Canon 7D and Noise issue

Associate
Joined
29 Jun 2005
Posts
1,013
Location
Newcastle
I don't know whether I'm being fussy here but I've only recently upgraded from my 40D to a 7D and was having a play earlier with a 50mm 1.4. Is it just me or does there seem to be a lot of noise in the shadows for an ISO 200 shot? This is the original RAW file with no post-processing.

Linky

After a bit of Googling it appears some people may have a dodgy piece of kit with the same issue, so hoped some second and third opinions might help make a diagnosis!
 
Soldato
Joined
8 Aug 2010
Posts
6,453
Location
Oxfordshire
Are you able to return it?
There is allot of colour banding/blotching, it's subjective but I really wouldn't be happy with it even though I like grain. Banding/blotching just looks terrible and really doesn't look like it's supposed to be there.
What I can't say is if this level of noise is typical of 7D's. I do know that 7d sensors are fab'd at different sites than other Canon camera's sharing 18mp APSC sensors, and that the 7D sensor suffers noise/banding that the others don't (at least as much). Such noise will of course look worse in areas that are underexposed and out of focus (noise is more visible in bokeh).

Personally I would return it and maybe try another one. If you decide to keep it however, overexposing 2/3's of a stop may resolve much of this issue. It's not ideal, it means your more prone to clipping highlights accidentally, but it may be the best solution given the level of banding.
 
Associate
OP
Joined
29 Jun 2005
Posts
1,013
Location
Newcastle
Yes, it was bought brand new from Warehouse Express I believe (a Christmas present), so getting it replaced shouldn't be too much of an issue. On another note, the lens doesn't seem to be anywhere near as sharp as I'd expect it to be? That was a refurb from a Canon reseller on eBay with a 1 year warranty.

As a comparison: I've taken a similar shot with my 40D on the same settings. I'd certainly be expecting at least similar from the 7D.
Linky
 
Soldato
Joined
8 Aug 2010
Posts
6,453
Location
Oxfordshire
Yes, it was bought brand new from Warehouse Express I believe (a Christmas present), so getting it replaced shouldn't be too much of an issue. On another note, the lens doesn't seem to be anywhere near as sharp as I'd expect it to be? That was a refurb from a Canon reseller on eBay with a 1 year warranty.

Hopefully some 7D users will chip in. I had a 550D and don't remember this level of noise, but the 550D wasn't reported to suffer this problem unlike the 7D.

Regarding the 50. If it's not front or back focusing, or it's isn't taking a long time to focus, then the lens is likely just fine. This AF mechanism can however be easily damaged on this lens.
Regarding sharpness, this lens isn't known for being that sharp at 1.4, neither is the 7D known for being particularly sharp either (more agressive than usual AA filter).
From what I'm told, the Sigma 50 1.4 is arguably the best 50 available for Canon.
 
Soldato
Joined
8 Aug 2010
Posts
6,453
Location
Oxfordshire
As a comparison: I've taken a similar shot with my 40D on the same settings. I'd certainly be expecting at least similar from the 7D.
Linky

To be a fair test and for the comparison to be useful, you should be comparing the two files at the same output size, thus you should try downsizing the 7d file to 10mp to match the 40d.

Don't expect miracles however over the 40D. The differences when viewed at the same output size (screen/print) should be practically indistinguishable.

http://www.dxomark.com/index.php/Cameras/Compare-Camera-Sensors/Compare-cameras-side-by-side/(appareil1)/619%7C0/(brand)/Canon/(appareil2)/180%7C0/(brand2)/Canon
 
Soldato
Joined
13 Feb 2003
Posts
10,631
Location
London
As a 7D owner, I'd say that it can start showing artefacts when using high ISO which can result in banding in the shadows, particularly in underexposed areas. I've never experienced it at ISO 200 though.

I'd certainly do some more test shots also, with other lenses and at different apertures.

The 50mm 1.4 is razor thin when used wide open, so its best to stop down a bit. Its also a lens which I find performs far better on the 5D, as on my 7D I always got purple fringing at 1.4.
 
Caporegime
Joined
18 Oct 2002
Posts
32,618
Yes, since the days of the canon 40d or even earlier canon hasn't regressed their sensor technology much at all in crop cameras, except n increasing the resolution. Almost all the canon crops sensors from the 40D onwards have a similar quantum efficiency and similar ADC readout circuitry and amplifiers. Therefore they have about the same noise floor and dynamic range. The higher resolution sensors have divided the same same signal between more pixels leading to greater per pixel noise. As an Exception says, if you down sample the higher resolution photos back to the 10mp 40d size you will find results to be about the same.

In general, sample variance between sensor I very low but not unheard of. You can try to return the 7D but if you are viewing images on your monitor at the same size as th 40d and see similar noise, or viewing both at 100% view and see more noise on the 7d then it is probably about normal for a 7d.
 
Associate
OP
Joined
29 Jun 2005
Posts
1,013
Location
Newcastle
It's about the amount of noise I'd expect in an ISO800 shot. In fact after a bit of playing around I'd say the shadow areas are about the same at 200 and 800 from the shots I've taken. Quite strange, you'd have thought I'd have Auto Lighting Optimiser on.
 
Soldato
Joined
8 Aug 2010
Posts
6,453
Location
Oxfordshire
The 50mm 1.4 is razor thin when used wide open, so its best to stop down a bit. Its also a lens which I find performs far better on the 5D, as on my 7D I always got purple fringing at 1.4.

Yes purple fringing should appear around 50-60% reduced on a 5d when viewed at the same output size as the 7d (at least in the centre).

Edit:
Also people don't buy a F1.4 lens to use at F2.
 
Last edited:
Soldato
Joined
13 Feb 2003
Posts
10,631
Location
London
It's about the amount of noise I'd expect in an ISO800 shot. In fact after a bit of playing around I'd say the shadow areas are about the same at 200 and 800 from the shots I've taken. Quite strange, you'd have thought I'd have Auto Lighting Optimiser on.

Like I said, give it some more tests and see what happens.

Yes purple fringing should appear around 50-60% reduced on a 5d when viewed at the same output size as the 7d (at least in the centre).

It was more a purple fringing that showed up on highlights all over the image, especially noticeable on light sources at night. It doesn't occur at all on full frame.

Edit:
Also people don't buy a F1.4 lens to use at F2.

True, but if I'm using the lens creatively where I wanted a shallow depth of field, I'd stop back up a couple of thirds to 1.6 or 1.8 as it gives a bit better sharpness, without loosing the bokeh.

If I'm shooting somewhere low light where f/1.4 is the only choice because the ISO has gotten far too high, then so be it.

Ultimately what I was originally trying to say is that despite the effective focal length difference, the 50mm f/1.4 works far better on full frame than on my crop.
 
Soldato
Joined
8 Aug 2010
Posts
6,453
Location
Oxfordshire
True, but if I'm using the lens creatively where I wanted a shallow depth of field, I'd stop back up a couple of thirds to 1.6 or 1.8 as it gives a bit better sharpness, without loosing the bokeh.

That works better for lenses with rounded aperture blades, because as soon as you stop down a 1/3 of a stop the bokeh takes on the shape of the aperture blades. Bokeh is subjective, but most people prefer circular airy discs than hexagonal ones (depending on number of aperture blades).
 
Soldato
Joined
13 Feb 2003
Posts
10,631
Location
London
The Canon f/1.4 has 8 blades so it actually produces quite pleasing specular highlights. My point though was that stopping down a 1/3 or so isn't going to significantly change the out of focus areas, so its worth it for a tiny bit more sharpness on your subject.
 
Soldato
Joined
5 Jan 2003
Posts
3,517
Location
Somewhere in the middle
Sorry can't help the OP, but I down loaded the samples out of interest and was quite surprised. Noise is not only in the shadow, but auto level the file and look how the highlights grain up as well.
Made me compare some of my files, my old Contax N Digital (Full frame) from early 90's who's max ISO is 400 !!! has less noise and super smoother film like grain.
My "old" 5D mk1 is much cleaner and no banding either......
Lesson for today, buy Full Frame or Nikon....or both !?!?!
 
Soldato
Joined
8 Aug 2010
Posts
6,453
Location
Oxfordshire
Sorry can't help the OP, but I down loaded the samples out of interest and was quite surprised. Noise is not only in the shadow, but auto level the file and look how the highlights grain up as well.
Made me compare some of my files, my old Contax N Digital (Full frame) from early 90's who's max ISO is 400 !!! has less noise and super smoother film like grain.
My "old" 5D mk1 is much cleaner and no banding either......
Lesson for today, buy Full Frame or Nikon....or both !?!?!

Try playing with this .nef
Underexposed 3 stops but no banding in sight from what I can tell.
 
Soldato
Joined
5 Jan 2003
Posts
3,517
Location
Somewhere in the middle
Impressive I have to say....... balances out and becomes a very useable shot from being 3 stops out....... Being a Landscape shooter sometime you have to rescue the shadows when the dynamic range is larger........ "Expose for the highlights and process for the shadows"
 
Soldato
Joined
8 Aug 2010
Posts
6,453
Location
Oxfordshire
Impressive I have to say....... balances out and becomes a very useable shot from being 3 stops out....... Being a Landscape shooter sometime you have to rescue the shadows when the dynamic range is larger........ "Expose for the highlights and process for the shadows"

Yup, that's what I do with landscapes (not that I'v done many), although haven't really been out to put the D800E through it's paces yet.
You shouldn't have any problem exposing for the sky during sunset, and then just using the gradient tool in LR to increase exposure of the foreground a couple of stops (no need for grads) and you then have full control over how soft of hard the gradient 'edge' is.

Edit:

Also tried the same with 5 stops of recovery, which also worked equally as well.
 
Last edited:
Back
Top Bottom