What ISO improvement is likely to be expected in an upcoming 5D3?

[JR]

This is the best I can do with ISO 12800. Taken during the day inside the house. No other light. 110mm f2.8 1/250s...when I zoom in on the raw file I can obviously see noise. It has noise reduction of 50 from Lightroom...

 

[JR]

Sorry, this was taken with my 5D mk II of course! With night shots of evening shots inside the quality become mediocre though as I need to apply too much noise correction to make it work in those condition, but to get a high shutter speed during the day, it seem to work...

 

[neuroanatomist]

So, that means that comparing the 1D X to 5DII (FF to FF) and comparing RAW image (which we haven't yet seen from the 1D X), I'd guess we can expect somewhere between 1/3 and 1/2 stop improvement, partly from improved on-sensor NR (pre-RAW) and partly from the gapless microlenses. I'll certainly take 1/3 to 1/2 stop...but that doesn't sound nearly as good as 2 stops, and Canon's marketing department knows that, which is why they use the 1D IV jpg as a baseline.

Are you ignoring the real life evidence that the 1D4 is giving at least 1 stop over the 5DII or are you saying that the 1DX will give 1/2 stop less than the 1D4?

What evidence is there that the 1D IV outperforms the 5DII by at least 1 stop for ISO noise on RAW images? You referenced DxOMark in regards to AF (still waiting for a link to that) - I can't find anywhere DxO reviews or rates AF performance, but they do compare sensor performance. For the 1D IV vs. 5DII, the low-light ISO ratings are 1320 for the 1D IV and 1815 for the 5DII - i.e., the 5DII is about 1/2-stop better than the 1D IV (note - that's consistent with slightly better on-sensor NR coupled partly offsetting the FF advantage of greater total light gathered).

Not 'real-world' enough for you? How about the TDP review - "Differences in noise between the 1D III, 1D IV and 5D II samples up through ISO 6400 are minor and insignificant to my eyes..." In the noise crops in the TDP review, I can see the lack of difference between the 5DII and the 1D IV. Where is this 'at least 1-stop advantage over the 5DII' you're talking about?

OTOH, in low light the 1D IV has a definite advantage in terms of AF sensitivity, and if that enables you to get the shot in those situations where you need ISO 6400, that certainly makes it the go-to camera as pwp states.

 

[torger]

So, that means that comparing the 1D X to 5DII (FF to FF) and comparing RAW image (which we haven't yet seen from the 1D X), I'd guess we can expect somewhere between 1/3 and 1/2 stop improvement, partly from improved on-sensor NR (pre-RAW) and partly from the gapless microlenses. I'll certainly take 1/3 to 1/2 stop...but that doesn't sound nearly as good as 2 stops, and Canon's marketing department knows that, which is why they use the 1D IV jpg as a baseline.

We'll see. My guess is that difference will show at high ISO, and then I expect it to be at least as good as Nikon D3s, or else Canon have certainly failed. The D3s is according to dxomark about 2/3 stops better than 5D2 in a print (i e taking into account noise reduction by averaging over the higher res on 5D2), and considerably better than that at extremely high ISOs. Dxomark does not take into account pattern noise though which can change the result a bit, the 5D2 would perform a bit less good then.

 

[briansquibb]

What evidence is there that the 1D IV outperforms the 5DII by at least 1 stop for ISO noise on RAW images?

I have posted pictures from the 1D4 at iso12800 without pp or NR. If you feel that the 5DII can match these then please show me

As a long term user of the 5DII I have never managed the equivalent over iso 6400. That to me says that the 1D4 delivers 1 stop faster than the 5DII. How else can you interpret that?

My photos are the real world, the evidence is in front of me - I have posted these real world pictures and yet you refer to a lab test as real world experience. I no longer consider iso 6400 as high iso with the 1D4. I have 7D, 5DII and 1D4. The 7D doen't cut it for usable high iso, the 5DII has been eclipsed and the 1D4 has taken the crown. I use all 3 models on a regular basis and I see the results from all three models. I know what their strengths are and also their weaknesses - this is how I choose which model to use for which photo.

This forum is not some high school debating society - it is a forum for passing on experience and their hopes and wishes for new features on upcoming equipment. My real world experience is on the table for all to see. I am a user and therefore what I do and where I have problems will apply to all forum members - and I expect to learn from them too.

IMHO the 5DII will have to at least match the 1D4 iso capability to stay in the game probably the AF too from other comments made on this and other simmilar threads. Fps? Well maybe up to 5 would be nice.

 

[NotABunny]

For some reasons which defy history (, most likely the need to feel that people pay through their teeth for top technology), people believe that the 1D line must have the absolutely best sensor possible. But as far as history shows, the inferior lines had similar noise levels per unit area of sensor. Why? Because it's probably not justifiable to design several sensors (for DSLRs), due to technological and cost reasons.

So, the 5D3 will have at least the same noise levels as the 1Dx (except if the technology used in the sensor is too expensive for a camera price below $5000). There is also another factor to consider: is the new Canon technology able to scale to resolutions in the 30s MP? Only they can answer that, but I would guess that they would not leave the mortal event photographers go without a camera capable of similar noise performance.

To me it looks like Canon is making technological leaps in this period, changing platform (DIGIC 5), sensor technology, lenses and flash technology. The only catch is that a lot of photographers can't see it because they would not benefit from it, because they work with a controlled environment.

 

[briansquibb]

To me it looks like Canon is making technological leaps in this period, changing platform (DIGIC 5), sensor technology, lenses and flash technology. The only catch is that a lot of photographers can't see it because they would not benefit from it, because they work with a controlled environment.

Are you talking about studio shooters here? rather than landscapers or street togs?

I would guess those are the 3 main categories of use for the 5DII?

 

[handsomerob]

So, the 5D3 will have at least the same noise levels as the 1Dx (except if the technology used in the sensor is too expensive for a camera price below $5000).

I seriously doubt that. 1Dx is created for sport photogs who need the ultimate low-light performance to stop action under any light conditions. 5DIII however will be aimed to a much different market. This doesn't mean it won't have great low-light capabilities, but it will simply not be in the same league as the 1Dx.

 

[briansquibb]

I think the 5DIII has got to be at least good at 12800 to match the current 1D4 - more from a marketing point of view than from a real world need. Studio shooters will have good lights, landscapers have tripods and longer shutter speeds.

I wouldn't have thought extreme low light performance would be high up the design list as the higher the ISO the lower the IQ. As the 5dxx range is about high IQ then high ISO is probably not working in the right direction.

What I would like to see is first class IQ at ISO 1600 with good IQ/low noise at iso12800

 

[neuroanatomist]

Posted this in another thread too - look at the exif of this picture that I took this evening in a dull hall and come to your own conclusions

http://www.squibb.org.uk/pictures/b09g8538x.JPG

I am getting usable photos at iso12800 - and there was no noise reduction applied.

Ahhh, I see. There aren't exactly lots of shadows or smooth detail in that linked image which make it a great one to compare noise.

And no noise reduction. Because shrinking a 16 MP image to a 0.5 MP image doesn't reduce noise. At all.

 

[pedro]

I fully expect the 5D3 to make 12800 a working ISO, if not 25600...

Thanks for posting your opinon. Quite revealing to me. And yes picsfor, hope it'll have a working ISO of 25600 and as some other poster mentioned: Hope Canon will follow their track laid by the 1Dx and keep the MP's at 24-28 or better 24 including sensor improvements trickled down from the 1Dx technique. If the 1Dx is a league of its own, why not add an 1Dsish body to it...? So a 5D3 could remain within a decent MP count. Even though the asian notion is not to fall behind competition (related to a rumored D800 36 MP monster), hope they keep to the formula presented with the 1Dx...the slow down the MP race a bit. But that's my two cents.

 

[NotABunny]

To me it looks like Canon is making technological leaps in this period, changing platform (DIGIC 5), sensor technology, lenses and flash technology. The only catch is that a lot of photographers can't see it because they would not benefit from it, because they work with a controlled environment.

Are you talking about studio shooters here? rather than landscapers or street togs?

I would guess those are the 3 main categories of use for the 5DII?

Not only, but anyone who can control the conditions of their shooting. A landscape shooter can take the time (including doing it in different days) to shoot a scene, plus the landscape doesn't move. An event is non-repeatable (unless it's staged, in which case it goes in the other category: controlled).

 

[wellfedCanuck]

Not only, but anyone who can control the conditions of their shooting. A landscape shooter can take the time (including doing it in different days) to shoot a scene, plus the landscape doesn't move. An event is non-repeatable (unless it's staged, in which case it goes in the other category: controlled).

Just a minor quibble, but what about landscape photos taken from an airplane? Given the expense and the aviation factors- it's not always possible to have a lot of control over the scene. AF, IQ and resolution are all important with aerial photography.

 

[briansquibb]

Posted this in another thread too - look at the exif of this picture that I took this evening in a dull hall and come to your own conclusions

http://www.squibb.org.uk/pictures/b09g8538x.JPG

I am getting usable photos at iso12800 - and there was no noise reduction applied.

Ahhh, I see. There aren't exactly lots of shadows or smooth detail in that linked image which make it a great one to compare noise.

And no noise reduction. Because shrinking a 16 MP image to a 0.5 MP image doesn't reduce noise. At all.

and the point of your email was.........

you saw a glossy black dog in the other thread, with shadows. Faced with real life pictures your only response is to try and score points?? The best way to do that is to show evidence to the contrary - which of course you haven't got.

 

[briansquibb]

I would like to see a real ISO 50 rather than an underexposed ISO 100.

Good point! The current iso 50 gives better IQ (according to some lab tests) but a native ISO50 could be stunning

 

[Meh]

I would like to see a real ISO 50 rather than an underexposed ISO 100.

Isn't it the opposite... ISO50 is ISO100 overexposed by 1 stop then pulled back 1 stop via software?

 

[Edwin Herdman]

I suppose that would work, though it definitely doesn't make sense to do that. An alternative would be to just expose at ISO 100, as normal...at this extreme low end of the ISO scale there will only be a problem if shooting a super-fast lens, but if the highlights are blown out you can't save them by pulling back an ISO 100 file. It would be better to just pull back brightness (which is all that ISO is controlling, after all, via an analogue for sensitivity, i.e. gain) in your favorite software package, or use an ND filter. It's definitely a (rare) problem, but the basic fact is that to get the higher ISO settings the baseline for sensor sensitivity seems to need to be set higher.

Since we've come this far, here's some amusing (but informative) reading, not for the easily offended though:

http://photography-on-the.net/forum/showthread.php?t=1081982

 

[KeithR]

and about 1 stop compared to a 7D

Errr... it ain't, you know. Not even close.

A 7D will happily give a 5D Mk II a run for its money at high ISO.

 

[KeithR]

For the 7D though, I found it excellent up to ISO 640, good at 800, 1000 was pushing it a bit and by 1600 I found it to not be so good, sometimes fine, bit more often than not way too noisy for me.

1600 ISO

3200 ISO

6400 ISO

8500 ISO (6400 ISO underexposed then pushed on conversion):

12800 ISO

The forum software has reduced the size of these - click on 'em to see all the "noise". Oh, and before anyone bothers, yes they're small anyway - you can take my word (or not) that they look superb printed big; and yes, not much shadow in them, but they're still in crappy light for the most part (Exif's in all of them).

Maybe I just had a dud

Well yes, that's one possible explanation...

 

[torger]

Here's a comparison to look at:

http://www.dpreview.com/reviews/studiocomparefullscreen.asp#baseDir=%2Freviews_data&cameraDataSubdir=boxshot&indexFileName=boxshotindex.xml&presetsFileName=boxshotpresets.xml&showDescriptions=false&headerTitle=Studio%20scene&headerSubTitle=Standard%20studio%20scene%20comparison&masterCamera=canon_s100&masterSample=img_0074.acr&slotsCount=4&slot0Camera=canon_s100&slot0Sample=img_0074.acr&slot0DisableCameraSelection=true&slot0DisableSampleSelection=true&slot0LinkWithMaster=true&slot1Camera=canon_eos7d&slot1Sample=canon7d_iso6400.acr&slot2Camera=nikon_d3s&slot2Sample=nikond3s_iso6400.acr&slot3Camera=canon_eos5dmkii&slot3Sample=img_0057.acr&x=-0.0959289424730138&y=0.03266440092941393

where you can see that the difference between 7D and 5Dmk2 is not very large. The 5D is a liiiittle better. Measurement-wise it is about one stop at 18%, and one stop more to D3s. The compact camera s100 is about two stops worse than the 7D, so there we see very real differences. Note that the D3s in this comparison looks a bit better than it is compared to 5Dmk2 due to more saturated colors in the raw conversion.

A large part of high ISO noise is photon shot noise which has no pattern, and thus is quite acceptable. Noise is usually less visible and disturbing in a print than one might think when looking at a 100% crop. I rarely use noise reduction when processing raws since it kills color and detail, and the noise is not so disturbing in the print.

Also note that if you don't mind short DoF there is a possibility on the fullframe to gather more light in the same amount of time. For example if you need 50/1.2 on the 7D for a certain view, you can use 85/1.2 on the 5Dmk2 to get the same view and then you have the option with shorter DoF (more light), in this case about a stop or so, which means that you may be able to use ISO1600 on your 5D with 85mm while you need ISO3200 on the 7D with 50mm. If you need the same DoF there will be no difference though, since you then need to use a smaller aperture on the 5D.