5d Mark III low iso poor performance.

Started Jun 20, 2012 | Discussions
Ogjetaknight Regular Member • Posts: 359
Re: You need to read all the posts in your thread

Alexandros Trichos wrote:

Hmmm looks interesting....

http://colorbyjorg.wordpress.com/2011/01/28/highlight-tone-priority-htp-choice-of-iso-and-noise/

That is something. And why is that with the broken isos? I ll try it...

This has already been suggested here in your own thread. You might want to read some of the suggestions people have made it will help answer your questions.

MayaTlab0
MayaTlab0 Senior Member • Posts: 2,736
Re: You dont need Sony-Exmor tech to have clean shadows

Alexandros Trichos wrote:

First of all I will assume this is a normal behavior of the camera, so this is not a bad copy that I have, thats how all 5dIIIs are.

Second, I can't se how some people are so blind. The camera is a tool, it should be working as expected, even if the man behind that is an amateur or a professional. This is a technical thing. There is noise at low isos where it shouldn't exist. I cannot understand why people imply that "A good photographer should not care for such things" or "we pay a lot of moony but we don't care about such things, because we are good photographers".

People aren't blind, they just thought you were looking for advices to minimise this and they gave you plenty. But apparently you're not here to improve what you can do with the 5d mark III but just complain about its sensor DR at low ISO. I'm with you on that one, but then either make it a specific thread with clearly stated intentions or go to another forum.

Now, noise at low ISO exists on every single camera. Even the D800 (shoot a deep blue sky with it, open it in Lightroom, and you'll see). What you dislike is the fact that the 5d mark III has slightly more read noise (which is a specific type of noise) than the average for non Sony / Nikon sensors and that it produces patterns, which is indeed a shame, but can minimised given the right technique.

And I don't believe that the problem is the underexposed images. It has nothing to do with it. This phenomenon, as I mentioned before, also appears in other properly exposed images.

It's sometimes true. But properly exposing your pictures and switching HTP off will always help minimise it.

Anyway... I don't really know what to do...

First, let me tell you what you SHOULD have done. Before spending that much money on a camera, it's common sense to do your homework and get as much information as possible about. The simple fact that you're able to post on DPreview tells me that you know how to use internet. That's a good thing, because there are plenty of websites talking about Canon's sensors read noise at low ISO, which cameras are affected (basically all of them), and how to minimise it. Then, drawing from your own experience and needs, you should have taken the right decision and NOT buy a Canon camera.

Second, if that is a deal breaker for you, just sell it, and buy a Nikon D800, which is undoubtedly better at dealing with read noise at low ISO, and not in a subtle way. But I suppose you bought the Canon for some reasons, and you might not find the equivalent functionnalities on the Nikon. Do also understand that the 5D mark III's sensor isn't "noisy" in general, in fact it's slightly cleaner than the D800's sensor at high ISO and doesn't exhibit as much colour shift in the shadows (something just as painful to correct as banding), so a D800 won't produce better pictures at higher ISOs. the 5D mk III just has poor read noise figures for a camera of this class.

BTW, it's not because it has poor read noise figures and ugly pattern noise that its sensor is "lagging behind" as a whole. As I said, I believe it is, with the 1DX, the only full frame sensor with an electronic first curtain for example. And it may have some innovation here or there that we don't directly see.

I'd love Canon to improve their sensors read noise figures, but unfortunately that's out of our hands. So the only things you can do are :
1 - do your homework before buying a camera.
2 - be rational in your purchase decisions.

Snaphappykiwi
Snaphappykiwi Forum Member • Posts: 52
Re: 5d Mark III low iso poor performance.

Alexandros Trichos wrote:

Am I the only one to think that 5d MK3 has very poor performance in shadows, even in low isos like 100 or 200 ? High color noise, weird noise patterns, vertical bands/stripes all without any pushing!!!
Straight out of the box the images look terrible in the shadow areas.
Why is that? I am very disappointed. VERY disappointed ...
Even my poor old 350d did better in that domain...

What an unusual name. Are you Greek?

 Snaphappykiwi's gear list:Snaphappykiwi's gear list
Canon PowerShot G1 X Canon EOS 5D Mark III Canon EF 50mm f/1.4 USM Canon EF 85mm f/1.2L II USM Canon EF 16-35mm F2.8L II USM +3 more
gigamel Regular Member • Posts: 178
Re: You dont need Sony-Exmor tech to have clean shadows

schmegg wrote:

gigamel wrote:

Denial:

"An unconscious defence mechanism characterized by refusal to acknowledge painful realities, thoughts, or feelings."

It is all about NOISE! - that you chooses to be blind is your problem!

It is only fair that the OP raises the question of the EXTREMELY NOISY 2012 Canon sensors that Canon has the nerve to charge $3500 for.

Yes - in fact, according to DXO, slightly less noise at all ISO's than the D700 and all but exactly the same noise as the D800.

So by your logic, Nikon has failed to improve their own design since the D700! LOL! You need to take some reality pills!

And you need to take some brain pills - look at the evidence:

http://dslr-check.info/2012/0511_5D3_ISO200+4EV.jpg

http://dslr-check.info/2012/0511_D700_ISO200+4EV.jpg

And this idiotic comment tells us ALL that you dont know what you are talking about:

"according to DXO, slightly less noise at all ISO's than the D700 and all but exactly the same noise as the D800."

Because the DxO "noise" you are talking about is SNR18 - do you know what that means? Guess not!

It means the signal to noise ratio at 18% middle gray - that is smack in the middle of the histogram and has nothing to do with the noise in the shadows!!

Take a look at the FULL SNR at Dxo (measurements, Full SNR)

http://www.dxomark.com/index.php/Cameras/Camera-Sensor-Database/Canon/EOS-5D-Mark-III

http://www.dxomark.com/index.php/Cameras/Camera-Sensor-Database/Nikon/D700

http://www.dxomark.com/index.php/Cameras/Camera-Sensor-Database/Nikon/D800

The SNR at 1% is

5D3 24 dB
D700 27 dB
D800 27 db (1 stop better than the 5D3)

All at the pixel level (!) making the D800 pixel just as good as the way bigger D700 pixels and both a stop (3 dB) better than the 5D3 at 1% gray

The SNR at 0,5% is

5D3 19,5 dB
D700 23,5 dB
D800 24,5 dB (nearly 2 stop better than 5D3)

The SNR at 0.1% is

5D3 6 dB
D700 10 dB
D800 15,5 dB (3 stop better than 5D3)

The SNR at 0.05% is

5D3 0 dB (noise and signal are equal)
D700 6 dB
D800 12 dB (4 stops better than the 5D3)

And again the D700 is not Sony-Exmor tech, but 4 year old Nikon tech - amazing it is so mush better than state of the art Canon tech in the $3500 5D3 and the $7000 1Dx.

And on top of that the noise of the Canon sensors appears as banding making it WAY more visible!

Ouch! - come again Canon when you know how to make noise free, non-banding sensors!

I actually mean this very friendly - you obviously didn't know what you were talking about - so have a nice day Schmegg and the same to the rest of you Canon-fanboys - we all want Canon to perform better to put pressure on the competitors

Rick Knepper
Rick Knepper Forum Pro • Posts: 16,960
Maybe next time, avoid the hyperbole and provocation

and you'll probably avoid the fanboi counter-insurgency (as some of the subject lines in the replies would suggest).

I can't comment on your tech problem as there is no real information provided here.

Alexandros Trichos wrote:

Am I the only one to think that 5d MK3 has very poor performance in shadows, even in low isos like 100 or 200 ? High color noise, weird noise patterns, vertical bands/stripes all without any pushing!!!
Straight out of the box the images look terrible in the shadow areas.
Why is that? I am very disappointed. VERY disappointed ...
Even my poor old 350d did better in that domain...

-- hide signature --

Rick Knepper, photographer, non-professional, shooting for pleasure, check my profile for gear list and philosophy.

 Rick Knepper's gear list:Rick Knepper's gear list
Pentax 645Z Canon EOS 5DS R Fujifilm GFX 50S Pentax smc D FA 645 55mm F2.8 AL (IF) SDM AW Pentax DA645 28-45mm F4.5 +9 more
schmegg Veteran Member • Posts: 5,768
Re: You dont need Sony-Exmor tech to have clean shadows

gigamel wrote:

"according to DXO, slightly less noise at all ISO's than the D700 and all but exactly the same noise as the D800."

Because the DxO "noise" you are talking about is SNR18 - do you know what that means? Guess not!

It means the signal to noise ratio at 18% middle gray - that is smack in the middle of the histogram and has nothing to do with the noise in the shadows!!

It does however represent an image level measurement of noise - much more useful in practical terms when you are interested in creating images rather than measuring parameters - something that certainly seems to escape you.

Here is the DxO graphs for image level SNR ...

Remarkably - these lines, or rather, the difference between them quite accurately represents the amount of difference you'll see in noise in the images from these cameras when the image has been exposed properly. It also accurately represents the difference in images you'll see on sites that actually test these cameras - rather than just measuring sensor parameters.

Now - if you are the type that regularly underexposes your subject by, say, 4 stops, then the difference in shadow performance might be important. But, fortunatley for most of us (though sadly not yourself it would appear), we manage to expose our subject much more accurately and effectively than that!

Alexandros Trichos
OP Alexandros Trichos Junior Member • Posts: 41
Re: 5d Mark III low iso poor performance.

What an unusual name. Are you Greek?

Yes ! I am !

 Alexandros Trichos's gear list:Alexandros Trichos's gear list
Canon EOS 350D Canon EOS 5D Mark III Canon EF 50mm f/1.4 USM Canon EF 70-200mm F4L USM Canon EF 100mm F2.8L Macro IS USM +2 more
fwampler Senior Member • Posts: 1,775
Re: 5d Mark III low iso poor performance.

You can't see the forest for the trees. Enjoy your new camera.

Alexandros Trichos wrote:

Am I the only one to think that 5d MK3 has very poor performance in shadows, even in low isos like 100 or 200 ? High color noise, weird noise patterns, vertical bands/stripes all without any pushing!!!
Straight out of the box the images look terrible in the shadow areas.
Why is that? I am very disappointed. VERY disappointed ...
Even my poor old 350d did better in that domain...

-- hide signature --

Fred

 fwampler's gear list:fwampler's gear list
Canon PowerShot G1 Canon EOS 5D Mark III Canon EOS 7D Mark II Canon EOS M3 Canon EOS-1D X Mark II +18 more
Alexandros Trichos
OP Alexandros Trichos Junior Member • Posts: 41
Re: You dont need Sony-Exmor tech to have clean shadows

1 - do your homework before buying a camera.
2 - be rational in your purchase decisions.

I will do my homework teacher, please don't tell mom.

Thanks for the good and useful advices though.

 Alexandros Trichos's gear list:Alexandros Trichos's gear list
Canon EOS 350D Canon EOS 5D Mark III Canon EF 50mm f/1.4 USM Canon EF 70-200mm F4L USM Canon EF 100mm F2.8L Macro IS USM +2 more
Schwany
Schwany Forum Pro • Posts: 10,169
Re: absolutely, definitively, and positiviely, not normal

Did anyone notice the shutter speeds used for ISO200 indoors? 1/2000th, 1/500th

Both shots are under exposed by quit a bit. It's normal for the Canon DSLRs I've used. They all do this to one degree or another. The 5DMkII and later series just happen to have this funky pattern noise.

Not worth getting excited about though. It's just a silly camera. One that doesn't work for the OP's photography style.

It doesn't look that bad to me. I've seen worse.

kevindar wrote:

this is absolutely not normal, and the wrong behavior. is this raw? jpeg? if raw, what are you using to process? this is definitely not normal, unless I guess you are shooting is 140 F degree temp, or something bizzarre. you can easily download dpr raw samples which have plenty of shadow, or just inspect them as they are. this is not normal.
--
http://razzi.me/kevindar/photos

 Schwany's gear list:Schwany's gear list
Canon EOS-1Ds Mark II Canon EOS 5D Mark II Canon EOS-1D Mark IV Canon EOS-1D X Canon EOS 5D Mark IV +14 more
John Sheehy Forum Pro • Posts: 21,426
Re: Take it back

Alexandros Trichos wrote:

So here are the raws.

http://www.sendspace.com/filegroup/nA3A16dCP9CjsnDtdsf2%2FQ

You can judge for yourselves!

I agree with you that it is ridiculous that this banding exists - it would be very easy for Canon to eliminate it by just making the sensor chip a little taller and dedicating rows of masked pixxels to detect the vertical stripes and remove them.

However, your exposure is not as standard as you suggest; these are a couple stops lower exposure than "standard" for the ISO, and about 3 stops lower than what would be optimum. As you are not protecting highlights here, there is no need for the weak exposures.

If you're going to use Canon DSLRs, and want to avoid such issues, you need to know a few things:

1) Highlight Tone Priority (HTP) is disastrous to shadows in the low range of ISOs because it brings the signal 1 stop closer to a very harsh post-analog-gain noise floor (it is totally harmless at high ISOs, with their more linear noise).

2) Automatic Lighting Optimization (ALO) can hide from you the fact that your exposure is weak, by automatically brightening the review/embedded JPEG.

3) Super-fast apertures like f/0.95 to f/2 cause the camera to do something very stupid; the camera under-exposes, not to protect highlights, but just so that your f-stop and shutter speed are "true" to the ISO exposure index, to cover up the fact that the camera really can't do those apertures, and loses a good deal of the extra light you are supposed to get opening those lenses all the way because the microlenses can't handle them. Rather than just under-expose, the camera mathematically scales the RAW data, and clips away any extra highlights that may have otherwise been retained.

4) On Canon DSLRs with 1/3-stop ISOs, other than the 1D cameras and the original 5D, ISOs like 160/320 are just 200/400 over-exposed, and ISOs like 125/250 are just 100/200 under-exposed, making this latter group show off the banding and high noise floor even more.

Combine #1, #3, and #4, such as f/1.2, ISO 250 with HTP, and you have a recipe for major shadow disaster.

-- hide signature --

John

Alexandros Trichos
OP Alexandros Trichos Junior Member • Posts: 41
Re: Take it back

John Sheehy wrote:
...

Combine #1, #3, and #4, such as f/1.2, ISO 250 with HTP, and you have a recipe for major shadow disaster.

Thanks! Very good observations and advices pal!

 Alexandros Trichos's gear list:Alexandros Trichos's gear list
Canon EOS 350D Canon EOS 5D Mark III Canon EF 50mm f/1.4 USM Canon EF 70-200mm F4L USM Canon EF 100mm F2.8L Macro IS USM +2 more
psuser New Member • Posts: 1
Re: 5d Mark III low iso poor performance.

Alexandros Trichos wrote:

Am I the only one to think that 5d MK3 has very poor performance in shadows, even in low isos like 100 or 200 ? High color noise, weird noise patterns, vertical bands/stripes all without any pushing!!!
Straight out of the box the images look terrible in the shadow areas.
Why is that? I am very disappointed. VERY disappointed ...
Even my poor old 350d did better in that domain...

i guess this is normal ... i figure out on all canon dslr.....not only 5d3
it not happen on jpg...its on raw without every function turned off
still look like iso smoothing by default

David Hull
David Hull Veteran Member • Posts: 6,334
Re: You dont need Sony-Exmor tech to have clean shadows

Alexandros Trichos wrote:

I DID NOT push anything.

Some people are blind...

I suppose so, but then… some people cannot set up a camera properly. This will be evident if you take a look at your histogram. When I look at your RAW files, I can push them 2 to 3 stops before there is any clipping on the top end. This means that you put yourself in the hole noise wise by at least two stops from the get-go. I can see absolutely no reason whatsoever why you would have chosen this exposure strategy for these two shots, perhaps you can explain?

The story is there in your histogram, but then, as you say, some people are blind

 David Hull's gear list:David Hull's gear list
Canon EOS M5 Canon EOS R
gigamel Regular Member • Posts: 178
Re: You dont need Sony-Exmor tech to have clean shadows

schmegg wrote:

gigamel wrote:

"according to DXO, slightly less noise at all ISO's than the D700 and all but exactly the same noise as the D800."

Because the DxO "noise" you are talking about is SNR18 - do you know what that means? Guess not!

It means the signal to noise ratio at 18% middle gray - that is smack in the middle of the histogram and has nothing to do with the noise in the shadows!!

It does however represent an image level measurement of noise - much more useful in practical terms when you are interested in creating images rather than measuring parameters - something that certainly seems to escape you.

Here is the DxO graphs for image level SNR ...

Remarkably - these lines, or rather, the difference between them quite accurately represents the amount of difference you'll see in noise in the images from these cameras when the image has been exposed properly. It also accurately represents the difference in images you'll see on sites that actually test these cameras - rather than just measuring sensor parameters.

Now - if you are the type that regularly underexposes your subject by, say, 4 stops, then the difference in shadow performance might be important. But, fortunatley for most of us (though sadly not yourself it would appear), we manage to expose our subject much more accurately and effectively than that!

HAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHA

You only photograph grey cards????????
ALL your pixels are exposed to 18% middle grey???????

HAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHA

ALL normal images contain a variety of grey levels!!!

Take a look at your histograms!!!

DOHHHHH!!!!

ron purdy Senior Member • Posts: 2,493
Re: You dont need Sony-Exmor tech to have clean shadows

You need to post the RAW file if you really want help. Also, as I said earlier, turn the stupid Auto Light Optimizer off. And use DPP.
--

ron purdy dot com

MayaTlab0
MayaTlab0 Senior Member • Posts: 2,736
Re: Take it back

John Sheehy wrote:

4) On Canon DSLRs with 1/3-stop ISOs, other than the 1D cameras and the original 5D, ISOs like 160/320 are just 200/400 over-exposed, and ISOs like 125/250 are just 100/200 under-exposed, making this latter group show off the banding and high noise floor even more.

Does this correspond to what one can see on this graph (The D800 curve is straightforward, while the 5D MKIII curve goes up and down) ?
http://home.comcast.net/~NikonD70/Charts/PDR.htm#EOS%205D%20Mark%20III

Is it possible on Canon cameras to force auto ISO to use only the optimal stops ?

schmegg Veteran Member • Posts: 5,768
Re: You dont need Sony-Exmor tech to have clean shadows

gigamel wrote:

HAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHA

You only photograph grey cards????????
ALL your pixels are exposed to 18% middle grey???????

HAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHA

ALL normal images contain a variety of grey levels!!!

Take a look at your histograms!!!

DOHHHHH!!!!

Gee. What a mature response!

Umm ...

Yes - I expose my images correctly so that the subject is not 4 stops under. In fact, most of my histogram that encompasses my subject is usually well toward the right hand side.

Seriously - take a look at the DPR RAW images. The difference is marginal at best. Pick ISO100 and use your anal abilities to look into the very shadowiest bits. There is SFA difference. Please don't make me post the crops and make you look even more ridiculous than you already do.

You are attempting to create an issue where none exists. And it's falling flat with all of us that have actually held a 5D series camera in our hands and used it to create images (something you quite obviously haven't done).

schmegg Veteran Member • Posts: 5,768
Re: Take it back

MayaTlab0 wrote:

John Sheehy wrote:

4) On Canon DSLRs with 1/3-stop ISOs, other than the 1D cameras and the original 5D, ISOs like 160/320 are just 200/400 over-exposed, and ISOs like 125/250 are just 100/200 under-exposed, making this latter group show off the banding and high noise floor even more.

Does this correspond to what one can see on this graph (The D800 curve is straightforward, while the 5D MKIII curve goes up and down) ?

Yes - exactly

Is it possible on Canon cameras to force auto ISO to use only the optimal stops ?

No. Auto ISO can uses whole stops or third stops. The choice is selectable. The difference in the final image is marginal at best and should tell you something about the magnitude of relevance that these lines have to your final images.

(unknown member) Senior Member • Posts: 1,378
Re: You dont need Sony-Exmor tech to have clean shadows

You seem lost!

Click the link for a solution.
http://forums.dpreview.com/forums/read.asp?forum=1032&message=41631709

Walter Sr

gigamel wrote:
SILLY YOU!! - The OP asked about "low ISO", Dohhh!

And that is VERY disappointing that Canon is still stuck with 2005 tech performance

And even the $7000 1Dx cant save Canon - same horrible banding as 5D3:

http://forums.dpreview.com/forums/read.asp?forum=1032&message=41684369

schmegg wrote:

gigamel wrote:

Yeah, but it is still VERY disappointing that 2012, $3500 Canon tech cant beat 2008 Nikon tech, right?!!

Wrong. The 5D3 shoots video. It has much higher resolution. Better LCD. Much better viewfinder. More capable AF system. Wider ISO range. Dual card slots. Higher burst rate. Better buffer. Lighter body ....

Oh - but you aren't considering any of that are you. Silly me!

-- hide signature --

I am out to take the perfect picture, if it exits!

Keyboard shortcuts:
FForum MMy threads