Open drphilmarshall opened 9 years ago
In detail, I see three things wrong with the image viewer set up in the Dashboard Tools:
I think these are hangovers from VICS82, and I suspect 2 and 3 are also affecting the quick dashboard too (and causing the "missed a sim" error reported in #293). Looks like directing your investigations along these lines might be a good idea, @chrissnyder - thanks! :-)
@chrissnyder did you have any luck with this, please do direct me to appropriate bits of code to check if that helps. I do think not having QD is a limitation in the classification, and missed by returning SW volunteers.
The code in question is in https://github.com/zooniverse/Lens-Zoo/blob/master/app/controllers/quick_dashboard.coffee. It's just about finding the proper settings for the parameters, which unfortunately I'm not very helpful with.
Thanks @chrissnyder, I did look at this script a week or two ago. I found the image scales are OK, but i guess there's some other settings that aren't right or that the headers are not being read in properly. I'll take another look to see if I can spot what's up. Thanks
I've compared the stage-two branch version of quick_dashboArd.coffee with the master to flag up differences - please let me know if any of these are suspicious
ma: source: 'http://spacewarps.org/subjects/raw/' st2: source: 'http://spacewarps.org.s3.amazonaws.com/subjects/raw/' does this contain VICS82 or CFHTLS files, I presume the latter otherwise we wouldn't see the right images so that should be OK.
scales are reversed in the master compared o stage-two but I think the master order is right
there is some browser specific text in the stage 2 version that Is just set to url = "javascripts/webfits-canvas.js" in master
stage-two gives width = dataunit.width height = dataunit.height but in the master it says {width, height} = dataunit are these equivalent?
slightly different versions of magnitude
do these make a difference Stage-two
parameters = @parameters[preset]
@wfits.setCalibrations(1, 1, 1)
master
# @wfits.setCalibrations(1, 1, 1)
@wfits.setCalibrations(@calibrations['g'],@calibrations['r'],@calibrations['i'])
I'm going to sleep now but suspect that the width o setCalibrations is doing somethign wrong
width = dataunit.width
height = dataunit.height
{width, height} = dataunit
These are equivalent, yep.
Stage-two
# Get a preset and apply to color composite
parameters = @parameters[preset]
@wfits.setCalibrations(1, 1, 1)
master
# CFHTLS:
# @wfits.setCalibrations(1, 1, 1)
@wfits.setCalibrations(@calibrations['g'],@calibrations['r'],@calibrations['i'])
These may or may not be equivalent, based upon the value of @calibrations
. I can try setting them to 1,1,1 to see if it makes any difference.
It's worth a try, but I guess this won't solve the non centred display but may explain alpha issue perhaps.not sure
it still have a feeling that somehow something is being read in with vics82 sizes, but can't spot where this could be other than the width/height which you said was ok
is there anything that might be in "javascripts/webfits-canvas.js"?, but I couldn't locate it to check
or is there some way vics82 data frame sizes, settings might be being read in instead of cfhtls?
Looks like it:
http://tools.zooniverse.org/#/dashboards/spacewarp/55ddf10a1766277dfb00130f
Zoom and pan are off, as are non-linearity and color balance. Files are correct though! Another hangover from the branch merge.