Noise pictures of the 10D are without NR- D60 with NR

  • Thread starter Thread starter David Martin
  • Start date Start date
http://web.canon.jp/Imaging/EOS10D/

Rich
-
"...In fact, reproduction is so acurate that noise reduction is
unnecessary, even with long exposures."
Presumably quoted from a Canon press release
--
DaveMart
I've e-mailed Dave asking him to confirm. I will report back to the
forum
Regards,
--
DaveMart
--
Drop by for more digital discussions..
http://www.dv-info.net/ib/index.php
He says that Canon told him that they NR had beenupdated from the D60 but he may have misunderstood.

He is contacting Canon and will update his review if needs be and will keep me informed.
I might even let you guys know what he says!
Regards,
--
DaveMart
 
From Phil's review it appears that the D60 has auto NR for long
exposures.
NR isn't specific enough.

There are two kinds of noise reduction that have been employed on Canon CMOS dSLRs:

1. Dark Frame Subtraction. Immediately after taking the picture, the camera makes another exposure with the shutter fully closed. It subtracts the data from the first picture. This clears "hot pixels" whose values are non-zero due to dark current flow.

2. Ongoing on-chip noise reduction, done during the exposure. Canon has a term for this, but I don't remember it right now. This form of noise reduction is what set the D60 (and now the 1Ds) apart from all predecessors and allowed exposure times of minutes instead of seconds.

You can tell if the camera is doing Dark Frame Subtraction because there is a delay immediately after taking a shot equal to the length of the shot. Take a 3 second shot with DFS on and the camera goes dormant for 3 more seconds immediately afterward.
  • The D30 employs only Dark Frame Subtraction noise reduction.
  • The D60 employs only ongoing noise reduction. Thus the amplifier heat noise and some hot pixels.
  • The 1Ds employs type ongoing noise reduction and has the option to add Dark Frame Subtraction. Best of both worlds, and sure enough folks are reporting "stellar" long exposure performance.
The 10D MAY employ ongoing noise reduction - I say this because I think it may be part of the CMOS chip design. It seems that the chip is not all-new but a close relative of the D60's chip.

There is no custom function I can see to add Dark Frame Subtraction on the 10D - a disappointment to say the least! Perhaps Canon could add it in a future firmware release - it's probably a feature managed by software and not the Digic - though I doubt Canon will add any more features at this price (they have not done so with the D30 nor D60). Perhaps they chose to leave it out to make the 1Ds a little more desirable.

-Noel
 
... Dark Frame Subtraction, that is. I did my own on the 1D, even though it has it. I can do it better!
KP
From Phil's review it appears that the D60 has auto NR for long
exposures.
NR isn't specific enough.

There are two kinds of noise reduction that have been employed on
Canon CMOS dSLRs:

1. Dark Frame Subtraction. Immediately after taking the picture,
the camera makes another exposure with the shutter fully closed.
It subtracts the data from the first picture. This clears "hot
pixels" whose values are non-zero due to dark current flow.

2. Ongoing on-chip noise reduction, done during the exposure.
Canon has a term for this, but I don't remember it right now. This
form of noise reduction is what set the D60 (and now the 1Ds) apart
from all predecessors and allowed exposure times of minutes instead
of seconds.

You can tell if the camera is doing Dark Frame Subtraction because
there is a delay immediately after taking a shot equal to the
length of the shot. Take a 3 second shot with DFS on and the
camera goes dormant for 3 more seconds immediately afterward.
  • The D30 employs only Dark Frame Subtraction noise reduction.
  • The D60 employs only ongoing noise reduction. Thus the amplifier
heat noise and some hot pixels.
  • The 1Ds employs type ongoing noise reduction and has the option
to add Dark Frame Subtraction. Best of both worlds, and sure
enough folks are reporting "stellar" long exposure performance.

The 10D MAY employ ongoing noise reduction - I say this because I
think it may be part of the CMOS chip design. It seems that the
chip is not all-new but a close relative of the D60's chip.

There is no custom function I can see to add Dark Frame Subtraction
on the 10D - a disappointment to say the least! Perhaps Canon
could add it in a future firmware release - it's probably a feature
managed by software and not the Digic - though I doubt Canon will
add any more features at this price (they have not done so with the
D30 nor D60). Perhaps they chose to leave it out to make the 1Ds a
little more desirable.

-Noel
--

29 lbs. of Canon stuff in a backpack that I carry everywhere. A closet full of things that are banned in Britain. A minivan and a Fender Stratocaster. A three bedroom ranch with three owls on an acre. An aversion to rumours. Also, absolutely no Canon 1200mm f/5.6. Yet.
 
From Phil's review it appears that the D60 has auto NR for long
exposures.
NR isn't specific enough.

There are two kinds of noise reduction that have been employed on
Canon CMOS dSLRs:

1. Dark Frame Subtraction. Immediately after taking the picture,
the camera makes another exposure with the shutter fully closed.
It subtracts the data from the first picture. This clears "hot
pixels" whose values are non-zero due to dark current flow.

2. Ongoing on-chip noise reduction, done during the exposure.
Canon has a term for this, but I don't remember it right now. This
form of noise reduction is what set the D60 (and now the 1Ds) apart
from all predecessors and allowed exposure times of minutes instead
of seconds.

You can tell if the camera is doing Dark Frame Subtraction because
there is a delay immediately after taking a shot equal to the
length of the shot. Take a 3 second shot with DFS on and the
camera goes dormant for 3 more seconds immediately afterward.
  • The D30 employs only Dark Frame Subtraction noise reduction.
  • The D60 employs only ongoing noise reduction. Thus the amplifier
heat noise and some hot pixels.
  • The 1Ds employs type ongoing noise reduction and has the option
to add Dark Frame Subtraction. Best of both worlds, and sure
enough folks are reporting "stellar" long exposure performance.

The 10D MAY employ ongoing noise reduction - I say this because I
think it may be part of the CMOS chip design. It seems that the
chip is not all-new but a close relative of the D60's chip.

There is no custom function I can see to add Dark Frame Subtraction
on the 10D - a disappointment to say the least! Perhaps Canon
could add it in a future firmware release - it's probably a feature
managed by software and not the Digic - though I doubt Canon will
add any more features at this price (they have not done so with the
D30 nor D60). Perhaps they chose to leave it out to make the 1Ds a
little more desirable.

-Noel
Agreed. There is certainly no DFS on the 10D.

The doubt seems to arise as whether there is any on-going NR either. Dave at IR is coming back to me on this.

If there isn't the images may (repeat may ) clean up to a greater extent than is possible on the D60, you just have to do it out of camera.

Certainly looking at the type of noise from the 10d posted here I find it difficult to imagine that any in-camera NR has been done. It's just too easy to take out.
I'll keep you posted.
Regards,
--
DaveMart
 
And do you think that Canon is going to care about your decision? This issue is not a FLAW. It is a misplaced set of expectations by some uninformed and unreasonable customers. This camera's noise performance is SIGNIFICANTLY better than the D60 - I know, I have both - and it is hot pixel free up to a couple / three minutes exposure - GOOD ENOUGH for 99.9% of serious amateurs and professionals alike.

Enjoy your D30/G3.

John
I decided to open up another thread about this.
From Phil's review it appears that the D60 has auto NR for long
exposures.
The 10D doesn't appear to have any NR.
Jerry thought that it was automatically on, but that doesn't appear
to be the case to me from Phil's review.
This on it's own should account for the differnces in the pictures,
and means that you just process the pics from the 10D out of camera
instead of in for NR and they should be pretty comparable.
several people have remarked on how easy it was to clean them.
The D60 is just doing it in the camera.
What do you think?
Regards
--
DaveMart
I think noise/hot pixels at long exposures in this camera must be
considered a serious problem. C'mon, even if its cheaper than the
D60, i can't buy a camera with an issue like that. I would feel
that my money was not well spent... Even if you can fix the image
using Photoshop, if i'm going to spend big bucks on a camera i want
it to work well in almost every condition with the least possible
flaws.

I guess we should start to flood Canon's email with reports about
this flaw, so that if its possible to solve this problem (firmware,
call the cameras back, whatever), this can be done as soon as
possible. ;)

I'm gonna stick with my D30 and G3 for now. If this problem is
fixed in a near future (VERY near please), than i will consider
buying this camera.
Apart from this problem, the camera looks very promising...

PS: Sorry about the bad english...i'm portuguese and i skipped the
English classes... :D
 
I decided to open up another thread about this.
From Phil's review it appears that the D60 has auto NR for long
exposures.
The 10D doesn't appear to have any NR.
Jerry thought that it was automatically on, but that doesn't appear
to be the case to me from Phil's review.
This on it's own should account for the differnces in the pictures,
and means that you just process the pics from the 10D out of camera
instead of in for NR and they should be pretty comparable.
several people have remarked on how easy it was to clean them.
The D60 is just doing it in the camera.
What do you think?
Regards
--
DaveMart
I think noise/hot pixels at long exposures in this camera must be
considered a serious problem. C'mon, even if its cheaper than the
D60, i can't buy a camera with an issue like that. I would feel
that my money was not well spent... Even if you can fix the image
using Photoshop, if i'm going to spend big bucks on a camera i want
it to work well in almost every condition with the least possible
flaws.

I guess we should start to flood Canon's email with reports about
this flaw, so that if its possible to solve this problem (firmware,
call the cameras back, whatever), this can be done as soon as
possible. ;)

I'm gonna stick with my D30 and G3 for now. If this problem is
fixed in a near future (VERY near please), than i will consider
buying this camera.
Apart from this problem, the camera looks very promising...

PS: Sorry about the bad english...i'm portuguese and i skipped the
English classes... :D
Your English is fine - so much better than my Potuguese!

You will, however, find that there are issues with every camera - check out recent posts on dust in the 1Ds, which costs lots of money.

It isn't even quite sure that there is any problem yet - check out my recent thread on NR on the 10D.

I think you will find that the extra resolution and better AF of the 10D would make you very happy whatever the result of this particular issue, but it never hurts to wait awhile and let others sort out any problems.
Best of luck whatever you decide.
--
DaveMart
 
... Dark Frame Subtraction, that is. I did my own on the 1D, even
though it has it. I can do it better!
Agreed. I didn't mean to imply that you couldn't. Sorry for any confusion.

I would rather have the option of having it done for me seamlessly, though. I can't imagine that there could be a better after-the-fact method than could be engineered into the camera, where the Raw data is available...

Actually, one could imagine being able to do it very well indeed OUT of camera with a raw image and a Raw dark frame... A new feature to request of the various Raw image converter makers. :)

-Noel
 
The doubt seems to arise as whether there is any on-going NR
either. Dave at IR is coming back to me on this.
I'm guessing there's at least SOME NR going on, or we wouldn't be seeing exposure times getting into the minutes at all. FYI, the press release said only this about noise:

"A new amplifier circuit boosts the S/N (signal to-noise) ratio to provide an extended sensitivity range from ISO 100 to 3200 and superior noise reduction at all ISO speed settings. "

-Noel
 
... A new
feature to request of the various Raw image converter makers. :)

-Noel
Noel,

I'm under the impression the dark frame has to be captured by the camera. For example I take a 3 minute exposure of the bridge at night. Then I put on a lens cap and shoot a 3 minute "dark frame". Then I do some diddling with both those pics in photoshop to subtract the dark frame from the first frame. That frame is unique to that shoot because it mimics (as closely as possible) exactly what kind of noise my camera creates.

Is your comment above indicating a desire for Raw image converter makers to create a simple thing to merge those two frames, or to create a dark frame for you?

Please clarify. And while you are at it, anyone who knows, please indicate how to "subtract" that dark frame in photoshop. Thanks so much.

Bob
 
... as a couple of products (such as Powershovel) will directly convert a RAW without interpolation ... you could take these and mix them. I may try that later, as I have a noise machine (1D).
KP
... Dark Frame Subtraction, that is. I did my own on the 1D, even
though it has it. I can do it better!
Agreed. I didn't mean to imply that you couldn't. Sorry for any
confusion.

I would rather have the option of having it done for me seamlessly,
though. I can't imagine that there could be a better
after-the-fact method than could be engineered into the camera,
where the Raw data is available...

Actually, one could imagine being able to do it very well indeed
OUT of camera with a raw image and a Raw dark frame... A new
feature to request of the various Raw image converter makers. :)

-Noel
--

29 lbs. of Canon stuff in a backpack that I carry everywhere. A closet full of things that are banned in Britain. A minivan and a Fender Stratocaster. A three bedroom ranch with three owls on an acre. An aversion to rumours. Also, absolutely no Canon 1200mm f/5.6. Yet.
 
Something else that caught my eye:
The main limitation of the camera here is that the autofocus system has > a hard time focusing at light levels below somewhere between 1/4 and > 1/2 foot-candle (in the general vicinity of 3-5 lux).
Improved autofocus is my reason #1 to upgrade to this camera. How good is it's autofocus anyway?

Cheers,
-Miro

http://www.chromasoftware.com
also he said this"While most other high-end digicams on the market
use a dark frame subtraction method to deal with image noise, the
10D (like the D60 before it) appears to be doing something very
different, as there's very little delay between the end of the
primary exposure and the writing of the image file to the memory
card. There's clearly no "dark frame" exposure involved. I suspect
that this advanced noise reduction processing in the 10D is another
consequence of the "active pixel" CMOS technology Canon developed
internally. Having active circuitry associated with each pixel in
the sensor array allows lots of fancy processing that would be
impossible otherwise, and it looks like Canon's noise reduction
system takes advantage of this"

I am genuinely interested and curious as a newcomer to dslr,
because I don't want to spend 1500 on something I may be sorry
for.....but which is correct....in your opinion was the canon rep
you spoke with knowledgeable and credible? thanks..t
looks to me like what you are saying might well be the case.
Imaging Resources certainly seems to feel that it is. See:
http://www.imaging-resource.com/PRODS/E10D/E10DPICS.HTM
and look under 'Low-light'
Cattle dog, however, found some Canon info indicating that there
was NO NR on the 10D
Cattle-dog?Sources?
My own gues is that they have NR, but not NR for long exposures as
the noise which remains look too easy to knock out with NR and I
can't see how any long-exposure NR routine could miss it.
Anyone? Help!
--
DaveMart
 
I tested a PRODUCTION 10D, a pre-production 10D and a D60, the results D60 and 10d almost identical alll the way to 4 minutes with a really slight edge to the 10D. The Canon reps 10D, truly awful beyond 2 seconds. The lesson here, don't trust any image on the net until there are production cameras the will allow you to prove if for yourself.
The doubt seems to arise as whether there is any on-going NR
either. Dave at IR is coming back to me on this.
I'm guessing there's at least SOME NR going on, or we wouldn't be
seeing exposure times getting into the minutes at all. FYI, the
press release said only this about noise:

"A new amplifier circuit boosts the S/N (signal to-noise) ratio to
provide an extended sensitivity range from ISO 100 to 3200 and
superior noise reduction at all ISO speed settings. "

-Noel
 
Hi Barry - Any other noticeable differences between the reps 10D and the Production 10D that you noticed?
I tested a PRODUCTION 10D, a pre-production 10D and a D60, the
results D60 and 10d almost identical alll the way to 4 minutes with
a really slight edge to the 10D. The Canon reps 10D, truly awful
beyond 2 seconds. The lesson here, don't trust any image on the
net until there are production cameras the will allow you to prove
if for yourself.
 
The issue of noise started when we opened for business so it was the focus for the day. In all honesty I don't think anyone who was at the demo found anything else to dislike at all (A few D60 owners were dismayed with the $700 reduction in price).
I tested a PRODUCTION 10D, a pre-production 10D and a D60, the
results D60 and 10d almost identical alll the way to 4 minutes with
a really slight edge to the 10D. The Canon reps 10D, truly awful
beyond 2 seconds. The lesson here, don't trust any image on the
net until there are production cameras the will allow you to prove
if for yourself.
 
that the focus of the day wasn't on how great the focus improvement was on the 10D over the D60.

I got distracted by it for awhile because a lot of energy was spent there by people who have a right to be concerned as they shoot a lot of long exposure stuff! I don't shoot long exposures on a weekly basis. I do shoot portraits, cars, behnd the scenes at trade shows and a lot of candids and don't have the means for a 1Ds so I was very interested in the 10D and it's capabilities!

From what I saw today the 10D will work well for 95% of my needs and I welcome the price reduction with open arms!
 
that the focus of the day wasn't on how great the focus improvement
was on the 10D over the D60.

I got distracted by it for awhile because a lot of energy was spent
there by people who have a right to be concerned as they shoot a
lot of long exposure stuff! I don't shoot long exposures on a
weekly basis. I do shoot portraits, cars, behnd the scenes at trade
shows and a lot of candids and don't have the means for a 1Ds so I
was very interested in the 10D and it's capabilities!

From what I saw today the 10D will work well for 95% of my needs
and I welcome the price reduction with open arms!
 
if i'm going to spend big bucks on a camera i want
it to work well in almost every condition with the least possible
flaws.
Geez, sounds like you're talking about the 1D, not the 10D. Try a five minute exposure with a 1D and THEN we can talk about flaws. And that's a camera that costs over twice as much as the 10D.

Regards
Stefan
 
I'm under the impression the dark frame has to be captured by the
camera.
That is correct. Sorry if I wasn't clear. I was implying that if you took a long exposure, say 30 minutes, then took a dark frame exposure for 30 minutes - both in Raw mode - then fed both of those files into a hypothetical new, improved converter that could subtract the dark frame in "internal" representation (i.e., 12 bit values, one per sensor site), that you might get a better result than if the work is done in Photoshop after interpolation.

It is also possible that a dark frame or frames could be kept on file, because at least some of the characteristics of your sensor don't vary much over time. Note, for example, that Maurice's two indoor shots have the "hot pixels" in the same locations. Thus a dark frame (or set of dark frames taken at different durations) - even taken after the fact - could be used mathematically to improve long exposure images. I believe QImage already provides this sort of functionality, though not with Raw files.
Please clarify. And while you are at it, anyone who knows, please
indicate how to "subtract" that dark frame in photoshop. Thanks so
much.
Here's what I would do:

1. Open the actual image in photoshop.
2. Open the dark frame image in a second window. Select all, copy.
3. Activate the first window again, and paste.
4. Go to the Layers window, and change the mixing mode to "Difference".
5. Layer-Flatten Image.

That's it! Have fun.

-Noel
 

Keyboard shortcuts

Back
Top