Help! What is this?


#61

Hum, i guess I was thinking of this thread when I suggested th W.W.S.D feature :P 


http://forum.eyewire.org/discussion/316/w-w-s-d-or-what-would-seung-do#Item_1 

#62

I've been working on this cell for quite some time... when I scroll, it looks correct, but I think the branching isn't really possible. I'd just like to know if my eyes are right here! The second screenshot shows the point of the really odd branching.


Help??



#63

@aimorai ,  you're right that in the 3D it doesn't look possible.  For everyone else I've traced in green the correct line of the piece. 




The way we stain the cells isn't perfect.  The area you had a problem with might not have gotten enough stain to show the boundary of the cell, or there might have been a bad slice.  It's possible for something to look right in the 2D but look and be wrong in the 3D.  On stuff like this it can be a tough call.  I would have left it off because it pretty clearly looks like two different cells.





#64

Today I encountered a cube, that went completely through a dark-grey area. That was the first time I saw this - normally the correct path only touches those areas slightly. Is it something special? It looked really good in the 3d view, so I assumed it is correct...


The JS console catched some requests to: 
http://cache.eyewire.org/volume/83026/chunk/0/1/0/0/tile/xz/96:128 and
http://cache.eyewire.org/volume/83025/chunk/0/1/1/1/tile/xz/32:64

So, maybe it is cube 83026? Or 83025? Or a complete different one?



#65

Another strange cube that I would like to see in an upcoming tutorial evil grin. A lot of holes in it and a lot of blurry parts, which I were not sure about...


Cell 29, Cube 45037


#66

Oh goodness, now that you guys can get cube numbers you'll be keeping me busy, won't you?


#67

Here is a snapshot of a displaced slide, viewed from side. It was simply moved about 20% of the y axis when viewed from above.


#68

Hey guys!  An update, you can now give us the cube # in addition to the screenshot.  It'll make it easier for us to tell what's going on with the particularly tricky ones.  Just type /debug into the chat.  The  Cube ID will be in the upper left corner.


#69

I screwed up and didn't get the cube number...sorry! It's part of the SAC private beta, though.


No matter how I turned and checked and rechecked this one, this absolutely looked correct, so I had to accept it. Do dendrites actually DO this?



#70

Yup, it's called an autapse, that is a neuron synapsing onto itself.  The google results for it are interesting if only because they include a paper by Sebastian and an image from the early days of eyewire.


#71

so, this is an example of one nueron making two separate trails through the cube. In this case, the small section is what the AI offered to start. Is there any reason to complete the second larger trail?


#72

Hi good42n,  You've given a good example of a seed piece that branches in two directions.  It's important to do both of them--if branches get skipped we can miss a big chunk of the neuron.  Your goal in playing shouldn't be to get to the edge of the cube as fast as possible, it should be to get ALL the branches that go to the edge of the cube.


#73

Ah, rprentki - I talked to good42n in chat about this yesterday. Forgot to put an answer here as well.


The "problem" was, that there are two grey pieces visible in the overview (The long one at the bottom and the small one in the upper corner) and if either both parts should be colored (even tough they are not connected directly) or only the AI one.

Delayed answer: Only complete the path the AI started for you. This has something to do with the hierarchy of those cubes. Especially in the Starburst Challenge it is possible that two branches are visible in the overview in one cube - but belong to different cells! Coloring both would confuse the AI and enrage the GrimReaper :)

#74

Thanks for the help Nkem.  Always a good idea to stay on the GrimReaper's good side.


#75

I seem to have an ongoing problem with adding too much. Here is an example. 


Any hints on how to distinguish?

#76

Is this a merger?


#77

Indeed that is a merger.  An ugly one too.




I've highlighted in red where I think you went wrong with this one.  It's a bit difficult to tell without being able to see the 2D, but that connection looks a bit weird to me.  It's also possible that that piece was just worth 40 points.  

#78


Are those boutons that fully encapsulate a smaller chunk something special? The outer hull is even brighter than the normal "inner grey". Are they just "dye-repelling"?

(btw: I think, it does not belong to the correct path - I left it uncolored)

#79

Merger in EyeWire!


#80

I don't understand this