Personal View site logo
Any opinions about DXOMark?
  • Generally, considering this:


    I wanted to make topic about DXOmark instead.

    Myself, I do not like their tests. Something is wrong with them. Under "wrong" here I mean general concept, not math error.

  • 15 Replies sorted by
  • im not a fanboy of any camera company out there, im a fan of any camera that does the job you want it to do. im sure canon fans hate dxomark, but i really feel canon never gets a fair rating, i dont see how the 7d/60d/t2i/t3i can all score so low, in my opinion they all take beautiful stills.

  • I find them to be one helpful part of researching a camera. Although, I tend to look at their individual scores and give no consideration to their overall one.

    The thing to keep in mind is, they are pretty much only useful for stills, not video. If you read them, you would think the GH1 slightly outperformed the GH2 in lowlight and that has been handily proven to be untrue for video.

    Anyway, the overall scores give more weight to dynamic range, ISO, than burst speed etc. so it is natural that for the last 3 years Canon has often scored lower. But it remains to seen if the trend will continue this year, since the newest releases show very different strategies for each company than last year.

  • @thepalalias I go about it the exact same way, I look at individual scores not the overalls. And as you said, it's all about still image quality at a given ISO, it doesn't not directly relate to video as downsampling/binning/processing implementation plays a big part in that.

    The D800 seems incredible though, I have a D700 and D3S, and I think I may have to pick one up :)

  • @joesiv It seems like the jury is really still out on D800 vs D4, though. :)

    I am definitely invested in Canon lenses so I hope the EOS-1D X can compete with D4. The ISO equation was so lopsided in favor of Nikon for the last round and I am curious how this one will go.

    Either way, I really enjoy shooting with Canon glass so my Canon is safe. :)

  • Considering this .... we have just one option:

    Let's start collect money for a D800 body to be hacked!

    About the DxO, I think that not always the best math give the best pleasant image to our eyes. It's all about calibration. Optical Psychophysics, Gestalt, cultural influences, social induction ... all change our perception of what "image quality" is.

  • You can buy an AKG mic capsule with a ruler flat frequency response, and I have a whole drawer full of them. But they aren't my favorite mics. If you shoot some video, presumably your eye will tell you what you need to know. In this case, a lot of people are very happy with their Nikon. But a lot of people are happy with other cameras as well. Some of my favorite photos were taken with a twin lens Rollei. And no lack of resolution. I have to figure how to get the taking lens on my GH2 one of these days--or maybe just buy a roll of 120.

  • Yeah interesting stuff about sensors. But most of the article is related to measuring for print 'resolution normal'. When it comes down to 72dpi 1920 x 1080 HD video recording for example , I wonder if the relevancies shown are much more difficult to apply with all the downsizing.

  • @driftwood

    Also interesting is that now only tiny number of photos is printed.

  • Btw is site mentioned in article with electron read noise, QE and capacity numbers calculated from DXOMark data.

  • Random Noise doesnt bother me so much. Maybe gigantic 4k LCD displays will make a difference, but even poor noise performers look great with a bit of NR and a good down sample to 1080p. Even prints need to get up to A4> to really notice the difference.

    FPN is a nightmare though, and it would be good for them to test for that.

    Dxo dynamic range figures are really quite useful though.

  • FPN is a nightmare though, and it would be good for them to test for that.

    It is not easy thing to test. In fact all such tests must mention much more, like sensor temperatures, etc.

    Btw, DR measured is for 8Mp images, not for 1080p. So, potentially, maximum possible DR will be more fo video (in ideal situation).

  • They still haven't reviewed the 20mm f1.7.

  • OK I have read that luminous-landscape essay. I really love the graph porn. In these graphs you can see why panansonic would have to use a sony sensor for gh3 (if it is). I agree that the digital path to video resolution requires a different testing procedure and this will be even more true as the processing power of cameras increases.

    Witness an announcement by Nvidia at CES on a new TEGRA4 platform - they highlight realtime HDR as a significant use case for this platform.

    The Tegra 4 also introduces new computational photography architecture, which adds a new engine to drive the image processing and significantly improve the amount of time it takes to calculate the necessary mathematics 10 times faster than current platforms. To show off its power, NVIDIA demonstrated HDR rendering on live video. The chip is also capable of implementing HDR in burst shots and with LED flash. The idea, NVIDIA says, is to eventually make our mobile cameras more powerful than DSLRs

    Perhaps Tegra4 will need its own topic if a video demo shows up.