K10D vs. 40D at ISO800/1600

[Jia Pu]

After seeing the endless debate about K10D and 40D, I decided to try both cameras myself. So I just got my hands on a 40D, and did some high ISO shots.

The images are on below page. Click each image to see it in full size.
http://homepage.mac.com/jia.pu/PhotoAlbum10.html

This comparison is by now means scientific. Since I don't even have similar lens on for the two bodies. And I don't claim which one is better, because some a claim is often subject and is certainly a flamebait. They have very different noise characteristics. I personally prefer 40D.

P.S. First, having got used to K10D, 40D feels quite big, if not a little awkward, in my hand. Second, which is probably predictable, 40D with 50mm f2 is a lot faster than K10D with 16-45mm (both at f4) on focusing speed and accuracy.
--
Think different.
http://www.flickr.com/photos/cheap_glass/

 

[lol101]

Hi,

What are we seeing here? jpegs straight from cameras or developed raw? 100% crops ?

Also, you mentioned that you're using a 50 on the 40D (1,6 crop= 80mm 24x36 equivalent) and a 35 on the K10D (1,5 crop= 52mm 24x36 equivalent) so I assume the pictures are not taken from the same place and don't have the same DoF, otherwise you'd have to crop the K10 image a lot more than the 40D to get the same view.

Am I correct in my assumptions?

On a side note, the 1600 ISO shot for the K10 should give you a nice entry for the "show your best VPN shots" thread:

http://forums.dpreview.com/forums/read.asp?forum=1036&message=25490489

 

[Dmitry Yu Oschepkov]

1. try to make the identical color - then compare

2. take near identical glass to compare the focusing speed. FA50/1.4 for Pentax is needed for example. It's far quicker then 16-45, of course!
--
Digital Zombie

 

[lol101]

... you can also post the 40D ISO 1600 shot to the VPN thread (except it's HPN ...)

 

[lol101]

OK, I see in the exif that you used the 35f2 on the 40D, not the 50 so my first question is irrelevant ;-)

I hope you realize that, even though the 40D has less noise, both cameras are able of much better high ISO quality if you expose the image properly in the first place.

That would, in particular, make the patterned noise visible on both shots (although it is faintest on the Canon) go away.

In high ISO shots, try to expose the scene so that the highlights are almost (but not quite) blown and you'll get much lower noise.

See this for a comparison of ISO 1600 shot on the K10 and 30D:

http://forums.dpreview.com/forums/read.asp?forum=1036&message=25516652

Also, note that, even though both are shot at ISO 1600, f4 and 1/10s, the histogram of the Canon shows it's 1/2 stop higher: always the same trick, Canon ISO 1600 is really ISO 2000 or something.

 

[Jia Pu]

40D with 50mm f2 is a lot faster than K10D with 16-45mm (both at f4)

40D with 35mm f2.
K10D with 16-45mm, and set on 35mm.

--
Think different.
http://www.flickr.com/photos/cheap_glass/

 

[Jia Pu]

What are we seeing here? jpegs straight from cameras or developed
raw? 100% crops ?

These are raw processed in Aperture with default setting.

Also, you mentioned that you're using a 50 on the 40D (1,6 crop= 80mm
24x36 equivalent) and a 35 on the K10D (1,5 crop= 52mm 24x36
equivalent) so I assume the pictures are not taken from the same
place and don't have the same DoF, otherwise you'd have to crop the
K10 image a lot more than the 40D to get the same view.

See correct in my self reply. The "50" was a typo. The 40D is using a 35mm f2.
--
Think different.
http://www.flickr.com/photos/cheap_glass/

 

[Jia Pu]

I hope you realize that, even though the 40D has less noise, both
cameras are able of much better high ISO quality if you expose the
image properly in the first place.

Yes. I realize that.

That would, in particular, make the patterned noise visible on both
shots (although it is faintest on the Canon) go away.

In high ISO shots, try to expose the scene so that the highlights are
almost (but not quite) blown and you'll get much lower noise.

See this for a comparison of ISO 1600 shot on the K10 and 30D:

http://forums.dpreview.com/forums/read.asp?forum=1036&message=25516652

Those are amazingly clear pictures. However, from an engineering background, I have some reservation on samples with such perfect exposure, simply because in reality, not every shot gets exposed perfectly. It can be the incompetence of the photographer, such as myself. Or simply because you don't have time to fine tune everything. Like most measurement in engineering field, performance in sub-optimal environment matters too. That is just my 2 cents.

 

[jm_mac]

I would love to hear your impressions on focusing speed and accuracy when you have had the chance to use the 40D more (especially with moving subjects). I have both systems (K10D and DRebXT), and need to determine if I should continue with my Canon system just for action events, or if I can go all Pentax.

There still seems to be a lot of debate as to whether Pentax focusing is as fast as Canon focusing (even with Pentax HSM lenses)

 

[lol101]

Those are amazingly clear pictures. However, from an engineering
background, I have some reservation on samples with such perfect
exposure, simply because in reality, not every shot gets exposed
perfectly. It can be the incompetence of the photographer, such as
myself. Or simply because you don't have time to fine tune
everything. Like most measurement in engineering field, performance
in sub-optimal environment matters too. That is just my 2 cents.

Yes, I understand your point.

It's clear that the Canon is better in this respect.

With the K10, you have to get the exposure right in the first place when you go to high ISO since there's not a lot of on-chip noise reduction like in the Canon.

Plus, I also noted that the 30D I'm using is generally giving 1 stop more exposure than the K10 in dark scenes when I use AE, so I definitely go manual exposure + spot metering when it's dark with the K10.

Fortunately, there is the hyper manual mode that makes shooting in manual as easy than shooting in AE modes!

 

[Jia Pu]

I was not impressed by K10D with 50mm f1.4. I ended up using that lens in manual mode most of the time.

I will play more with 40D with 35mm f2, and see how it feels.

--
Think different.
http://www.flickr.com/photos/cheap_glass/

 

[Dmitry Yu Oschepkov]

Nevertheless, the lens with fixed focal length and/or faster (F2.0 vs F4.0) has faster focusing speed - it doesn't depend on system. So your conclusions on faster focusing speed of C*non is totally erroneous couse are made with the lenses which are totally different in focusing speed even on the same system.

I can accept that C*non has faster focusing speed, or any other result in case everything is done fair. If it's not - then it's just fake, so all of the results that you presented here is most likely a fake... try to be correct and fair while trying to compare things...

--
Digital Zombie

 

[Vaards]

However, from an engineering background

From engineering background you would be shamed to post it. That name "engineer" is not given just for blue eyes. But to prove ones ability to calculate and consider all most important factors and things.

Like most measurement in engineering field, performance
in sub-optimal environment matters too. That is just my 2 cents.

It would matter in case if your wife would use one camere and you other one. And then editin pics form the same place you would notice that wife got better pics. So - there is som indicator for you - to improve your skills, to buy better lens, to reconsider your taste or to change wife.

In your case there is just slapdash simulation of optimal environment.

My $7.99

I hope you realize that, even though the 40D has less noise, both
cameras are able of much better high ISO quality if you expose the
image properly in the first place.

Yes. I realize that.

That would, in particular, make the patterned noise visible on both
shots (although it is faintest on the Canon) go away.

In high ISO shots, try to expose the scene so that the highlights are
almost (but not quite) blown and you'll get much lower noise.

See this for a comparison of ISO 1600 shot on the K10 and 30D:

http://forums.dpreview.com/forums/read.asp?forum=1036&message=25516652

Those are amazingly clear pictures. However, from an engineering
background, I have some reservation on samples with such perfect
exposure, simply because in reality, not every shot gets exposed
perfectly. It can be the incompetence of the photographer, such as
myself. Or simply because you don't have time to fine tune
everything. Like most measurement in engineering field, performance
in sub-optimal environment matters too. That is just my 2 cents.

 

[lock]

I had the opportunity to test the two systems as well ( all with sigma 17-70). Though limited by time (a couple of hours only), I quickly found out that the 40d (as well as the 30d because I did not see much of a difference) is far better in terms of noise when there is non- even distribution of (reduced levels of) light. So, with very dark areas in contrast to somewhat brighter ones, the k10d is as at least one stop worse ( iso 1600 K10d comes close to iso 3200 on the 30d/40d). This is what you could see in the second row of your pictures too. although HPN is in the 40d picture too.

I have to admit that I did not really worry about focussing speed, so I will not discuss it here. But considering everything else, price/quality the Pentax still beats Canon.

Why ? I've been going back and forth whether to decide to change brands. I decide not to. First of all, I found out I could gain quality (less noise) with faster lenses in circumstances I needed high iso most. It's not everything, but still it's defintitely a step up. Of course this will cost some, but in the end it would cost some and more if I would change brands. Second, I found out that even the 30d with fast zoom (17 mm at F2.8 is the fastest I tested) still doesn't handle the job adequately in every situation I wished it did ( did not test the 40d with that). Though I did get faster shutter speeds, there was a lack of shutter speed so evident, it made the non-is pictures on the 30D completely useless. Though slowers speeds were obtained with the K10d, more pictures were actually potential keepers. So, the conclusion was that whatever I want, I needed fast glass with IS on a Canon too. This made the choise of switching brands much easier, cause now I would face a minimum of €2000,- to get a 30d combi to suit my needs (40d : +€300,-). Of course some money would come back from selling the k10d and two lenses, but if it would at least mount up to a minimum €1200,- investment. What I need in my current setup most with the k10d is a fast 70-200 zoom (F2.8). I can do very well with the Sigma 17-70 in lower ranges for the time being, so I do not need an upgrade there ( FA 16-50). Probably within a couple of months Tamron will deliver the 70-200 F2.8 FF lens, under €700,-. That leaves me at least €500,- to spend on a new Pentax body ,which will certainly have better high iso performance than the k10d and most likely an improved focussing algorithm as well.

That should be enough. I will get a more or less similar quality set as with a new with Canon 30d ( not 40D). But the escape from a new learning curve, the slighly worse handling of the 30D in my hands (personal, very personal..), the fact that the 30d is not a new cam compared to a k..d, and the once in a lifetime investment into in-body IS is enough for me to decide to buy into Pentax again, e.g. the Metz AF 48 flash.

lock

 

[PentaxNick]

At a recent wildlife event tracking moving tigers I found the Pentax K10D slower than a friend's Canon 30D when in AF-S mode but just as fast in AF-C. It was noisier though.

Other's have suggested that the Pentax AF-S mode is more accurate and hence slower.

Nick

 

[Anastigmat]

After seeing the endless debate about K10D and 40D, I decided to try
both cameras myself. So I just got my hands on a 40D, and did some
high ISO shots.

The images are on below page. Click each image to see it in full size.
http://homepage.mac.com/jia.pu/PhotoAlbum10.html

DP Review also has high ISO comparisons between the two cameras.

This comparison is by now means scientific. Since I don't even have
similar lens on for the two bodies. And I don't claim which one is
better, because some a claim is often subject and is certainly a
flamebait. They have very different noise characteristics. I
personally prefer 40D.

I think the consensus is that the 40D is better at high ISO settings. Of course the 40D is much more expensive, nearly twice as expensive in fact. Nevertheless, Pop Photo's latest figures show that the 40D is outselling both the K10D and the Nikon D80 even though it is more expensive than either.

P.S. First, having got used to K10D, 40D feels quite big, if not a
little awkward, in my hand. Second, which is probably predictable,
40D with 50mm f2 is a lot faster than K10D with 16-45mm (both at f4)
on focusing speed and accuracy.
--
Think different.
http://www.flickr.com/photos/cheap_glass/

That is strange, both cameras should be about the same size. Personally I will save my money and upgrade to a full frame DSLR, despite the good performance of cameras like the 40D. No doubt full frame is the future, and the future is the new Canon 5D replacement.

 

[ManuH]

At a recent wildlife event tracking moving tigers I found the Pentax
K10D slower than a friend's Canon 30D when in AF-S mode but just as
fast in AF-C. It was noisier though.

AF-C is not really faster, it just let you press the shutter even if it's not completely in focus.

Other's have suggested that the Pentax AF-S mode is more accurate and
hence slower.

You cannot take the picture if the focus is not locked onto the target.

--
Manu



http://flickr.com/photos/10506176@N07/

 

[ManuH]

2. take near identical glass to compare the focusing speed. FA50/1.4
for Pentax is needed for example. It's far quicker then 16-45, of
course!

It's more complicated than that. Yes the f1.4 will let enter more light for autofocusing, but I found that in practice the FA50 f1.4 is often slower than my other lens (a Sigma 28-105 f2.8-4) in low-light. My guess is that the FA50 has very low contrast at f1.4 and it slow down the AF.

Thus to get a real comparison, we should use the exact same lens. Let's say a Sigma 17-70 at different focal lengths. And test at different light levels. It would be much more relevant than comparing apples and oranges like the OP did.

--
Manu



http://flickr.com/photos/10506176@N07/

 

[Jia Pu]

That is strange, both cameras should be about the same size.

I was surprised, because on paper the their difference is 1/10 inch. Quite different in my hands.

Maybe because the L shaped handle part of 40D is too protruded.

--
Think different.
http://www.flickr.com/photos/cheap_glass/

 

[GordonBGood]

I had the opportunity to test the two systems as well ( all with
sigma 17-70). Though limited by time (a couple of hours only), I
quickly found out that the 40d (as well as the 30d because I did not
see much of a difference) is far better in terms of noise when there
is non- even distribution of (reduced levels of) light. So, with very
dark areas in contrast to somewhat brighter ones, the k10d is as at
least one stop worse ( iso 1600 K10d comes close to iso 3200 on the
30d/40d). This is what you could see in the second row of your
pictures too. although HPN is in the 40d picture too.

lock, it should be noted that with different raw processing than Canon typically uses for JPEG's, one can completely eliminate Scan Pattern Noise, which is horizontal (HPN) in the Canon cameras; this was true for all previous models, I haven't yet checked the 40D raw files in depth. Little can be done with the VPN of the Pentax K10D without some loss of detail in filtering because the black frame offset correction where the pattern is injected into the image has already been applied to the raw data capture. Canon raw files are more "raw" with the offset still in the raw data for better processing by improved algorithms.

I'll dig into the Canon 40D raw files later with a hex editor and confirm if they still leave the raw data not dark frame offset corrected and report here later.

For those technophiles that want to know the difference in techniques, Canon typically take a running average of the masked to light dark frame offset detemination area of the sensor to apply as the dark level offset correction for each row. Any random variation of this running average, for instance due to hottish photosites, will be injected into the image as an offset error that applies to the entire scan row and will be somewhat systematic in that it will continue to affect adjacent scan rows. The alternate technique is to determine the best fit straight line of offset variation vertically across all of the black level offset rows, which can be programmed to reject inputs that are significantly different than the average and will tend to reject spurious readings anyway in finding the best fit line, then use this straight line formula to apply consistent offsets as determined on a row by row basis. This will eliminate random offset variations, especially if these calculations are carried out at an increased bit depth such as 16 bits.

It should be noted that many raw convertors of Canon raw data files still continue to use a dark frame offset technique similar to that of Canon in that they use a running average (ie. dcraw and derived convertors use the running average technique).

Regards, GordonBGood