Forgot Password
Pentax Camera Forums Home
 

Reply
Show Printable Version Search this Thread
01-10-2010, 02:50 PM   #1
Veteran Member
pcarfan's Avatar

Join Date: Sep 2007
Location: Dayton, Ohio
Photos: Gallery | Albums
Posts: 2,960
Unscientific K-7 ISO testing

I was impressed with the 'unscientific' testing done by Mr.Turnip with his K-x. It is frigidly cold outside and I am cocooned indoors, and decided to test my K-7 to see how it retains detail. IMO, noise test ignoring fine detail preservation means nothing, and thus my reason to be impressed with the test posted.

Images taken in RAW opened in LR and saved. NO NR unless specified. ALL EXIF intact.

I think iso 6400 is totally atrocious in the K-7, but up to iso 3200 is usable with great detail preservation.

iso100

iso200

iso400

iso800

iso1600

iso3200

iso6400 (Some banding)


100% crops- The dots inside the lens diagram shows very fine detail.
iso100

iso200

iso400

iso800

iso1600

iso3200 (The fine detail and noise is left behind, and now it is up to me to balance noise and detail like below)

iso3200 with LR chroma NR with practically no loss of resolution

iso3200CS2 luma NR with some loss of detail (choice to preserve detail - I love to have this choice left to me)

Now, High NR without worrying about fine detail (choice to remove noise-if the camera did this, I will have no choice to go like the image above)

iso6400

iso6400 with NR with severe loss of detail


Hopefully this shows detail is everything and leaving detail and noise is always better than removing noise at the expense of detail. Evaluating noise level has to be tightly tied to it's effects on detail like Mr.Turnip has shown. This loss of detail is not even apparent unless viewed close to 100%, check these full images at iso 3200 and iso 6400 with NR to eliminate almost all noise and it is sort of difficult to see loss of detail.

iso3200 with almost all noise removed

iso6400 with almost all noise removed (chroma noise still evident)



Last edited by Damn Brit; 01-10-2010 at 04:26 PM. Reason: Added imgwide tags
01-10-2010, 03:01 PM   #2
Veteran Member




Join Date: Dec 2006
Location: Ontario
Posts: 750
When I first got my k-7 and compared it to my old k10, I was amazed by the difference between the two cameras in a lot of areas, but I was very surprised at how well the k-7 did at ISO 400, 800 and 1600 compared to my k10, which I wouldn't use past ISO 400.

Thanks for the samples.
01-10-2010, 03:34 PM   #3
Site Supporter




Join Date: Mar 2009
Location: Gladys, Virginia
Photos: Gallery
Posts: 16,230
Thanks for this. I think sometimes in the slamming of the high noise performance of the K7 is lost the fact that it really is pretty good. Sure it loses some to the Kx in the dynamic range department, but compared to, say, the K10, it is pretty impressive. At the same time, whatever the camera, high iso is reserved for situations where you have to get the photo and don't have any choice.
01-10-2010, 04:18 PM   #4
Veteran Member
pcarfan's Avatar

Join Date: Sep 2007
Location: Dayton, Ohio
Photos: Gallery | Albums
Posts: 2,960
Original Poster
I am glad a couple of people found it useful, I also did this for my own use. I am tempted to upgrade to the K-x for it's 1 to possibly 1.5 stop high iso noise advantage, but kept going towards getting more lenses instead. I got the 35/f2 and will soon be getting the DA15 and sigma 30/1.4, wanted to make sure I am making the right choice in choosing more and more lenses instead of the k-x, as I really do think the K-7's approach to iso 3200 is impressive and I don't need anything higher at the expense of losing some major features.

01-10-2010, 04:26 PM   #5
Damn Brit
Guest




Please use IMGWIDE tags for large images, thanks.
01-10-2010, 04:49 PM   #6
Veteran Member
pcarfan's Avatar

Join Date: Sep 2007
Location: Dayton, Ohio
Photos: Gallery | Albums
Posts: 2,960
Original Poster
QuoteOriginally posted by Damn Brit Quote
Please use IMGWIDE tags for large images, thanks.
Will do, .....I wasn't too sure at what resolution it is considered too long. If it fits my screen I leave it as it is , but this was probably too long for most monitors. Thanks for the change.
01-10-2010, 04:54 PM   #7
Damn Brit
Guest




QuoteOriginally posted by pcarfan Quote
Will do, .....I wasn't too sure at what resolution it is considered too long. If it fits my screen I leave it as it is , but this was probably too long for most monitors. Thanks for the change.
The forum limit is 900 wide so anything over that is best given IMGWIDE tags.
That is what I believe to be correct anyway.
01-10-2010, 06:22 PM   #8
Pentaxian
jgredline's Avatar

Join Date: Oct 2007
Location: LosAngeles, Ca.
Photos: Albums
Posts: 10,587
Bad enough that the flagship k-7 lags behind the K-X but it still lags behind the K20D that it replaced.

01-10-2010, 06:56 PM   #9
Veteran Member




Join Date: Dec 2006
Location: Ontario
Posts: 750
QuoteOriginally posted by jgredline Quote
Bad enough that the flagship k-7 lags behind the K-X but it still lags behind the K20D that it replaced.
Not having shot with a k20d, I'd like to see just how much of a "lag" there really is between the two. I doubt there's a lot of difference at all.

ETA: http://www.dpreview.com/reviews/pentaxk7/page29.asp

"When looking at the K-7's high ISO output it is obvious that Pentax, despite deploying an updated version of the K20D sensor in the new model, hasn't made a huge step forward in terms of noise reduction processing. In the samples below the K20D shows more chroma noise than the K-7 but at its default settings noise reduction is switched off while on the K-7 it is set to 'Medium'. Activating it on the K20D will get you very similar results to the K-7."
Pentax K-7 Review, October 2009
by Lars Rehm, Don Wan and Richard Butler

Last edited by Andrew Faires; 01-10-2010 at 07:05 PM.
01-10-2010, 09:22 PM   #10
Pentaxian
JohnBee's Avatar

Join Date: Sep 2008
Location: front of computer
Posts: 4,620
The banding starts at ISO1600 and gets worst from there.
This was in part the main reason why we didn't upgrade to the K-7 this year. As the K7 performed worst than our K20D's during our own ISO tests.

Having said that, it's not entirely fair to judge a camera on one attribute alone. And so I think it's only fair to say that as far as the K7 goes, noise is only part of the picture.
01-10-2010, 10:50 PM   #11
Veteran Member




Join Date: Nov 2008
Location: Trabzon/Turkey
Photos: Gallery
Posts: 1,010
QuoteOriginally posted by pcarfan Quote

Hopefully this shows detail is everything and leaving detail and noise is always better than removing noise at the expense of detail. Evaluating noise level has to be tightly tied to it's effects on detail like Mr.Turnip has shown. This loss of detail is not even apparent unless viewed close to 100%, check these full images at iso 3200 and iso 6400 with NR to eliminate almost all noise and it is sort of difficult to see loss of detail.

iso3200 with almost all noise removed
4263954630_c78c4b9bcb_o.jpg
iso6400 with almost all noise removed (chroma noise still evident)
4263955078_96742736c2_o.jpg
Thanks much for the efforts,

I have a question, in the last two pictures, was NR applied in camera or did you use an external application?
01-11-2010, 05:19 AM   #12
Veteran Member
pcarfan's Avatar

Join Date: Sep 2007
Location: Dayton, Ohio
Photos: Gallery | Albums
Posts: 2,960
Original Poster
QuoteOriginally posted by cbaytan Quote
Thanks much for the efforts,

I have a question, in the last two pictures, was NR applied in camera or did you use an external application?
External application with noiseware.
01-11-2010, 06:30 AM   #13
Junior Member
Mr.Turnip's Avatar

Join Date: Jan 2007
Location: Denmark
Posts: 33
QuoteOriginally posted by pcarfan Quote

iso3200 with almost all noise removed
This is most impressive...Can you quickly tell how you did it?
01-11-2010, 08:51 AM   #14
Veteran Member
pcarfan's Avatar

Join Date: Sep 2007
Location: Dayton, Ohio
Photos: Gallery | Albums
Posts: 2,960
Original Poster
Claus, I use the noiseware paid version (US$50) plug-in with photoshop CS2. This plug-in allows to control noisereduction strength for highlight, midtone and shadows, and much more including confining various degrees of NR to certain color only as well. It also has various pre-sets.

I think I chose the 'weak noise' pre-set for iso 3200 image as I thought the noise wasn't much (chose FULL noise reduction for iso6400 which obliterates detail), but the dark shadows still had noise left, and the percentage of noise reduction on the shadows was already 100%, but strength was set at the minumum, so I maximized the strength on the shadows until the noise disapperaed there. (This is not a panacea, as if the image has a lot of transition zones between midtones and shadows, then you will get blotches in those areas by doing this).

I've done other iso tests before, but I liked your approach where you see how noise and detail fare as iso is boosted in 1-stop increments.

I find iso 3200 very usable with proper NR, I've used it indoors in available light and found it very usable. But, for ultimate IQ I only like iso100 and may be 200, and that is where you can actually feel the images and almost reach and touch it with the proper lenses.

Some test shots....lowlight (iso3200, f2, 1/15th handheld)


100% crop (Well using this 'click to see a larger image' doesn't work aat all, as even when clicked it's not the 100% crop I posted)


Even withour NR iso 3200 is ok
in good light (iso3200, f2, 1/100) light background with NO NR

In bad light (iso3200, f2, 1/8) with dark background to show nosie easily and handheld, but no NR.



I've never seen this banding that showed up in my first psot before and I did quite a few iso3200 shots before and it wasn't there. I wonder whether it's related to the firmware updates? I've pushed iso 3200 images by 1.5 stops for a iso 9600 images and didn't see banding, I am going to check those old files and see (P.S: I don't think even the pictures here taken at iso3200 shows any banding, do they??). I will report back with some of thoise images if I can find any. Also, as this banding looks uniform I will try dark frame subtraction and see whether that will help.

Last edited by pcarfan; 01-11-2010 at 09:08 AM.
01-11-2010, 09:01 AM   #15
Veteran Member
pcarfan's Avatar

Join Date: Sep 2007
Location: Dayton, Ohio
Photos: Gallery | Albums
Posts: 2,960
Original Poster
QuoteOriginally posted by jgredline Quote
Bad enough that the flagship k-7 lags behind the K-X but it still lags behind the K20D that it replaced.
By how many stops do you think the K-7 iso 3200 image lags behind the K20D image ? I think any difference is not field relevant.

K-x is indeed the king of high iso. I wish the K-7 is better by 1 to 1.5 stops than it is, but it is not, and it's not a major impediment for me.
Reply

Bookmarks
  • Submit Thread to Facebook Facebook
  • Submit Thread to Twitter Twitter
  • Submit Thread to Digg Digg
Tags - Make this thread easier to find by adding keywords to it!
camera, choice, detail, dslr, iso, iso3200, iso6400, k-7, loss, noise, nr, photography, test
Thread Tools Search this Thread
Search this Thread:

Advanced Search


Similar Threads
Thread Thread Starter Forum Replies Last Post
K-5 vs K-7 First ISO performance testing (ISO-6400) starscream Pentax News and Rumors 95 09-25-2010 07:02 PM
Pentax DA* 55mm 1.4 SDM (unscientific test) asdf Pentax SLR Lens Discussion 7 01-29-2010 02:20 PM
Unscientific Kx ISO testing Mr.Turnip Pentax DSLR Discussion 26 01-13-2010 02:31 PM
Very, Very unscientific, but of interest Ed in GA Pentax DSLR Discussion 41 06-21-2008 12:00 PM
The big unscientific RAW Converter Comparison HogRider Photographic Technique 0 02-21-2007 11:27 AM



All times are GMT -7. The time now is 07:41 PM. | See also: NikonForums.com, CanonForums.com part of our network of photo forums!
  • Red (Default)
  • Green
  • Gray
  • Dark
  • Dark Yellow
  • Dark Blue
  • Old Red
  • Old Green
  • Old Gray
  • Dial-Up Style
Hello! It's great to see you back on the forum! Have you considered joining the community?
register
Creating a FREE ACCOUNT takes under a minute, removes ads, and lets you post! [Dismiss]
Top