1. Welcome to Mu-43.com—a friendly Micro 4/3 camera & photography discussion forum!

    If you are thinking of buying a camera or need help with your photos, you will find our forum members full of advice! Click here to join for free!

Is M43 really so bad at high ISO? Comparison to A7Sii and X-T1

Discussion in 'Back Room' started by WGPhotography, Nov 22, 2016.

  1. WGPhotography

    WGPhotography Mu-43 Regular

    185
    Oct 21, 2016
    Wayne
    So the other night I had a sports event to shoot (see my boxing thread in the Panasonic forum for samples here) and I ordered a couple of other cameras to compare against my (beloved) GX8.

    I was concerned about the noise performance as I knew I'd be around ISO640 and above.

    Unfortunately, the A7Sii didn't arrive on time, but I did manage to try out the X-T1 and I noticed slightly better ISO performance, but skin tones that were much more yellow and less appealing compared to the Panasonic. This was inadvertently backed up by the findings of the Camera Store TV test on YouTube last week.

    Anyway, today I managed to compare all three cameras, but not in the regular way I see most comparisons. For me, the test for high ISO noise is not from in-camera JPEGs, or OOC RAW, but edited RAW files. How far can I push the file before it looks too bad etc.

    So on my way home, I stopped the car and shot all three cameras on a tripod, doing my best to keep the same FoV and settings. The exposures between the camera were oddly quite different, so I decided to use the histogram of each camera instead of keeping the settings the same. If this adversely affects the images then please tell me; I suspect it may do, but some images with the same settings on one camera would look COMPLETELY different on another camera. Also, one camera, I forget which had a shutter speed of 1/320 whereas another would want 1/350 etc. So I thought keeping the histogram looking the same would be a better bet... I was freezing as well and wanted to get back in the car lol.

    I shot RAW, and the in-camera settings were all totally dropped. So in terms of noise reduction, sharpness, colour saturation etc, all settings were lowered in-camera as much as possible.

    All cameras are running the latest firmware on both their bodies and lenses, and I picked the same focus point for them all, which was a small patch of snow around a third of the way into the image, towards the left third of the frame. Weirdly, the Fuji image seems soft throughout, so I don't know if it missed focus, but if it did, it did so on the several images I shot with the camera. I focused on the same point each time with the AF. Maybe I have a soft sample of the lens?

    Speaking of lenses, I wasn't too bothered about sharpness, but I wanted to keep some kind of uniformity, so I chose the 12-35mm Panasonic, 18-55mm Fuji and 12-240mm Sony zooms. I set them all to f5.6 and I zoomed in to approximately 27mm FF equivalent, which was the minimum the Fuji would go to, and therefore it dictated the length a little. In hindsight, this could be the cause of the Fuji softness, so perhaps I should have shot them all around 35mm or so?

    I imported the RAW files into Ligfhtroom CC (latest update) and did nothing but raise the shadows up to 100% and the highlights down to 100%. Dynamic Range wasn't on my mind hugely as it's been a dull day and there wasn't a lot of range in brightness anyway. Saying that, I've never struggled with the supposed limited DR on m43, especially with the GX8.

    I've attached the exported JPEGs below. These were all exported at full size with no adjustments made, so let me know if I need to adjust the size to match them all to the 12MP of the A7Sii here. They are at 300dpi.

    If this test is useless, or I should have done something else, please let me know, but hopefully they offer some kind of information to people like myself who generally PP their images.

    The images are:

    652/654/656 - Panasonic GX8 - ISO800/3200/6400
    955/957/958 - Sony A7Sii - ISO800/3200/6400
    3155/3157/3158 - Fuji X-T1 - ISO800/3200/6400

    I don't know if the site will compress the attachments. If it does, please suggest a way to upload the full size files and/or RAWs elsewhere.

    If there's anything I've over-looked, please also suggest things, but I have to say, up to 800, m43 performed better than I imagined. It appears there';s a stop or so benefit to APS-C over m43, and a further stop or so from APS-C to FF. Above 6400, the A7Sii clearly takes the lead, but it's very rare I would shoot at such an ISO personally.

    I did take a shot with the Sony at ISO32000. It's horrible of course but it's much cleaner than expected. Still a bit pointless though.

    As an aside, the GX8 is still the best camera I've ever owned for handling and functionality. I've had many cameras, but from the tilting EVF to the menu system, it's pretty flawless. It's my clear favourite in terms of usage.
     

    Attached Files:

    Last edited: Nov 22, 2016
    • Appreciate Appreciate x 1
  2. barry13

    barry13 Mu-43.com Editor Subscribing Member

    Mar 7, 2014
    Southern California
    Barry
    Hi, the forum will resize larger images down to 1600px.

    If you link to Flickr, the full size can be available there.

    You could also use Google Drive or Dropbox or similar.

    Or just post 100% crops smaller than 1600px.

    Barry
     
  3. kimballistic

    kimballistic New to Mu-43

    2
    Nov 12, 2016
    Brian Kimball
    I find that ISO comparisons at normal light levels are pretty much useless. They all will look good.

    Since the goal is to bring out the noise, personally I would redo the test in a very dark setting, which is when most people will be using high ISO anyway.
     
    • Agree Agree x 3
  4. Clint

    Clint Mu-43 Hall of Famer Subscribing Member

    Apr 22, 2013
    San Diego area, CA
    Clint
    For me the usability of a camera can trump technical aspects of others. What use is a camera if you don't like using it?
     
    • Agree Agree x 6
  5. jyc860923

    jyc860923 Mu-43 Hall of Famer

    Feb 28, 2012
    Shenyang, China
    贾一川
    What I'd like to see m43 to improve, is the IQ after push/pull manipulations, the raw file just doesn't contain enough info. Since I like to give the images a dynamic look yet being realistic, I can't remember when I ever had a clean image with m43.
     
  6. kimballistic

    kimballistic New to Mu-43

    2
    Nov 12, 2016
    Brian Kimball
    I'm considering investing in m43 (currently a Fuji X-T1 owner, and really wanting the Sync IS video capabilities of the EM1.2 + 12-100).

    IQ is a hot topic I've found. Would it be too much to ask if you could post an example or two?

    I've been pixel peeping full resolution JPGs and the occasional RAW for weeks and I'm seeing results all over the map-- amazing noise free shots and horribly noisy shots.
     
  7. jyc860923

    jyc860923 Mu-43 Hall of Famer

    Feb 28, 2012
    Shenyang, China
    贾一川
    Here I'll pick a few full-res shots that you can download from my flickr, TBH I can make the images appear noise free but NR isn't something I'd like to do because of the loss of detail. And there are very possibly issues with my pp technique so please just take it with a grain of salt.

    https://c2.staticflickr.com/6/5641/30804685530_a2041ed84c_o.jpg
    https://c2.staticflickr.com/6/5616/31019478081_6bd6fd8402_o.jpg
    https://c2.staticflickr.com/6/5589/30554708260_bf8f3dd6b7_o.jpg
    https://farm6.staticflickr.com/5689/30990290392_52fe727cf4_o_d.jpg
    https://c2.staticflickr.com/6/5620/30097569276_4dbae395c0_o.jpg
    https://c2.staticflickr.com/8/7554/29462492693_5180158ca9_o.jpg
    https://c1.staticflickr.com/9/8530/29095913863_e59ab3b251_o.jpg
     
    Last edited: Nov 24, 2016
    • Informative Informative x 1
  8. DanS

    DanS Mu-43 All-Pro

    Mar 8, 2016
    Central IL
    A lot depends on the ISO, apeture and quality of the glass.

    This is one I took a little over a month ago in DC.
    • Panasonic G7
    • Olympus 60mm
    • f/11
    • 1/640
    • ISO 400
    • Raw with a little NR, sharpening.
    https://www.dans-hobbies.com/wp-content/uploads/2016/10/P1050118.jpg
     
    • Like Like x 2
  9. kimballistic

    kimballistic New to Mu-43

    2
    Nov 12, 2016
    Brian Kimball
    Thank you! Those are all shots I would be very happy to have in my personal collection. I see the noise and I don't mind it at all.

    The landscape-oriented cityscape is the only one with borderline-unacceptable levels of noise, and yet I could only see the noise when I viewed it at 100%.

    I'm going to stop worrying about IQ. Thanks for the examples!
     
    • Like Like x 2
  10. kimballistic

    kimballistic New to Mu-43

    2
    Nov 12, 2016
    Brian Kimball
    Fantastic. Super sharp, no noticeable noise. Thanks for sharing!
     
  11. jyc860923

    jyc860923 Mu-43 Hall of Famer

    Feb 28, 2012
    Shenyang, China
    贾一川
    Thanks you. I find that halo, or grey-ish borderline annoying too
     
  12. Ramsey

    Ramsey Mu-43 Top Veteran

    745
    Jan 9, 2013
    Zagreb, Croatia
    Honestly, i saw a huge difference between my E-PM1, E-PL5 and E-M10 ii. Maximum of 1200, 1600 and 2500 were used, respectively. I feel like i get cleaner files with E-M10 ii on 2500 then i did with E-PM1 on 1200.

    That said, base ISO is still noisy, and after push-pull, even with good light, NR is usually required.
     
    • Like Like x 1
    • Agree Agree x 1
  13. It can depend on your colour and what you had to protect in the initial exposure. Even working at ISO200 with perfectly controlled lighting (two controllable TTL flashes, polariser, multiple positionable LEDs) you can still end up with stuff that is going to just be super rough without some specific noise control above just leaving the image alone. In this case a butterfly with pure white scales - they have structural polarised colour so there is really no getting around having radically different exposure on different parts of the scales no matter how you light it, at least for a live subject.

    Here's basically fine -

    14kxbd.jpg
    Subscribe to see EXIF info for this image (if available)


    and here on the background where I couldn't quite light it hard enough without blowing something else ... is gonna need a trip to PS when I'm done

    2o9aqx.jpg
    Subscribe to see EXIF info for this image (if available)


    as with any camera you're going to run into your worst cases where you have a scene with compromised or natural lighting, highlights to protect, shadows or dark areas that are gonna look wrong if they're totally crushed down blacks, and particularly dark blue or red colours. This one (whenever the image caching here is fixed :p ) -

    secret bee rituals

    was a total nightmare to do due to being at min shutter 1/200, having a minimum acceptable macro equivalent aperture of about f16-f22, and having both a dark subject and a constantly changing degree of blown out highlights. And that was days of work to get and to fix to that level, and if you look close, it can still be pretty rough, even with going in and manually re-noising bits with higher resolution noise.
     
    • Informative Informative x 1
  14. pdk42

    pdk42 One of the "Eh?" team Subscribing Member

    Jan 11, 2013
    Leamington Spa, UK
    Just to emphasise what many have said... high ISO comparisons only make sense if they are done in difficult lighting situations. Not only is this the obvious time when you'd need high ISO, but you'll likely be experiencing slow shutter speeds and poorer contrast. Both will apparently soften your image and accentuate noise. Then you'll likely have more challenging shadows where PP will be needed.

    u43 noise will be worse than APS-C which will be worse than FF which will be worse than MF. To find out what's your own personal limit you'll need to shoot some examples in sitations that matter to your style of shooting. If you're photographing music gigs, then a Sony A7s is probably where you need to be!
     
    • Agree Agree x 2
  15. 50orsohours

    50orsohours Mu-43 Hall of Famer

    Oct 13, 2013
    Portland Oregon
    Can you post some examples? I don't seem to run into noise issues.
     
  16. Klorenzo

    Klorenzo Mu-43 All-Pro

    Mar 10, 2014
    Lorenzo
    Agreed. And also on correct exposure (vs shadow recovery), subject tone (dark/bright after correct exposure), the kind of details of the subject (skin or clouds vs fur or branches) and denoise software and sharpening settings.

    And also how big is the subject in relation to the frame that often translate to how much you crop.

    And, on a different level, the kind of shots: night shots, street, concerts, rough sports, B&W vs wildlife, macro, etc.

    I also have the feeling that any kind of denoise improve the details in a comparable amount across different sensors, somehow leveling a little the playground. The smaller sensor benefits more from good denoise. In other words the difference in the unprocessed RAW seems bigger then what you see in the final images.

    I'm usually happy up to ISO3200 for most shots, 1600 is very good (unless I have to do much recovery...).
     
  17. Ramsey

    Ramsey Mu-43 Top Veteran

    745
    Jan 9, 2013
    Zagreb, Croatia
    I'll try to post an example after work
     
    • Like Like x 1
  18. Clint

    Clint Mu-43 Hall of Famer Subscribing Member

    Apr 22, 2013
    San Diego area, CA
    Clint
    How large are you viewing these images to see the noise?
    - If you are viewing the images larger than the size to be displayed on a monitor - Why? Who is going to be viewing them at that that size?
    - If you are viewing at 1:1 - that is like printing the longest edge near 50" (127cm) in length and then walking right up to photo to look at it. Who does that with prints?
    - Even with 60" monitors and the whole image being viewed, noise is typically a non-issue.
    - Have you made large prints? If not, you are probably in for a real surprise. Noise does not reproduce in printed photos the way it does on monitors! View the print at a size that it would typically be viewed at, I don't see people walking up to roadside billboards to see how clear an image is.

    At 30" (76cm) I did not see any noise issues with those photos that @jyc860923 posted.

    If you want noise free images, maybe you are using the wrong media and should be using a vector art program (Illustrator/CorelDraw) or Photoshop to draw your images so that you end clean. When illustrating, noise is typically added to provide a more photorealistic image!

    I just do not understand this mentality about making images that appear so sterile or clinical.
     
    Last edited: Nov 24, 2016
    • Agree Agree x 2
    • Like Like x 1
  19. WGPhotography

    WGPhotography Mu-43 Regular

    185
    Oct 21, 2016
    Wayne
    Thanks for the replies. I'll re-shoot asap in darker conditions
     
  20. Ramsey

    Ramsey Mu-43 Top Veteran

    745
    Jan 9, 2013
    Zagreb, Croatia
    i tried to find an example of base ISO noise, but i can't explain it better or find a better example of what @piggsy@piggsy already posted.

    in the end, i agree with @Clint@Clint , while i do think there is some noise, it's minimal. Also, pixel peeping is bad and we should all feel bad :)  And looking at my first post, i made a mistake in saying "noisy". What i wanted to say "certain/minimal level of noise at base ISO".

    All that said, it is possible that i'm mistaking certain things in the OOF area with noise.
     
  1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.