Stardust@home Suggestion Thread

Discuss your experiences with and ideas about Stardust@home here.

Moderators: Stardust@home Team, DustMods

Orion_0169
Posts: 153
Joined: Wed Aug 02, 2006 8:17 pm

Post by Orion_0169 »

A set of threads specific to movie numbers would cut down on new threads each and every time someone finds something they think is a track.

Check out... 10000-14999
Check out... 15000-19999

Or somesuch.

<shrugs>

I'd like to see a list somewhere of all the movies people have considered suspect. It must be massively large by now. Added bonus of this, would be, when you see a movie that you think other people might have, but you choose not, you could go see how others felt, without starting a thread to ask.

I'm sure soon enough someone will start a thread asking for #'s of events! ;-)
Je ne peux pas regarder la Mer sans me demander qui vit au-delà de cela.
cosmicrocker
Posts: 11
Joined: Sat Aug 05, 2006 8:10 am

Post by cosmicrocker »

That's a good idea. A little organization in the forum would improve it a great deal. Starting a new topic for each new question simply fragments communication.
...Tom

visit #space on IRC.freenode.net
Ziggy
Posts: 10
Joined: Wed Aug 02, 2006 4:39 pm

Improving the quality of results

Post by Ziggy »

I have a few suggestions to make about the VM, and more specifically the calibration movies, that may help to improve the assessment of the average user. (N.B. Apologies in advance if you have already considered such amendments to the VM.)

Additional Calibration Suggestions:

As far I can tell (after viewing over 400 calibration movies) they are only designed to check a user's ability to confirm, or deny, the presence of a valid track within a 'well-formed' movie. (i.e. One where the surface is easily located and the user has scope to focus below the surface level.)

I believe that if additional calibration movies were introduced to encourage a correct response for the more complex examples, that are seen within the genuine data set, then a higher accuracy of movie assessment would be attained.

Movies which don't appear to change throughout the focus range:
Movies where the surface focus level is too low to reveal whether a track exists.
http://stardustathome.ssl.berkeley.edu/ ... e_id=48802

(N.B. The surface anomolies at top of picture only in focus at lowest position... therefore, imposisble to ascertain whether a track exists under the surface at this position in the movie.)

Correct response: BAD FOCUS
Unable to determine the surface focus level
Movie seems to be an amalgamation of two movies - but picture only changes at the very bottom of the focus range.
http://stardustathome.ssl.berkeley.edu/ ... e_id=13723

Correct response: NO TRACK - as user to able to give an accurate appraisal of the image.
Movie seems to be an amalgamation of two movies - picture only changes halfway down the focus range.
http://stardustathome.ssl.berkeley.edu/ ... e_id=27903

Correct response: BAD FOCUS - user unable to ascertain whether a track does / doesn't exist in movie.
Ziggy
Posts: 10
Joined: Wed Aug 02, 2006 4:39 pm

Post by Ziggy »

I'm wondering: are any calibration movies Bad Focus? Because I had one where there was nothing ... slight shift in gray scale, no indication of Surface ... and I think I got scored wrong.
I have just posted a suggestion on the "Problems and Support" forum on this very topic because I don't think there are any 'BAD FOCUS' calibration movies... but perhaps there should be.

http://stardustathome.ssl.berkeley.edu/ ... =improving
mover
Posts: 14
Joined: Sat May 27, 2006 5:11 pm
Location: Malden, The Netherlands

Post by mover »

I think I've seen a "Bad Focus" calibration movie.
At least when I clicked the Bad Focus button my score went up by one...
"Bad Focus" real movies don't seem to be counted.
See the universe in a grain of dust...
icebike

Post by icebike »

Orion_0169 wrote:A set of threads specific to movie numbers would cut down on new threads each and every time someone finds something they think is a track.

Check out... 10000-14999
Check out... 15000-19999

Or somesuch.
Actually I'd like to see these come to an end.

There is really no need to post every suspected find, there 1.6 million movies, we don't need threads started for each one.

The interesting oddities are fine to post, or those where you have a serious question.

But just because you are excited because you THINK you found something doesn't mean we all need a new thread. Its not practical to do this, and it bogs down the forum with yet more trivia.

Just my grumpy 2 cents....
mwhiz
Posts: 95
Joined: Tue May 23, 2006 3:58 pm
Location: Seattle, Washington, USA

whole project stats??

Post by mwhiz »

just wondering if the stardust team is going to put up some stats for the whole project... like: number of tiles scanned, number of tiles quality checked, number of tracks found, % complete and all that jazz. would be nice to see how fast the project is progressing. the individual stats are quite good though.
~peter
"The Earth is the cradle of mankind, but one cannot live in the cradle forever."
~Konstantin Tsiolkovsky
DustBuster
DustMod
Posts: 694
Joined: Wed May 17, 2006 8:12 pm
Location: Horsetown, USA

Post by DustBuster »

I agree with the icebike, if every (currently +/-7000) user posted their finds, we would be drowning right now... but it seems a number of people just can't help themselves- so if Orion's idea can limit these to a small number of topics that everyone can post and view as they please without extending the topic list to 17 pages, I'm all for it!
No dessert for you- ONE MONTH!
Orion_0169
Posts: 153
Joined: Wed Aug 02, 2006 8:17 pm

Post by Orion_0169 »

lol I too would rather not see them, but I suspect curbing the enthusiasm won't fly so I suggested the organisation. ;-) Most of the "look at this" I don't bother with now. Unless it's ballerinas with no heads and dogs. I've actually been writing down some amusing ones... haven't got to posting them yet. I have an awesome one of a gargoyle eating the Eiffel tower.
Je ne peux pas regarder la Mer sans me demander qui vit au-delà de cela.
Blarrrg
Posts: 47
Joined: Wed Aug 02, 2006 5:48 pm

Post by Blarrrg »

You have to admit, the communist one was cool.
Orion_0169
Posts: 153
Joined: Wed Aug 02, 2006 8:17 pm

Post by Orion_0169 »

Yeah yeah! It's the "I found a real one" ones that are... well communist ones are much better. I still say graceful ballerina though. Dog and all.
Je ne peux pas regarder la Mer sans me demander qui vit au-delà de cela.
sriley
Posts: 2
Joined: Sat Aug 05, 2006 12:33 pm
Location: Shingle Springs, CA
Contact:

Post by sriley »

I'd like to see a way to look at calibrations that I failed on, to see what it was I missed, to hopefully learn to do better in the future. I don't always see that the failed calibration counter has incremented until later - so have no easy way to go back to review what happened - or where I may have gone wrong.
icebike

Post by icebike »

sriley wrote:I'd like to see a way to look at calibrations that I failed on, to see what it was I missed, to hopefully learn to do better in the future. I don't always see that the failed calibration counter has incremented until later - so have no easy way to go back to review what happened - or where I may have gone wrong.
Try your Back button.

But seriously, you are doing well to not fixate on the counters. Just do your best on each movie, and the counters will take care of themselves.

Everybody misses a few.
largeeyes
Posts: 3
Joined: Sat Aug 05, 2006 11:44 pm

Post by largeeyes »

How about a top 10 of movies with the highest number of "hits" that people have seen trails on?
icebike

Post by icebike »

largeeyes wrote:How about a top 10 of movies with the highest number of "hits" that people have seen trails on?

Better yet, get the dust team started reviewing the top ten and if these are indeed found to be tracks, USE THEM FOR CALIBRATION MOVIES.
Post Reply