Results 1 to 13 of 13

Thread: noise comparison ISO1600

  1. #1

    Default noise comparison ISO1600

    hi people,

    got some ISO1600 samples from the following site

    http://digitalcamera.impress.co.jp/0.../index_iso.htm

    ran neat image (default settings) through some of the ISO 1600 images. here are the output for comparison. Please note not just the level of noise, but also the level of detail.
    The 5D and D200 is downsampled to 6meg. All 100% crop


    canon 5D


    canon 350


    nikon D200


    nikon d50


    s3pro
    Last edited by wind30; 30th May 2006 at 07:56 AM.

  2. #2

    Default Re: noise comparison ISO1600

    pentax ds2


    oly E500

  3. #3
    Senior Member
    Join Date
    Dec 2004
    Location
    NTU and Wdls
    Posts
    2,622

    Default Re: noise comparison ISO1600

    I don't think there's a need to run through neat image. Otherwise you can't give an objective test. For example, if you take the oly E500 image, and actually run through neat image to actually clean up noise, I doubt much details would be left. *points to the shadow region, notice the colour patches there?*

    I assume no neat image for 5D.
    Neat image is used to remove noise.
    If you want to use it, best for you to redo, clean up the images to 5D standard of noise free-ness, then compare.
    Last edited by unseen; 30th May 2006 at 02:26 PM.

  4. #4

    Default Re: noise comparison ISO1600

    I RAN ALL the images through neat image, including 5D.

    Olympus E500 has the problem with the color blotches in the shadow region at ISO1600 which cannot be removed by neat image. But at ISO800 without the color blotches, its noise performance is not really that bad and probably better/similar to D50/s3pro.

    Look at D50. A LOT of reviews (dpreview) says it is very low noise, but it is also very low details too. And canon 350D, people say canon has plastic look but it has the highest detail for the cheap DSLRs.

    The point I want to make is when you judge ISO performance, it is better to run the image file through neat image (or similar). Some makers choose not to do much noise reduction in camera to preserve the details and that doesn't mean the ISO performance is bad.

  5. #5

    Default Re: noise comparison ISO1600

    this looks more like a demonstration of how good neat image is rather than an objective comparison of noise at ISO 1600.

  6. #6
    Senior Member
    Join Date
    Feb 2003
    Location
    The heart of the Abyss
    Posts
    2,307

    Default Re: noise comparison ISO1600

    Quote Originally Posted by madmacs
    this looks more like a demonstration of how good neat image is rather than an objective comparison of noise at ISO 1600.
    Precisely.

    What are the parameters for Neat Image? Are they the same for all (thus not optimized for the camera?) or are they specific to the camera (are you sure that this is the best?) Did you use the noise profiles secific to the camera?

    Lots of questions and variables, quite meaningless at the end of the day without the variables locked in.

    Note the color blotchiness of the 5D, 350D and D50 which is not seen on the D200.
    Last edited by Watcher; 30th May 2006 at 04:41 PM.

  7. #7

    Default Re: noise comparison ISO1600

    Quote Originally Posted by wind30
    I RAN ALL the images through neat image,

    ...

    The point I want to make is when you judge ISO performance, it is better to run the image file through neat image (or similar). Some makers choose not to do much noise reduction in camera to preserve the details and that doesn't mean the ISO performance is bad.
    Your reasoning is flawed.

    To judge a camera's ISO and noise performance, you need to take the raw image from the camera, and analyse it without and noise reduction programs, whether they are built in the camera or thru the computer. Any noise reduction program will degrade the picture quality. Hence, we want the lowest noise sensor possible to avoid image quality degradtion.
    deadpoet
    my portfolio

  8. #8
    Senior Member
    Join Date
    Jan 2002
    Location
    Singapore
    Posts
    12,938

    Default Re: noise comparison ISO1600

    Agree with Deadpoet and Witness.

    Also, downsampling will introduce additional processing which lead to less comparable results.

  9. #9
    Senior Member jOhO's Avatar
    Join Date
    Apr 2003
    Location
    Singapore
    Posts
    6,485

    Default Re: noise comparison ISO1600

    if u're comparing detail.. pls also choose a subject with more detail, in all regions, high mid lows...

    A for effort but probably Z for result.

  10. #10

    Default Re: noise comparison ISO1600

    ya lar. I did not do the test. The best test is like what some of you say, RAW developed using ACR with zero noise reduction. Then people will question the RAW convertor used as ACR is not optimized for all raw formats. But if you use the camera maker's raw software, you cannot turn off NR for some of these software.

    My point is to show that people sometimes conclude certain camera has lower noise due to out of camera Jpeg has lower visible noise, like D50 but failed to notice a huge drop in details. I cropped the region with fine detail (not many in the test pic) to show this.

    What I am saying is that with a good NR software like Neat image, it is better to get those higher noise cameras as you can retain the detail in the jpegs.

    As to which camera is better, you cannot tell by one test.

    The neat image is run using Auto profile, default setting for all. This is actually the recommended/best setting for neat image. As you can see a simple default run can remove a LOT of noise, much better than what the camera can do. So actually cameras like D50 is actually quite bad as it removes details which is gone for ever. Same for the s3pro. The jpeg developed for s3pro using ACR is REALLY noisy and shows how much NR s3pro does for high ISO pictures.
    Last edited by wind30; 30th May 2006 at 09:24 PM.

  11. #11
    Member
    Join Date
    Oct 2005
    Location
    Bt. Timah
    Posts
    1,354

    Default Re: noise comparison ISO1600

    I suppose it can be regarded as what you can achieve with the images produced by each camera at ISO 1600.

    May not be totally fair or objective, but it's a different perspective to the usual direct 100% crop comparison which every other site has.

    Also, images will invariably be downsized (unless you always print big) & manipulated, so if you regard the test in isolation, it does show how the image stands up to a series of edits to achieve the same final product, which is really what we all do.

    You can consider a no-holds-barred give me your best shot test.

  12. #12
    Senior Member
    Join Date
    Feb 2003
    Location
    The heart of the Abyss
    Posts
    2,307

    Default Re: noise comparison ISO1600

    There many points in this post and I'll try to give my view.
    Quote Originally Posted by wind30
    ya lar. I did not do the test. The best test is like what some of you say, RAW developed using ACR with zero noise reduction. Then people will question the RAW convertor used as ACR is not optimized for all raw formats. But if you use the camera maker's raw software, you cannot turn off NR for some of these software.
    That said, ACR has a different quality output for different cameras. Are you testing the LCD (Lowest common denomination, ie ACR) or the best quality output? I understand what you mean but that is the issue here. No major review only put one part of their review using RAW. If you see Phil Askey, Bjorn Roslett, or even Michael Reichmann, they all are very careful in describing their RAW conversion.

    I would have preferred the best quality rather than LCD.

    Quote Originally Posted by wind30
    My point is to show that people sometimes conclude certain camera has lower noise due to out of camera Jpeg has lower visible noise, like D50 but failed to notice a huge drop in details. I cropped the region with fine detail (not many in the test pic) to show this.
    That may be true. However, you forget that the target audience is different for each price range of the camera. The price and performance is different. You compared the output, performand and target audience of 5D (about S$5k) to D50 (about S$1.2k)? It is like comparing a Toyota Starlet, BMW 5-series, Ferrari and a Hummer.
    Quote Originally Posted by wind30
    What I am saying is that with a good NR software like Neat image, it is better to get those higher noise cameras as you can retain the detail in the jpegs.

    As to which camera is better, you cannot tell by one test.
    That is correct. But when you read Phil Askey's review, you would never see him pit a 350D vs 1DMkIIN in terms of performance or image quality. It is simply an unfair comparison.

  13. #13
    Senior Member
    Join Date
    Jan 2002
    Location
    Singapore
    Posts
    12,938

    Default Re: noise comparison ISO1600

    Quote Originally Posted by wind30
    My point is to show that people sometimes conclude certain camera has lower noise due to out of camera Jpeg has lower visible noise, like D50 but failed to notice a huge drop in details. I cropped the region with fine detail (not many in the test pic) to show this.

    What I am saying is that with a good NR software like Neat image, it is better to get those higher noise cameras as you can retain the detail in the jpegs.
    Have seen the original images. Without Neat Image, the shadow noise in ISO1600 images by E500 is really quite bad (even running through Neat Image, the images were still noiser than some others straight from camera). If there were fine details in the shadow regions, they would been masked out by the noise.

    I'm not sure whether the details were really lost in the D50's pic or just a difference in focussing (or lens used). The top of the head region seemed to have more details than the pic from E500.
    Last edited by mpenza; 31st May 2006 at 11:19 AM.

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •