Sunspotter Talk

Fuzzy images

  • ElisabethB by ElisabethB

    Hey everyone,

    First of , it's really great to have Sunspotters up and running again !

    But what is with all the fuzzy images ? Almost half of the images are very fuzzy. Now, this doesn't really interfere with determining which image is the more complex one. But, unconsciously, I think people are more prone to dismissing the fuzzy image (because it is fuzzy, ie not right). What is the reason behind these images ?

    Thanks for any clarification ! 😄

    Posted

  • diablosbud by diablosbud

    The images appearing on the mobile site with my iPod Touch 5G are also blurry/fuzzy, and none of my results are going through (I'm not increasing in number of passes when I'm certain I'm correct). For reference, this is on Safari in iOS 7.1.1.

    Posted

  • hamptongray23 by hamptongray23

    I believe that the images are just pixelated.

    The Sunspotter scientists are just interested in how complex the sunspots are not how large they are, so smaller sunspots are probably included along with the huge ones. I would imagine that you would have to zoom in for the smaller sunspots (although likely still quite large in comparison to earthy objects) to appear big enough for citizen scientists to rate them well.

    That is my best guess.

    Posted

  • dvdgc13 by dvdgc13 translator, scientist

    Hello all, @hamptongray23 is in the right direction. We are trying to get rid of any possible bias on the classification, so far we are taking care of two that makes the images a bit fuzzy. One is size, as @hamptongray23 said, we are interested in the complexity and we try to avoid people choosing the larger just because is larger. Of course, as @ElisabethB has said this could make the people to choose the less fuzzy one... that's something we are aware and we will take care when interpreting the results. The other change that may make the images a bit fuzzy is that we have de-projected the images so all of them seems as observed from the same point of view.

    @pahiggins has written a blog post for this new dataset that explains with detail these changes. Feel free to ask questions about all that if you are interested 😃

    Posted

  • pahiggins by pahiggins

    Hi all,
    Just to reiterate what everyone pretty much said already:
    So, last time we cropped the images so that all of the sunspot groups would be at the same scale (and have the same resolution). We thought that people might be biased toward choosing the bigger sunspot group as the more complex one, even if it wasn't really. This time we wanted to try and fix that bias by making the sunspot groups the same size on your screen (even if they are not the same size physically). So the new problem if that one is fuzzy and the other isn't!

    I considered using an interpolation technique to make the images less pixelated- but then they would just look smoother, and still fuzzy (maybe that would have been less distracting).

    Do you think that having the images blown-up (or shrunken) to the same size makes it easier or harder to discern complexity?

    Thanks for your comments!

    Posted

  • cherrycoke by cherrycoke in response to pahiggins's comment.

    Hello pahiggins,
    The pixilation does through me off when I am choosing, however I am try not to let it. I think (for me at least) the them being fuzzy is more distracting then if they were not the same size since you are asking for complexity and not size. Complexity, involves details which are harder to see in the fuzzy images. Therefore I think having them the exact same size is not as big of a priority as having them clear so as to better see the details and discern the complexity.

    Posted

  • ElisabethB by ElisabethB

    For me it is better to have them the same size. But I have to consciously look at the fuzzy images and not dismiss them because they are fuzzy. And squinting helps ! 😄 It really took some time getting used to the fact that the fuzzy images were 'good' images.

    Posted

  • pahiggins by pahiggins

    Thanks for the comments. I think next time (when we use SDO/HMI data) I will try interpolating, so at least they will be smooth...

    Posted

  • shocko61 by shocko61

    Hi ElisabethB , shocko61 ,Ian here for the first time the above discussion answered my question thank you , it is all very interesting . I have been looking at space weather for a very long time . Cheers , Ian

    Posted

  • ElisabethB by ElisabethB

    Hi Ian ! Sorry for the late reply ! Happy to be of help !

    I, personally, find Sunspotter to be a very relaxing Zoo-project ! 😄

    Posted

  • pahiggins by pahiggins

    It's nice to hear that you find it to be relaxing! As a scientist behind the scenes, I find it stressful to do classifications using the interface... I am so used to automating data analysis, that doing things by hand makes me nervous 😃

    Posted

  • ElisabethB by ElisabethB

    it could just be me, of course ! 😄

    Posted

  • misibacsi by misibacsi

    Fuzzyness is quite confusing. If one image is fuzzy and the other is a sharp image I tend to choose the sharper image as more complex, so there is a bias in choosing because of fuzzyness. Why don't you leave them out? Anyway how do these images become fuzzy? I thought all images would be crystal clear.

    Posted

  • ElisabethB by ElisabethB in response to misibacsi's comment.

    Hi misibacsi

    Did you read the explanations on the previous page ? .

    Posted