canon is just faster for my taste.
There are other tiny, little, maybe for you unimportant, things like, well call me stupid, but for example I like the orange backlit at dim light compared to the greenish one from the Nikon Topdisplay.
My hand and feel is not made for the Nikon design, thats what I noticed again!

Regarding DR,
I can and will compensate some DR in post processing.
I like the HDR blending, autoalign(!) of the 3 photos, for me, who frequently quickly does handheld shoots, its a new step forward.
I like the 2 shoots comparison on the large screen.

When I touched the 5D, I was reminded of one of the reasons I went with Canon, its the haptics, grip & control.
For me making photos is not a job, its a joy.
Still my budget is tight, so I'll have to wait a bit. But I think everybody who is currently concerned about what to do,
Try To get hold of the cam and really focus on the things you really need and prefer. Then forget about all the forums.
In the end I noticed its me taking the photo, not the cam,
if I have fun then I'm happy.

Just my 2 cents I wanted to share

 

[LetTheRightLensIn]

It is to early to say for sure, but it seems like there may be a way to filter out most of the banding from the 5D3 unlike with say a 20D,30D,50D,5D2, etc. It seems like they may have eliminated most of the banding that varies shot to shot and mostly left it with a type of banding that may be relatively consistent shot to shot. It is definitely way to early to be sure though. But there are hints of promise.

You'd need a RAW converter that had a new option to feed certain information or a pre-processor to alter RAWs before feeding them to regular programs (as to the former, who knows if any companies would offer the options, as to the latter it can certain be coded no question at all). But again, still to early to say.

 

[etto72]

I would like everyone to make a simple consideration.....
Looking at the last 4 Canon sensor generation
-1Ds mk II, 5D,
-1Ds mk III,1D mk III,5D mk II, 40D
-1D mk IV, 7D
-1DX, 5D mk III
Lets look at the pixel pitch and at the equivalent DR values for each generation:
5D Pixel size 8.2 = 9.2 DR
1DS mk III Pixel size 6.4 = 11.3
5D mk II Pixel size 6.4 = 11.1
40D Pixel size 5.7 = 11
7D Pixel size 4.3 = 9.8
1D mk IV Pixel size 5.7 = 12
1DX Pixel size 6.9 = ?
5D mk III Pixel size 6.25 = ?
As you can see each generation gave an increase in DR except with the 7D wich has an extremely smaal 4.3 pixel pitch size
Last generation sensor gave a 12 EV with 5.7 pixel size
The mk III has 6.25 pixel size
Even if Canon didn't progress in DR with the newest sensor engineering...
..... I have really hard time to believe that with the 5d mk III with 6.25 they can't get same or better DR results with a sensor that has 5.7 !!!!
Time will tell....
Thanks
Ettore

 

[wickidwombat]

So here is a question

Does anyone actually know what the useable dynamic range of any of these cameras is?
I'm not talking about the infinite naval gazing number crunching analytical sort i'm taking about where the line gets drawn between "yeah that looks good" and "OMG that pattern noise just tore my left eyeball out"

Its prettysubjective isnt it? so the line is going to be different for everyone

and then there is all the third party Noise reduction gear out there which can salvage something that was fairly poor and make it acceptable extending that range of use. I dont think comparing things like this soley based on some numbers is very wise.

 

[well_dunno]

Last generation sensor gave a 12 EV with 5.7 pixel size
The mk III has 6.25 pixel size
Even if Canon didn't progress in DR with the newest sensor engineering...
..... I have really hard time to believe that with the 5d mk III with 6.25 they can't get same or better DR results with a sensor that has 5.7 !!!!

+1 to that, mark 3 has bigger photosites thanks to the gapless microlens sensor tech which should benefit both DR and ISO performance of the cam. If there is no improvement on those, I do not think it would make 5D mark 3 a bad camera still, considering all the other improvements, but it would make it difficult to justify the price tag IMHO...

 

[Radiating]

I would like everyone to make a simple consideration.....
Looking at the last 4 Canon sensor generation
-1Ds mk II, 5D,
-1Ds mk III,1D mk III,5D mk II, 40D
-1D mk IV, 7D
-1DX, 5D mk III
Lets look at the pixel pitch and at the equivalent DR values for each generation:
5D Pixel size 8.2 = 9.2 DR
1DS mk III Pixel size 6.4 = 11.3
5D mk II Pixel size 6.4 = 11.1
40D Pixel size 5.7 = 11
7D Pixel size 4.3 = 9.8
1D mk IV Pixel size 5.7 = 12
1DX Pixel size 6.9 = ?
5D mk III Pixel size 6.25 = ?
As you can see each generation gave an increase in DR except with the 7D wich has an extremely smaal 4.3 pixel pitch size
Last generation sensor gave a 12 EV with 5.7 pixel size
The mk III has 6.25 pixel size
Even if Canon didn't progress in DR with the newest sensor engineering...
..... I have really hard time to believe that with the 5d mk III with 6.25 they can't get same or better DR results with a sensor that has 5.7 !!!!
Time will tell....
Thanks
Ettore

I think it's worth mentioning that using raw data to discuss low iso banding and dynamic range is VERY VERY misleading.

The usual testing methodology simply determines the lower end of the dynamic range as the point at which the noise overwhelms the rest of the data, ie a signal to noise ratio.

The issue with low iso banding is not so much that Canon's cameras have excessive noise at low iso, they do not have especially excessive noise, and the data proves this at only slightly more than a 1 stop difference between Canon and say Nikon. The issue however is that this low iso banding noise CANNOT be removed because noise reduction algorithms treat pattern noise as detail and do not remove it.

Noise reduction algorithms are currently good for about 2-5 stops of noise reduction depending on how tolerant you are of losing detail.

So while on paper the 5D II/III have aproximately a fantastic 11+ stops dynamic range. In the real world they have an earth shattering 3-6 stops less DR than tested because NR is ineffective on low iso banding.

So here is a question

Does anyone actually know what the useable dynamic range of any of these cameras is?
I'm not talking about the infinite naval gazing number crunching analytical sort i'm taking about where the line gets drawn between "yeah that looks good" and "OMG that pattern noise just tore my left eyeball out"

Its prettysubjective isnt it? so the line is going to be different for everyone

and then there is all the third party Noise reduction gear out there which can salvage something that was fairly poor and make it acceptable extending that range of use. I dont think comparing things like this soley based on some numbers is very wise.

Hope that also answers your question.

 

[LetTheRightLensIn]

I would like everyone to make a simple consideration.....
Looking at the last 4 Canon sensor generation
-1Ds mk II, 5D,
-1Ds mk III,1D mk III,5D mk II, 40D
-1D mk IV, 7D
-1DX, 5D mk III
Lets look at the pixel pitch and at the equivalent DR values for each generation:
5D Pixel size 8.2 = 9.2 DR
1DS mk III Pixel size 6.4 = 11.3
5D mk II Pixel size 6.4 = 11.1
40D Pixel size 5.7 = 11
7D Pixel size 4.3 = 9.8
1D mk IV Pixel size 5.7 = 12
1DX Pixel size 6.9 = ?
5D mk III Pixel size 6.25 = ?
As you can see each generation gave an increase in DR except with the 7D wich has an extremely smaal 4.3 pixel pitch size
Last generation sensor gave a 12 EV with 5.7 pixel size
The mk III has 6.25 pixel size
Even if Canon didn't progress in DR with the newest sensor engineering...
..... I have really hard time to believe that with the 5d mk III with 6.25 they can't get same or better DR results with a sensor that has 5.7 !!!!
Time will tell....
Thanks
Ettore

none of those results are normalized

 

[LetTheRightLensIn]

So here is a question

Does anyone actually know what the useable dynamic range of any of these cameras is?
I'm not talking about the infinite naval gazing number crunching analytical sort i'm taking about where the line gets drawn between "yeah that looks good" and "OMG that pattern noise just tore my left eyeball out"

Its prettysubjective isnt it? so the line is going to be different for everyone

and then there is all the third party Noise reduction gear out there which can salvage something that was fairly poor and make it acceptable extending that range of use. I dont think comparing things like this soley based on some numbers is very wise.

the relative, which does better than what, is easily defined (ignoring banding) but sure the amount you find acceptable is subjective, taht's something else (but if one has more measured than another it will also have more by whatever your subjective opinion is than the other one, so that is why it is wise, because it makes it easy to compare in standard way)

 

[Tammy]

http://www.arekgmurczyk.pl/tag/roadshow-nikon-d4-i-d800/

are you able to calculate the D800 DR from this set?

 

[jrista]

I would like everyone to make a simple consideration.....
Looking at the last 4 Canon sensor generation
-1Ds mk II, 5D,
-1Ds mk III,1D mk III,5D mk II, 40D
-1D mk IV, 7D
-1DX, 5D mk III
Lets look at the pixel pitch and at the equivalent DR values for each generation:
5D Pixel size 8.2 = 9.2 DR
1DS mk III Pixel size 6.4 = 11.3
5D mk II Pixel size 6.4 = 11.1
40D Pixel size 5.7 = 11
7D Pixel size 4.3 = 9.8
1D mk IV Pixel size 5.7 = 12
1DX Pixel size 6.9 = ?
5D mk III Pixel size 6.25 = ?
As you can see each generation gave an increase in DR except with the 7D wich has an extremely smaal 4.3 pixel pitch size
Last generation sensor gave a 12 EV with 5.7 pixel size
The mk III has 6.25 pixel size
Even if Canon didn't progress in DR with the newest sensor engineering...
..... I have really hard time to believe that with the 5d mk III with 6.25 they can't get same or better DR results with a sensor that has 5.7 !!!!
Time will tell....
Thanks
Ettore

Where in the world did you get those numbers? They are inconsistent with both primary sources of data for whats technologically possible (DXO) and most likely possible in the real world (DPR). Real-world data is more subjective, so if we just take DXO data, which is more objective, the numbers become:

Gen.	Model	Pixel Pitch	Screen DR	Norm. DR	Q.E.
1	5D C	8.04	10.77	11.13	25%
1	1Ds II	7.05	10.79	11.34	30%
2	40D	5.63	10.99	11.18	33%
2	5D II	6.39	11.16	11.86	33%
2	1D III	7.20	11.50	11.70	29%
2	1Ds III	6.30	11.25	11.97	31%
3	7D	4.16	11.12	11.73	41%
3	1D IV	5.70	11.46	11.95	44%
4	5D III	6.25	??	??	(47%)?
4	1D X	6.95	??	??	(47%)?

 

[jrista]

looks like the DR is the smae regardles sof the body. There is not advantage in FF over 7D in IQ and dynam,ic ranger!

On a normalized basis, not much. On a native basis, yes, DR has improved by over 2/3rds of a stop over 3 generations. Quantum efficiency has also consistently improved generation over generation (I think I've been conservative with my estimates for Canon's latest.) If we extend that to a fourth generation, one would expect a minimum of about 1/3rd of a stop improvement in DR for the next generation unless Canon has some physical limitation in their electronics that simply prevent achieving 12 stops total.

 

[LetTheRightLensIn]

looks like the DR is the smae regardles sof the body. There is not advantage in FF over 7D in IQ and dynam,ic ranger!

Not really for dynamic range. For FF though you can get better SNR what with the 2.5+ times larger area over which to collect light.

 

[Aglet]

[size=14pt]Pixel-Peepin' Pushing Plaid Shadows update
[/size]

Preliminary subjective evaluation conclusions are farther below.

I've compiled dark frame raw files for the following cameras:

Canon DSLRs: 40D, 50D, 60D, 7D, 5D2, 5D3, 450D, 1000D
Canon PowerShots: G3, G5, G6, G11, G12
Nikon: D90, D5100 (seeking D800, D300/s)

As much as possible, these were all done using Manual mode and the following settings:

- T = 1/200s
- A = f/16
- iso base + all std values, 100, 200, 400, etc
- WB = daylight
- all in-camera noise reduction disabled
- all in-camera tone-curves disabled (Auto Lighting Optimiser, D-Lighting, etc.)
- capped lenses and as dark an environment as possible (G-series and 5D3 were under less than ideal conditions)


Also, in-camera dark frame subtraction was tested using iso 100 and 800 wherever possible:

- On SLRs this was a 2 second exposure at both 100 and 800 iso with, and without, long exposure noise reduction.

- On the G-series it was done at base ISO and +4 EV iso at 1 second and 2 second exposures. These cameras automatically turn on dark frame subtraction at 1.3 seconds or longer, if I remember correctly.


Post processing was done with ACR, using DNG converter 6.7rc1 for the 5D3 files.

Supported raw file vs converted to DNG files compared look virtually identical on the ones I tested so let it roll as equivalent.

Import processing was preset to:

- 0 black level clipping (so as not to lose any of those noisy pixels)
- +4 EV exposure compensation
- 2010 Adobe default standard style instead of any of the camera-specific tone/color options
- imported as 16b/color
- WB as shot (DL) (some anomalous K amounts showed up despite the daylight settings, looking into that later (4800-5600K)


- further +3 EV added under PS CS5 Image/Adjust/Exposure to really bring up the dark noise to very visible levels


Comparing the overall amount of noise, and its subjective quality, was very illuminating!

WOW!

I can now really see the weaknesses and strengths of various camera bodies when it comes to different kinds of shooting conditions and post-processing expectations. Since I often shoot scenes with very high DR I want to be able to take a single shot, exposed to retain hilite info and PP to bring up shadow details. I don't want to do this with multiple-shot HDR methods since it's often not suitable or convenient for non-static scenes.


As I'd expected, the noise quality of my old 40D is more amenable to large shadow pushing amounts than newer bodies like the 7D.
The 60D was somewhere in between, 50D was kind of like the 40D but much higher overall noise. 450D was surprisingly decent, 1000D was pretty limited.
PowerShot G11 showed considerable room to push shadows while D90 and D5100 Nikons showed far more room and virtually no detectable patterning to the noise at even moderate to high levels of shadow push. 7D shadows could not be pushed much at all at low iso due to tremendous amounts of fine vertical banding (due to it's doubled readout channels?) but is better at 800 iso and up as random noise starts to dominate. This fine banding can be visible in large prints.

When comparing 5d2 to 5d3, the 5d3 shows only about 0.3 EV less overall low-iso dark noise as perceived by overall brightness and histogram of resulting pushed image. 5d3's noise character, however, is somewhat less patterned (retains vertical, almost no horizontal banding) and may allow it to be pushed slightly more in post, perhaps an optimistic 1 stop improvement. 5d3's hi iso performance is pretty good, tho that's not something I normally need beyond iso 1600.

OTOH, D5100 can be pushed an extra 4.5 EV before its low iso files match the same level of total noise as the 5d2 but the D5100's noise is less patterned so could allow for even further pushing before hitting the limits.


This is all a rather subjective comparison, and does not lend itself very well to deriving actual DR numbers for any given camera but this is fairly representative of the kind of post-processing workflow I use for some kinds of shots that I do. Being able to now see these cameras compared like this gives me a good idea which one will be "the right tool for the job."
If I could reshoot all these bodies on a nice static scene example, like imaging-resource does,then an even better idea of the practical shadow pushing limits, using the hi-key areas as a normalizing exposure level, could be obtained. That would be closer to approximating real-world use and limits.

When I get more time to compile and organize actual image samples from all these tests into a layout where you can more readily see the effects I'll start a new thread on the subject using Pixel-Peepin' Pushing Plaid Shadows as the topic.


The recent release of DxOmark data on the D800 is tantalizing, but I still want my own dark file samples if I can get my hands on one.
And let's hope Canon can meet the challenge that Nikon/Sony have thrown down... SOON.

 

[Astro]

There are slight differences in banding vs. fixed pattern...I believe they are caused by different gains at different parts of the pipeline. The fixed pattern stuff is finer and cross-hatched, and pretty much always the same in every frame. Whatever causes banding is different, and the banding can change frame to frame. As far as I can tell, the fine, FIXED PATTERN noise is gone in the 5D III.

FPN is caused by physical non-uniformities at sensor level.
it can be easily removed because it does not change.

 

[sndan]

http://www.flickr.com/photos/trentchau/6861371280/#sizes/l/in/set-72157629637172995/

A GREAT DIFFERENCE

 

[Aglet]

random banding is more likely caused by power supply instabilities or other system noise creeping into the sensor signal path. comes down to proper electronics design and shielding.
I found noticeable random horizontal banding in some older bodies I tested, like the 40D, when pushing shadows.
Makes me wonder how well it'd perform if the internal power supplies and shielding were improved...