|
|
| April 29 |
- added code - notMovingNum + catchUserNum >disObeyMax
- test 1 - out of patience so danceDuration2 = 30 - didn't seem to
be
- added code to make 1 teachDance clock and now the third third problem
is I think fixed
- let see with out of time - timing of the compliant is right although
4 secs seems very quick?
- and make sure we are getting the correct emotions etc with both
version - if not compliant they should have to do the whole of the danceduration2
period - so I'm taking the (or patience) element out of that section in
the outofTimeorPatience()
- dance 3 starts with a repeat - I think that's OK though it seemed
long - the emotionales seemd correct
- height of our friend?
- autnomous button
- this is good I think - no resets tested
- suddenly I can't ci or co tgTeachDance.cxx- urgh
- put in the new copy me instruction
- put in the cut to the trial
- nextdancestep only has one instruction per eweight and one aiff
file - I need variety
- what about stuff like newmoves coming - where are those files and
could I at least use them temporarily
- I'm making some called nextstep
- newmove4_sad.aiff - can be used as is
- also newmove_sad.aiff
- some of the dance steps do not move right hand much! -lets check
them all
- one more for me - together manic too quiet(sound file levels!)
- It would be nice to have a relative stayclose (with time delay)
for the long catcherRemonstration speeches
|
| April 28 |
- New check
- at some point I am getting several dance steps being taught - not
sure where - after a runaway? I think what it does is a step, another step
and then a watch user dance without any intervening instructions - can't
repeat the error easily - got it again
- getting - notteachDanceAction error - which couts in newBrain
- triggered, I think from nextdancestep instruction -
- I think the eweight was mad - and I have a mad action - in tgEmotion
we were passing the string angry!!!
- in general don't feel totally convinced?
- thing too short - bit tall now- 0.6
- I suspect one remonstrate is in face
- not sure if we can go on saying that wasn't right in third third
- changed
- Seems a little bit too samey - one right, one wrong, one right, one
wrong - the comply also has to be randomized a bit
- randomize comply and then depending on whether its dancedbadly or
well randomize the eweight
- check whether its randomizing
- changed firts-thrid to go between hap and manic
- Systematic
- Dancer complies all the time - how does it feel - really reactive?-
how do we get the sense of "pickier" - do I actually need a PICKIER
actionStore ?
- check how it "feels" - maybe it should always be sad if
ypu're wrong?
- check 1 - it seemed stuckhap in 2 and sad in 3
- check 2 - a few mads but not in danced_badly
- wrong and sad
- right and angry
- I checked the randome# generators and they are fine so this is all
serendipity
- Dancer runs all the time
- seemed uniformly mad
- added randomness and now I like it
- Combination
- But anyway I think the disObeyNum needs to increment if the user doesn't
dance - I need to tie the disObey num to the catchUserNum - how?
- can I internalize it's entire calcluation to the Emotion?
- need to make the dancetimer and onlyonce stuff work with the autonomous
system
- need to make something change us from autonmous to !autonomous
|
| April 27 |
- Called and got seats and vegggie meal
- How to test my autonomy
- RUN - see if it crashes or whatall happens
- 120 dancetime?
- yes it crashes! - its because I turned eweight into a pointer and
don't allocate it any memory.
- I've made it into a pointer but now I feel that all eweights point
to each other bu no eweight has any memory - so I still think we'll crash
- or its a problem waiting to happen - Dave says make it a eweight_ = strdup("hap");
I should have no problems unless I try to "strcpy or sscanf into the
string." He suggests I get around it by making #defines or enums instead
- but I don't want to.
- So far when I run I run! -OK its running
- Says great whether I moved or not
- I ran away twice - first time it was I think manic and the second
time it simply ended everything
- I THINK I SHOULD BE MAKING A MORE SYSTEMATIC PLAN
- Prob should also shortcut my RUN - so each time I check it doesn't
take 1min
- took many of the actions out(made thingscene backup)
- changed teleport in script1
- The Plan - first test what happens if the user doesn't move
- is it doing what I think I coded it to do?
- is it working more genreally
- Doesn't Move
- Now we have the Thing responding from tthe notMoving ActionStore
- but the emotion doesn't seem to change?
- Let's Check the time stuff first - is the mood changing over time?
- time fixed - I had the > round backwards -> and now the emotions
change as prescribed - check again -
- so - in the notMoving function - the timeStates works and change
the emotion
- but if I move away - things screw up - we do not get to the rest
of that function
- RunAway - called from tgTeachDance checkUserDist
- In tgTeachDance checkUserDist- I added
- stopMovement Check
- getEweight
- seemed to do work with the movingaway -
- check running away ->recap (we start getting comply- = 0 which
is dancing)
- Recap - I don't want it to go to recap after Running Away which
means changing teachDance
- In doCatchUserRemonstration I replaced setState(RECAP) with userMovingTooMuchUpdate
- will it fly? - no
- Instead of RECAP - I am going to go to WE_DANCE_INSTRUCTION - but
is the "from the beginning implicit in the CATCH_USER_REMONSTRATION
- although they do say from the top, start over and from the beginning
- I think we can take that to mean the beginning of this step - and let
them be - OK this is now good
- Moving
- what is the random #? - 1 and 0
- the first time it was good - then bad
- what is the relationship between comply_ and moving_?
- actually this is correct by the second move I was in the SECOND_THIRD
- maybe the time has to be longer - I need two or three correct dances
before I get picky - lets look at time!
- hmm I got i dance in FIRST THIRD, 3 in SECOND - changed numbers
to .5 and .75 - now I got 1-2, 2-2, 3-2
- I think I need to make it longer
- 180 secs, 1-0.4, 2-0.7, n - good
- also see how changing the proportions effects the rest
- ProblemWand Check - if the whole body of the person moves, the n the
comply will kick in, should I compare it to the body movement (head)
|
| April 26 |
- on the 23rd I planned the changes to the teachdance and emotion
classes
- also start gathering info (hotels, phone #'s etc to bring to Yerp)
- also resend fax to cesnit ( answer yesi)- busy!- sent 27th
- check with Michelle that the fedex went to ARs - received confirmation
by email 27th
- today I must write up that code and start to debug it
- first make all calls from teachdance and relevant alterations
- done except what to put into setTimeState function
- second make empty functions in emotion- make and test
- fill functions and make
- what happens to movementCheck and moving flags if the danceaction
is interrupted? - I've made sure both flags are turned off if we get into
getEweight - if runningAway
- t making with all functions filled
|
| April 22 |
- PC will fit in large grey case 90x54x33cm - emailed Yesi
- wrote Hiroshi and Mike to make PC
- don't forget to call and get veggie food and book seats!
- thing Project changes
- thing teachdancee/emotion
- made .olds of both
- it would be cool if you could swop between automated and non automated
behavior
- issues are
- changing the emotionclass
- figuring out when/where to call it - I think I only want to call it
when the user has just been dancing or been expected to be dancing...or
when they are running away
- figure out how to make it work for the first dance time and then for
the subsequent dance times
- I printed out the code and need to go over how to fit what I want to
do into what exists.
- new first dance action
- shorten trial
|
| April 21 |
- avid time 7-9 pm
- put out still images - one field and de-interlace in fotoshop
- ask Custer again about puttingthings out to quick time?
- gave ars package to Michelle she will fedex it tomorrow
- 200 word abstarct.
- ent email to rebecca
- abstract on web
|
| April 19 |
- do four pages of text for aec!
- 200 words for the English thing
|
| April 16/17 |
|
| April 13 |
- redigitized thing1_316 thing2-316 thing3-316
- however I am not seeing the sequence
- put the hi-8 footage 3/31 ->betacam
- I need to get Dave to do the voice over
|
| TeachDance |
- Within x seconds of the end of the teach dance the Thing needs to be
angry or sad (40 secs)
- Divide total time into 3 or 4 sections?
- EPath 1 - The dancer - one who is trying to obey
- simply get pickier over time
- maybe I need another actionStore - "that wasn't it exactly"
rather than no - or maybe the happy and sad AS's should simply say
that (Check what they do say)
- 40 secs of happy -> manic
- any movement is a yes, you got it right
- 40 secs of pickiness (sad, happy no's and angry/manic yes's)
- every (other) movement is not accepted - picked on
- 40 secs of pickierness (angry no's and sad, angry yes's)
- I don't like that we can go from some kind of instruction to dance
to the exit - this needs to be cleaner
- EPath2 - The not dance - one who runs away and/or doesn't dance
- I think we also add one to the "runaway" count if the
user doesn't dance
- first run/not dance -> happy
- second -> angry/sad
- third -> manic
- fourth -> angry/sad
- fifth angry and immediately ->exit
- What happens if someone jumps from EPath1 to EPath2 - or what defines
such a jump? - or can they simply run in tandem? Make provisions for:
- if the running starts in the second time period
- if the running starts in the third time period
|
| TO Dos |
Things the Walker showings taught me
- Check - the run away thing - seemed to go on too long/too many times
before the thing got mad
- People can get through the walls of the 4things arena too easily -
mayb the invisible collision detection has to be further in - maybe the
thing's shouldn't run so close to the walls
- On the Idesk its hard to manouver in that circular space - should I
make it more linear?
- Sometime a tracking glitch means the rock doen't sit on user's head.
- Timing
- I cut the dance down to 2 minutes - I think if the point (the vagury
of the Thing is got across fast) this maybe long enough
- I cut the second dance moment down to 15 secs - I also think this should
be long enough
- The Trial scene is too long
- Think again about the way the navigation is enabled and disabled
- I think I can work out 1/2/3 general paths that the Things emotions
will take that should account for nearly every situation during teach dance
-since there are Mainly two types of people
- those who dance and those who don't
- some don't get that you should move body - so MUST be more explicit
- I think that maybe we do away with the repetition of steps - until
we come back after rock chase
- Still have runtoRoc avDCS mess
|
| April 12th |
- Scheduled Avid for this week
- Tues 1-3, 7-9
- Wed 4-8
- Thurs 5-9
- Fri 2-6
- Research list of conferences to submit papers to with deadlines
- Write to that Patrick Maun guy
- Check put the Sunderland fellowships - deadline April 28th - see Dan's
mail Mar 13th - I put it in the jobinfo folder
- Make list of things the Walker taught me
|
| April 5-10 |
- 5 - Flew to the Walker Center for the Arts in Minneapolis and set up
Idesk - well Johnny had already set it up!
- We met the fabulously nice Sarah Schultz and the great Steve Dietz
and other good Walker people, Lisa, David, Rebeccae, Lonny
- 6 -
- Dan Sandin gave a lecture for their What's New sereies
- We showed poverty Island on the Idesk
- We realised that the calibration (tracker) was off - the one of us
(Johnny) who had the secret knowledge for fixing this - didn't actually
have the knowledge so it stayed uncalibrated all week
- Wand # one had a glitch that made the joystick stick, wand # 2 had
a very stiff joystick
- 7-8
- MDolinsky and I showed our work to various groups of students plus
interested walker staff
- met Patrick Maun, who is on the board of Leonardo Electronic Almanac
- he said they can fix it so I can link to my papers their without going
through the registration thing - he had studied with Peter Weibel in Austria
and his work sounds great. He is not convinced by any VR he's seen,
not even to the extent of wanting to do stuff in VR himself
- 9
- We were one of the performances/events at the Walker's after hours
party and showed Poverty Island, Th thing Growing and Window Pane
- All in all learnt more about Idesk !
|
| April 3/4 |
- wrote sketch for SIG99
- Dave rewrote and submitted it with images
|
April 2
|
- sound mixdown
- render effects as per Tom k mail
- get tapes and lay black or not
- sound levels checks
- put out to tape
- make lables
- send in
|
| April 1 |
- finished edit
- digitized at hi res
|
| Deadlines |
Sig Video - April 7th
Sig Sketch
Ars - April 25th - video
Japan - May 10th -video
also keep in memory this new conference Drew has mentioned in email today
Eurographics -short papers - April 30th
Visual -Narrative Matrix - Southampton -Nov 18- Deadline April 30
IEEEVR paper deadline Sept 1st
VSMM - May 1st
Comtec
Smart Graphics
Dave's site
|