enough is enough

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

Moderators: Stardust@home Team, DustMods

Post Reply
Chuck Crisler
Posts: 30
Joined: Wed Aug 02, 2006 9:44 am
Location: Windham, NH

enough is enough

Post by Chuck Crisler »

During phase 1 I asked a question about track vs. focus depth. The answer was that the expected dust particle energy would yield a track depth significantly greater than the VM focus depth. Hence, all tracks would be visible from the surface to the bottom of focus. Did someone change their mind about that???

Reference images 5133556V1 and 3773933V1. Both images are relatively clear. However, at the bottom of both images you don't see anything having any focus. Hence, according to info given out, there aren't any tracks in these movies.

However, they are calibration movies, both of which I missed.

I understand the need to make the calibration movies difficult. I applaud that for many reasons. However, shouldn't they follow constraints of the physical process? They are teaching/reinforcing tools. By examining them we learn about a broader range of visual clues. Now, a clue that I was explicitely told exists in the physical world doesn't seem to exist in the calibration movies.

Some of the calibration tracks are just bearly visible (on purpose). However, many of the real images are of VERY POOR QUALITY. If we assume that those miniscule tracks represent the threshold of what we are searching for, then we should mark as 'bad focus' every movie that we feel we could not find one of those smallest tracks on if it existed. Using that criteria, we should be rejecting 15-20% of the movies. Are the staff prepared to re-scan that many tiles?

What gives???
fjgiie
DustMod
Posts: 1253
Joined: Sat May 20, 2006 8:47 am
Location: Hampton, SC, US

Re: enough is enough

Post by fjgiie »

Chuck Crisler wrote:During phase 1 I asked a question about track vs. focus depth. The answer was that the expected dust particle energy would yield a track depth significantly greater than the VM focus depth. Hence, all tracks would be visible from the surface to the bottom of focus. Did someone change their mind about that???
Hi Chuck,
Here is that thread you referenced I believe.
Reference images 5133556V1 and 3773933V1. Both images are relatively clear. However, at the bottom of both images you don't see anything having any focus. Hence, according to info given out, there aren't any tracks in these movies.

However, they are calibration movies, both of which I missed.
Here are both of those calibration movies:
5133556V1 The speck at 187, 201 is still visible at bottom focus. From center of movie the speck is about an inch toward 8:00 o'clock.

3773933V1 The shadow at 157, 228 is still visible at bottom focus. Shadow is located almost in the same place as other movie, left of center and down some.

Calibration movies are only to keep you awake and give you points. They are more difficult in phase two than in phase one, This is on purpose.

"The Phase 2 calibration movies will be much more challenging than those of Phase 1, because they are made using one of the more subtle actual candidate tracks. So, don't expect Phase 2 to be the same old data and the same old search. You will probably have lower sensitivity and specificity scores than you had in Phase 1. We are doing this on purpose. We were amazed at how good the dusters have been during Phase 1 at finding even the smallest calibration tracks, but it meant that we could not really determine our detection threshold. Now some calibration tracks really will be almost invisible!"

Thanks,

fjgiie
bmendez
Stardust@home Team
Stardust@home Team
Posts: 530
Joined: Wed Apr 19, 2006 11:28 am
Location: UC Berkeley Space Sciences Lab
Contact:

Post by bmendez »

Hi Chuck,

I'm sorry for your frustration. But that is the nature of science, when dealing with the unknown there will be plenty of things that do not conform to our expectations. In examining several of the candidate tracks you should note that some of them do not appear as deep as we would expect. But also note that we do not call them "tracks" they are candidates. They may or may not be tracks and we don't know until we check them out. You'll note in today's blog from Andrew that a few of the candidates that were being extracted turned out to be nothing, but that wasn't clear until we dug in and started the extraction process.

Now to the calibration movies... I must reiterate what I have stated about them before. They are not for the purpose of teaching/training you, but to calibrate each individual's level of sensitivity/specificity. The Phase 2 calibration frames are based on one of the candidates, all of which had little resemblance to what we initially expected to find. So it makes sense to use one of them to continue to calibrate each duster and measure the level at which they are sensitive to small features of interest.

When we begin Phase 3, we will have yet more information to work with and will be able to again adjust the calibration movies to measure each individual's sensitivity/specificity.

Thanks,
-Bryan
"I am made from the dust of the stars, and the oceans flow in my veins"
- RUSH
jsmaje
Posts: 616
Joined: Tue Aug 15, 2006 8:39 am
Location: Manchester UK

Post by jsmaje »

bmendez wrote:You'll note in today's blog from Andrew that a few of the candidates that were being extracted turned out to be nothing, but that wasn't clear until we dug in and started the extraction process.
May I venture to suggest (as stated here) that, whilst this may have come as a surprise to the team, it could hardly have been a surprise to most dusters!

And I'm still waiting to be convinced that the first extracted track (if such it be) from 38843 / 9471219V1 is in fact the same feature dubbed by DustBuster as 'Alpha', consensually dismissed by forum contributers as surface debris.
Bryan, you have promised video evidence to the contrary here and here. Can we please know if, when, and in what manner we might actually get it?

Yours, trying-not-to-become-too-despondent, John :)

Meanwhile note that there appear to be only six people (one of whom may or may not be be a computer program) sustaining the current rate of dusting! I know that given a sufficient number of volunteers (>20,000 at one point in phase 1) no individual need work that hard to analyse any particular movie a sufficient number of times. But what if you were to suddenly find yourselves left with just those six? :shock:
bmendez
Stardust@home Team
Stardust@home Team
Posts: 530
Joined: Wed Apr 19, 2006 11:28 am
Location: UC Berkeley Space Sciences Lab
Contact:

Post by bmendez »

jsmaje wrote:
bmendez wrote:You'll note in today's blog from Andrew that a few of the candidates that were being extracted turned out to be nothing, but that wasn't clear until we dug in and started the extraction process.
May I venture to suggest (as stated here) that, whilst this may have come as a surprise to the team, it could hardly have been a surprise to most dusters!

And I'm still waiting to be convinced that the first extracted track (if such it be) from 38843 / 9471219V1 is in fact the same feature dubbed by DustBuster as 'Alpha', consensually dismissed by forum contributers as surface debris.
Bryan, you have promised video evidence to the contrary here and here. Can we please know if, when, and in what manner we might actually get it?

Yours, trying-not-to-become-too-despondent, John :)

Meanwhile note that there appear to be only six people (one of whom may or may not be be a computer program) sustaining the current rate of dusting! I know that given a sufficient number of volunteers (>20,000 at one point in phase 1) no individual need work that hard to analyse any particular movie a sufficient number of times. But what if you were to suddenly find yourselves left with just those six? :shock:
Hi John,

I did try to locate a movie of the extraction of Alpha. While they stream video directly from JSC to Berkeley during the extraction, it seems they didn't preserve that video. Andrew tried sending me some stills, but for some reason they didn't come through email correctly. I haven't had time since then to retrieve those images another way. But I did tell Andrew about your concerns and he assured me what was extracted was indeed the feature that is identified in the candidates list. He said they recognized it quite clearly.

-Bryan
"I am made from the dust of the stars, and the oceans flow in my veins"
- RUSH
jsmaje
Posts: 616
Joined: Tue Aug 15, 2006 8:39 am
Location: Manchester UK

Post by jsmaje »

bmendez wrote:I did try to locate a movie of the extraction of Alpha. While they stream video directly from JSC to Berkeley during the extraction, it seems they didn't preserve that video. Andrew tried sending me some stills, but for some reason they didn't come through email correctly. I haven't had time since then to retrieve those images another way. But I did tell Andrew about your concerns and he assured me what was extracted was indeed the feature that is identified in the candidates list. He said they recognized it quite clearly.
-Bryan
So that's OK then :roll:
Post Reply