Page 15 of 19

Posted: Sat Sep 30, 2006 11:33 pm
by CHudnall
jsmaje wrote:You're absolutely right that the eye and brain can't take in everything at once, and that scanning quadrants in a systematic way is the best way to go. You can then take another close look at anything of potential interest, and finally make a holistic assessment (repeatedly focussing up & down can help spot deep features behaving differently from surface ones, even in one's peripheral vision).

An astronomer on this forum who is used to viewing astrophotos said exactly that (sorry can't find the link - the so-called Search facility on these forums is pretty-much useless, as also pointed out by others before). As an ex-doctor, that was the way I was taught to scan an X-ray, and as an ex-private pilot, that was the way to scan the sky out of the cockpit window (the four most vital cockpit instruments were also scanned systematically, though in a T-shape in that case).

It's a matter of training yourself to do it that way, and once in the habit it becomes instinctive. Since there are already going to be at least 700,000 focus movies to be looked at by the time they're finished, I don't think multiplying that number by four would really be much help.

And by the way, from the Tech FAQ: "We recommend using at least 1024 x 768 resolution."
That is the best way to scan, in quadrants. I'm not the astonomer that you speak of, but this is how I search, not only in the telescope but also my images (yup, astrophotographer also).

Instead of changing the program to present 1/4 of the image, why not just do what astronomers have been doing for years, make do with what you have. Meaning, if you only want to see the top left quad, take a piece of paper, cardboard, whatever, and cut it to shape. Wah-lah, when you hold it up, you only see one quad.

I'm not trying to put down the suggestion, I'm just saying that not everyone should have to see only one quad at a time. What I do is focus on one quad at a time, running focus up/down numerous times. I learned to do that searching the sky, where to aim my focus. It gets easier with time. :)

----------------------

Christine

Re: Pixel/micron rulers

Posted: Sat Sep 30, 2006 11:49 pm
by CHudnall
Crow T Robot wrote:With minor changes to its HTML code, Stardust could display neatly-drawn horizontal and vertical rulers. *Snipped - removed image from quote.*

Shall I propose these to the Stardust overlords? Mostly they just look pretty, but they would also make it easy to describe the location of subtle features.
If you want to describe a location on a movie, why not just give the coordinates to that feature you want to show?

Example: Movie # so&so: upper right quadrant: 452,36

If you look at the address bar at the bottom of the screen as you hover your mouse over area, the coordinates show on the address, at the very end.

Hope that helps. :)

---------------------------

Christine

Re: Pixel/micron rulers

Posted: Sun Oct 01, 2006 3:30 pm
by jsmaje
CHudnall wrote:If you want to describe a location on a movie, why not just give the coordinates to that feature you want to show?

Example: Movie # so&so: upper right quadrant: 452,36

If you look at the address bar at the bottom of the screen as you hover your mouse over area, the coordinates show on the address, at the very end.
OK, I've found that the coordinates appear in Netscape/Mozilla on the bottom bar, but not (as far as I can make work) in my version of IE 7.2.

Since the former browsers load much quicker, I've now decided to use Netscape. But, for some as-yet unexplained reason to do with javascript handling(?) you don't get the "Are you sure?" confirmation box after clicking on something. It's not much of a problem though - who would click on something they didn't think there was there? - except it may prompt you to double-check.

Aren't these browser inconsistencies really annoying?

Posted: Sun Oct 01, 2006 5:57 pm
by Mighty Pete
jsmaje, I kind of like the pop up not working..

We don't really need the extra questions:

You cannot delete the recycle bin ! Are you sure you want to delete the recycle bin and put it in the recycle bin ?

Posted: Mon Oct 02, 2006 2:58 pm
by jsmaje
Mighty Pete wrote:jsmaje, I kind of like the pop up not working..

We don't really need the extra questions:

You cannot delete the recycle bin ! Are you sure you want to delete the recycle bin and put it in the recycle bin ?
Indeed - a backwards recursive loop leading to minus infinity! :lol:

Please move post if in wrong place....

Posted: Fri Oct 13, 2006 4:04 pm
by Howie
Hi There,

I don't know about all of you, but, I am getting tired!

I am waiting for the team to give us some info as to weather we are heading in the right direction, so as not to waste their time...

Reading all the boards and learning from all the posted examples etc.
I know for keeping us honest and alert they are making the CM's smaller and trickier, especially the one without a track!

I know they are busy and a small group, but, I think by now they know what we are clicking on, and, they should be telling us what not to click on
like the inclusions, many, and sure would like to see, unless I missed it, an update on the latest possible hot tracks, like a poosible real one,,,

I have not heard of anyone with a second comment, that would be nice to here?

I am working my butt off everyday couple hours to stay at a rank of 180 and that means I have 2000 movies to get to 100!!!


Statistics for Howie

Your Overall Score: 4429
Total Movies Viewed: 16566
Your Rank: 180 out of 16226
Specificity: 97.53%
Sensitivity: 99.26%

Posted: Sat Oct 21, 2006 1:38 am
by Senator
Howie, you can see the best candidates found so far at: http://stardustathome.ssl.berkeley.edu/ ... .php?t=879

This project should be fun and I figure that if people would quit worrying about their score, they could spend more time working on the quality of their searches. Just my opinion 8)

Posted: Tue Oct 24, 2006 2:31 am
by steenkh
Senator wrote:Howie, you can see the best candidates found so far at: http://stardustathome.ssl.berkeley.edu/ ... .php?t=879
No. That was posted on August 16, and a lot has happened since then!

Why not point to the track in a calibrate movie I get wrong?

Posted: Sat Oct 28, 2006 3:37 pm
by Mikey
It would be very helpful to see what I got wrong on a calibration movie.

Why not have the next screen following an error be the calibration movie with an arrow pointing to the spot?

wrong avaliation as track

Posted: Sat Nov 04, 2006 6:44 am
by gilmano
I've seen in your Forum that "focus" occurrence (problem) is highly mentioned. I agree ...

Also, I think that you should show the user the wrong "track", classified by him as a TRACK, since there is no way to improve our classification if we don't see what our error was :?: :?

Seeing what we missed

Posted: Sat Nov 04, 2006 6:59 am
by fjgiie
Hi gilmano, welcome to the forum :)
Yes, there are some bad focus movies. See this link to review when to click the bad focus button:
http://stardustathome.ssl.berkeley.edu/ ... .php?t=645

When we miss a calibration movie we can hit the back button and see
if we can find what we missed. See this link:
http://stardustathome.ssl.berkeley.edu/ ... .php?t=467

Thanks,

fjgiie

Posted: Thu Nov 09, 2006 1:41 am
by ac220
Mr_Zeno wrote:I'd like to see a button that says "No track but has an interesting feature" :)

The reason is that I clicked on something I thought was an impact line of a slow small object, just to find out that it was a callibration movie. I lost my 100% because of it, even though I still think it was interesting :( I hate getting things wrong :(

Jason
I'd second that. In fact I think that there should be 4 options:
1) Obvious track
2) Possible track
3) Unknown object
4) No track
Maybe with 1 and 3 available after certain calibration score...

But you'd need to solve firefox compatibility problems to implement that.

search suggestion

Posted: Fri Nov 10, 2006 6:24 pm
by WorWizard
Just a suggestion. Could you make the searchec, instead or completely random, contiginious. In other words instead of getting 1,1000,2345,654,1 you get 1,2,3,4,5 and at the end repeat 1,2,3,4,5 ( if you reashed the end that is). this way you would be more encouraged to see the next slide or the opposing slide. I've see that same slides many times, I think they need ew eyes to focus on them because if I see them over ond over again I tend to click the same things that someone else may not. or may.

Ken

Add out of focus as a CM metric

Posted: Sun Nov 12, 2006 5:58 pm
by Tcarey
I suggest that a new metric be added to CM scores.

Correctly identifing a movie that is out of focus or the focus range is not sufficient for proper evalutaion of tracks.

I see a number of movies that do not have sufficient down focus to detect a track like the ones routinely shown in the CMs.

It makes me wonder how many frames are being classified as no tracks when they should be tagged as bad focus.

An additiional comment about the request for failed CM tests to show the missed target. If nothing else, it would be good if a failed CM at least played a sound to let you know to press the back button to reexamine the CM to find the missed target.

I try to look at the numbers for errors to see if I missed one, but at 1024 x 768 that number is below the edge of the viewable window and I have to scroll the screen to see it every frame. I frequently don't do that and then miss the the error and the opportunity to learn from my mistake.

Re: Add out of focus as a CM metric

Posted: Mon Nov 13, 2006 3:42 am
by GelDelve
Tcarey wrote:I suggest that a new metric be added to CM scores.

Correctly identifing a movie that is out of focus or the focus range is not sufficient for proper evalutaion of tracks.

I see a number of movies that do not have sufficient down focus to detect a track like the ones routinely shown in the CMs.

It makes me wonder how many frames are being classified as no tracks when they should be tagged as bad focus.
"Bad focus" includes movies that do not focus below the surface enough to decide if there is a track or not, so I think that another choice would only make it more confusing, rather than adding a benefit.

You are correct about the scores showing up below the viewable window. I have made it a routine to scroll down so that the focus bars are at the top of my viewable window as soon as a new movie starts to download, and, at least, on my screen I can see enough to know if my "answered incorrectly" calibration score has increased while I can also see if the new movie has fully downloaded or not. It would be terrific if there were a sound effect to let us know we got a CM wrong, especially if it could differentiate between 1) You clicked on a track and there is none, 2) You didn't click on a track and there is one, 3) You clicked on a movie with a track, but your click wasn't on that track. Unless you are watching your score by scrolling down and also remembering what you did on that last movie, this is the only sure way of knowing what to look for when you go back to see what went wrong. But then you'd have to know what each sound meant, too. I don't think we're going to see any changes like these anytime soon, so for now I'm going to just keep track of it the way I always have.