Downsampling to reduce noise, but by how much?

Comments Off

Posted on 26th November 2008 by Phil Askey in Camera reviews |Software |Syndicated Press

There has been much contention recently that we're being unfair on high megapixel cameras which show higher levels of noise than lower megapixel cameras. The almost universal argument is that "you can downsample the high megapixel image to reduce noise". This statement is often made with no evidence or example, this article is intended to provide some examples so that you can make your own mind up.

Firstly lets be clear here (and not blind everyone with science), downsampling four pixels into one averages noise - that makes sense without understanding the maths. But downsampling four pixels means you're halving resolution, effectively turning your twenty megapixel camera into a five megapixel (or your G10 into a 3.7 megapixel), ignoring the improvement in sharpness you should see.

In order to provide some samples I took our standard noise test shot in both JPEG and RAW from a Canon PowerShot G10 at ISO 800 (a sensitivity which is a stretch for almost all compact cameras). I downsampled these 14.6 MP images to three specifically chosen resolutions:

  • 10.0 MP (3648 x 2736)
  • 6.5 MP (2944 x 2208) *
  • 3.7 MP (2208 x 1656) **

* 1.5 x 1.5 input pixels for one output pixel
** 2.0 x 2.0 input pixels for one output pixel

To be sure I've got a fair cross-section of what the average user (not the average fanboy) would do in downsampling I chose five different methods of downsampling:

  • JPEG: Photoshop Bicubic
  • JPEG: Photoshop Bicubic Sharper
  • JPEG: Photoshop Bilinear
  • JPEG: Canon Digital Photo Professional
  • RAW: Canon Digital Photo Professional

So lets see what effect this has on noise levels (standard deviation of middle gray):

G10downnoisegraph_2

Lets start with the JPEG image (red / orange / green lines), all methods except Bicubic Sharper (no surprise) result in a reduction in noise but it's hardly significant. Indeed even at 3.7 MP (a 4 pixel into 1 reduction) we're seeing very little reduction in measured noise. In RAW things are a bit stranger, using default noise reduction settings DPP delivers a noisier image at full resolution which suddenly dives to 'JPEG levels' at ten megapixels and then tracks as we'd expect it to 3.7 MP.

If we take the most commonly used downsampling method (Photoshop Bicubic) we get a 4% reduction in standard deviation at 10.0 MP, at 10% reduction at 6.5 MP and a 20% reduction at 3.7 MP. Twenty percent is a nice number and it sounds good except don't forget you've now got an image which is a quarter of its original size.

Enough graphs and figures, lets have a look at the images, as ultimately that's the most important thing (although some may classify debating as being more important). Below are (a) the most common (Photoshop Bicubic), the best performing (DPP JPEG) and RAW (DPP RAW)

Photoshop Bicubic downsampling

(left to right: 14.6 MP original, 10.0 MP downsampled, 6.5 MP downsampled, 3.7 MP downsampled)

Canon DPP downsampling

(left to right: 14.6 MP original, 10.0 MP downsampled, 6.5 MP downsampled, 3.7 MP downsampled)

Canon DPP RAW downsampling

(left to right: 14.6 MP original, 10.0 MP downsampled, 6.5 MP downsampled, 3.7 MP downsampled)

As with all of our reviews / articles we provide you with the samples and will let you draw your own conclusions, mine is that you have to downsample a long way (like 4 into 1) before you get any really noticeable gain and even then noise is still visible and you've got a much smaller image. At the end of it all downsampling is no substitute for larger sensors or larger photosites.

Why theory is great but grain size isn't

One of the reasons that theories about downsampling reducing noise don’t appear to work in practice is that the theory assumes noise is random. Unfortunately, this isn’t necessarily true. Noise at a single photosite will effect adjacent pixels as part of the demosaicing process. So noise doesn’t occur as individual pixels but as grain. The mathematical theory may tell you that downsampling works but it won’t if your noise grains are any larger than one pixel (and they nearly always are from a camera with a bayer color filter array).

In the example below we have three noisy images: one with a grain size of 1 pixel (commonly used for non-real-world demonstrations), one with a grain size of 1.5 pixels and one with a grain size of 2 pixels. If we downsample each of these by 50% (using Photoshop Bicubic) we see noise drop substantially for the 1 pixel grain image but much less so for the 1.5 and 2.0 grain size images.

Original images (crops)


(1 pixel; 11.4 std dev, 1.5 pixels: 11.1 std dev, 2.0 pixels: 11.1 std dev)

After 50% Photoshop Bicubic downsampling (crops)


(1 pixel; 4.9 std dev, 1.5 pixels: 8.0 std dev, 2.0 pixels: 9.9 std dev)

Is this thing on?

Comments Off

Posted on 6th November 2008 by Phil Askey in Syndicated Press

From the bet-you-didnt-see-that-coming dept., your eyes don't deceive you, we've really started a blog. To be more accurate, two blogs, one for our editorial team and one for our developers. Over the last few years we've come under criticism for not being more open with our readership, there have been many reasons for this, not least of which a simple lack of resource.

The good news is that we now have a team, and a good one at that, six on the editorial team, two developers and often floating (bouncing?) between the two, my good self. So here's a channel for us for be more open with you, to try to keep you up to date with what's going on behind the scenes and as a place for us to answer common or topical questions in a more permanent / easily referable manner.

So dig in and enjoy, we've primed the blog(s) with some news and I'll be cracking the whip to ensure we have a steady flow of entries here.

Compact Camera Reviews – they’re coming (in a way)

Comments Off

Posted on 5th November 2008 by Simon Joinson in Camera reviews |Reviews |Syndicated Press

Around a year ago we had our new reviewers trained to the point where they were ready to produce their own reviews. Before we even started recruiting we knew we'd start them off with compact cameras - partly because a compact camera review is considerably less involved than an SLR, partly because there are so many compact cameras on the market. Perhaps predictably every compact review we released was met with a torrent of complaints suggesting that we shouldn't be wasting time on 'point and shoots' when there were so many SLRs in the queue.

By the time spring arrived two things were obvious; that we were going to struggle to keep up with the flood of new SLRs, and that the compact camera market had reached something of a plateau, with 'new' models being nothing more than 'old' models with marginally bigger screens or vital new features such as smile detection. So we took the decision (and, I should add, not lightly) to concentrate our limited resources on covering all the SLRs in depth before we went back to compacts.

To be honest it wasn't that difficult a decision; we get at least 20x more people reading an SLR review than a compact review (with most compacts sold being in the sub-$150 bracket I think it's safe to assume that the amount of pre-purchase research buyers are doing isn't considerable).

Fast-forward to today and we've finally scaled the mountain of new SLRs and are, save for the couple of models currently mid review (the Panasonic G1, for example), up to date. So what to do about the scores of compact cameras we turned a blind eye to over the last year or so? Compacts still make up the majority of camera sales and we certainly haven't abandoned them (we've already posted our full Panasonic LX3 review and have several others, including the Canon G10, in progress), but the sheer number released each year makes giving them a full review impossible (and, at the entry level, the cameras simply don't warrant a 10 or 12 page review).

We therefore decided to try an idea Phil and I have been toying with for several years now; a roundup/group test that will, we hope, give a snapshot of the compact camera market as we enter the peak holiday buying period. So we spent a couple of days researching and shortlisting, and ended up with a list of 35 or so cameras to include. We've now got them all in the office and I've spent a week in the studio with them doing the first round of tests. We hope, within the next four weeks or so, to publish all five of the group tests (they're split into budget, 'style', 'advanced', 'high end' and 'SLR-style').

To give you a taste of what to expect here's the list of cameras we've picked; don't expect a full review of every one of them (though there are some that will eventually make it to full review state), and don't be too surprised if a few of them change if we can't get hold of them....

  • Samsung L210
  • Canon A470
  • Kodak C1013
  • Sony W120
  • Nikon CP L18
  • Fuji FinePix Z20fd
  • Canon A590IS
  • Panasonic LZ8
  • Olympus FE360
  • Canon PowerShot SD790IS / IXUS 90
  • Casio EX-S10
  • Fujifilm Z200fd
  • Nikon CoolPix L210
  • Nikon Coolpix S60
  • Olympus Stylus 1040
  • Panasonic FX37
  • Pentax Optio S12
  • Sony DSC-T700
  • Canon SD880 /IXUS 870
  • Fuji  F100fd
  • Nikon Coolpix S710
  • Pansonic FX150
  • Samsung NV1000HD
  • Sony W300
  • Canon G10
  • Kodak Z8612 IS
  • Nikon P6000
  • Panasonic LX3
  • Ricoh R10
  • Pansonic TZ5
  • Canon SX110 IS
  • Sony H10
  • Canon SX10 IS
  • Fujifilm S2000HD
  • Fuji S8100FD
  • Kodak Z1015 IS
  • Nikon P80
  • Olympus SP565
  • Panasonic FZ28
  • Sony H50

Painting pictures (in 1000 words or fewer)

Comments Off

Posted on 5th November 2008 by Richard Butler in Syndicated Press

We’ve been, from time-to-time, accused of using overly flowery language in the reviews on this site. We try to tread that fine line between trying not to say exactly the same thing time-after-time and lapsing into the incomprehensible. And, on the whole, I think we get it about right. But, as you'd expect, there is the occasional dissenting voice on the forum demanding that we must just state the facts as dryly as possible (But then the forums would be a much less 'vibrant' place if everyone agreed on everything).

However, our linguistic forays - even at their most florid - are nothing compared to the adventures in prose undertaken on behalf of some of the camera companies. Perhaps it’s because much of the underlying material has been translated from unrelated languages – a process that can result in some beautiful, if somewhat enigmatic, imagery. Or maybe because there are few jobs more thankless than having to compose three pages-worth of compelling text about a camera that differs from its predecessor only in the inclusion of a larger screen, a more densely-packed sensor and a smattering of new scene modes (surely soon to include ‘Friends eating Lancashire Hotpot on a Wednesday evening in summer' mode).

Statements such as: “The subject floats serenely in its own world, like a misty memory or the landscape of a dream” are not unheard of, making me slightly concerned that the next time I order prints I’m going to be confronted with the question: ‘What finish would you like? Gloss, Matte or Whimsy?’

My favourite of late, however, must be the press release that suggesting that the camera it was pushing included a mode that can create: “a sense of déjà vu.” Which sounds lovely, unless you make the mistake of thinking about it too hard. I suspect that, if you take photos for yourself, then you’re probably doing so to preserve memories, in which case the sensation of having experienced the scene before should really come as standard; and those who’re taking photos to show or sell to other people probably aren’t hoping for the response: “I’ve already seen this.”

Adobe Camera RAW in our reviews

Comments Off

Posted on 4th November 2008 by Lars Rehm in Camera reviews |Reviews |Software |Syndicated Press

We have a brand new blog which desperately needs some content before it can be launched and following our recent Canon EOS 50D review there has been quite a lot of discussion in the forums about the use of Adobe ACR as a RAW converter in our reviews. These, I thought, were two good reasons to get into blogging mode and write a brief article that explains why Dpreview is using ACR and not proprietary or other third party converters for its tests.

So why is it then that we use Adobe Camera RAW and not one of the other converters?

Firstly we aim to create a level playing field and to make results - as much as possible - comparable between cameras. For this reason we simply have to use a third party converter that can handle most of the many different RAW formats that are out there rather than a proprietary converter. ACR is used as a plug-in in Adobe Photoshop, Photoshop Elements and Lightroom and has the highest market share of all third party converters which makes it the obvious choice. More importantly though Adobe updates ACR very frequently (and we're involved in early releases) to include the latest camera models. This allows us to review a new model fairly promptly after its launch (we often get the cameras before anyone in the RAW development arena, including Adobe).

We know that Adobe Camera RAW applies different levels of sharpness to different cameras and because of that we use the following workflow when processing our box shots:

Load RAW file into Adobe Camera RAW (Auto mode disabled)

  • Set Sharpness to 0 (all other settings default)
  • Open file to Photoshop
  • Apply a Unsharp mask: 80%, Radius 1.0, Threshold 0
  • Save as a TIFF (for cropping) and as a JPEG quality 11 for download

Even by using this workflow we cannot entirely avoid any processing differences between cameras but it's certainly as good as it gets in terms of creating a 'level playing field'.

Secondly, especially when analyzing image noise, we want to show a camera's image output in its 'purest' form, i.e. as captured by the sensor and as little as possible altered by sharpening algorithms, noise reduction or any other forms of image processing. Again, ACR does a pretty good job in this area. To demonstrate this I have included example crops from two images which both have been processed from an ISO 3200 Canon EOS 50D RAW file, one in ACR 4.6, the other in Canon's Digital Photo Pro 3.5. Sharpening and noise reduction were set to zero in both converters, then we applied an identical unsharp mask to both images, so in theory we would expect pretty similar results.

Dpp_acr_comp_4

Reality looks slightly different though. The ACR image is less sharpened and shows visibly larger amounts of chroma noise. That tells us two things. Firstly DPP is doing quite a good job at noise reduction and is probably the better choice for cleaning high ISO pictures of the 50D (ACR's NR can't match DPP's even if you turn it up). However, it also means that DPP applies at least some chroma noise reduction (and sharpening) even when NR is set to zero which renders the software pretty much less useless for our purposes.

Nevertheless, if you're interested in results that can be achieved with different RAW converters, you can have a look at the 'Software' page in our DSLR reviews where we usually post images of our resolution chart and the studio scene which have been processed in different RAW converters (if available at the time of the review).

Upcoming lens reviews

Comments Off

Posted on 3rd November 2008 by Andy Westlake in Lens reviews |Reviews |Syndicated Press

The level of interest in our Canon 18-200mm review has persuaded us that we should test a few more superzooms in comparison  (the great strength of the widget is that we don't need to have the lenses in-house simultaneously for side-by-side shooting). So expect to see some leading 18-2x0 mm zooms interspersed with our still-ongoing series of 50mm primes. Lenses on the shortlist for reviews in the near future include the following:

50mm Lenses:

  • Pentax 50mm F1.4 (on K20D)
  • Sony/Minolta 50mm F1.4 (on FF and APS-C)
  • Canon 50mm F1.8 II
  • Nikon 50mm F1.8

Superzooms:

  • Sigma 18-200mm OS
  • Tamron 18-270mm VC
  • Sony/Pentax/Tamron 18-250mm

(The Panasonic/Leica 14-150mm would be an obvious comparison for Four Thirds users, but its limited availability and high price will necessarily affect the final decision on whether we review it.)

At some point we still hope to test the Sony 70-200mm F2.8G on both FF and APS-C, to see if it can justify its stratospheric price-tag in comparison to the competition from Tamron and Sigma.

African Photography Blog – South Africa Photographic News, Articles, Tutorials is Digg proof thanks to caching by WP Super Cache