Broken Nodes!

[SRez]

Hi,

I have two points I would like to discuss;

First, this perhaps has been refer to (not sure) in ‘Bug that stops interpolation requester in tracing mode #598’; however, while tracing, the nodes on the skeleton disappear when one switches between layers!! Fragmented tracing.. it’s not even recorded in the table of nodes. Is this something that you have noticed as well?

Second, I have a request. Is it possible to have a ‘contrast slider’ – this is very important for us. There is already a ‘brightness slider’, which is a blessing.

Thank you,
Reza

[tomka]

Hi @SRez, thanks for your bug report. What exactly do you mean by "switching layers" (e.g. going through the stack in Z, manipulating the opacity of different overlays)? And do nodes you create first appear and than vanish again?

For the contrast slider: This is certainly something we want to have, too. I once created infrastructure for having such sliders which let the back-end manipulate the images. This is, however, not practical, as you can easily kill the server with many image manipulating requests. Instead, I will look into letting the front-end (browser) do that. The whole topic is covered in issue #186.

[SRez]

Thanks for the reply; yes, I meant through the z-stack up and down - tracing - when I place a node (green dot) on a neuron and move on to the next layer, often times a node appears out of nowhere, without me clicking on the neuron; I continue tracing normally. When I return back to the traced layers to check my work, some of the nodes (dots) have disappeared or have moved slightly to a different position! I should go up and down through the z-stack and see the neuron I clicked on with a 'dot' on it, exactly where I put the node, correct? --just like TrackEM.

Thanks for looking into these issues;
with much appreciation,
Reza

[SRez]

I saw your reply to the post: https://groups.google.com/forum/?hl=en#!topic/CATMAID/wnNhxx7EYNo
Perhaps we need to do an update.
Which browser, in your opinion, is best for CATMAID?
With respect to ‘contrast slider’, a front-end solution would be fine – I would be happy with any type of control over contrast.
The reason a contrast slider would be of great aid in tracing is that often times one gets folds on a section as dark lines; and while the fold might appear dark (product of image acquisition and exposure settings), there is actually information there. By changing the contrast, one could, more often than not, see the information beneath the fold. It is also a useful tool for deciphering synaptic densities.
Thank you again,
Reza

[tomka]

Most people seem to use CATMAID with Chromium or Chrome. So that is what I would recommend. Firefox should work as well, at least I use it usually when working on CATMAID.

For looking into this, it would be good to know what Git version your CATMAID instance is running on.

Regarding the contrast sliders: let's continue discussing this in issue #186. This issue (#606) should deal only with the problem you reported and not the feature request.

[tomka]

So I pushed a fix for #598 (the one you refer to in the first post) last Friday. Would you able have your CATMAID upgraded and see if this fixes this issue, too?

[SRez]

That's great, Tom; thank you. I will let you know how it goes. cheers

On Mon, 31 Mar 2014 08:46:49 -0700
Tom Kazimiers notifications@github.com wrote:

So I pushed a fix for #598 (the one you refer to in the first post) last Friday. Would you able have your CATMAID upgraded and see if this fixes this issue, too?

---

Reply to this email directly or view it on GitHub:
#606 (comment)

[SRez]

Hi Tom,
Sorry it took so long for me to reply about the Catmaid upgraid and user feedback; (many levels of admin to go through!) Well, after the upgrade, the problem still persists - the nodes don't appear in every layer while tracing. Also, I've been reading through the posts trying to make sure I don't complain about the same thing - However, now the 'Object Tree' icon has been replaced by a 'Star' (Navigate Neurons) icon and I can't find the option to create new neurons! It's a bit confusing; not as intuitive as the 'Object Tree' for some reason! Any ideas?

[acardona]

Hi SRez,

1. Have you done a deep reload of your CATMAID webpage after the update? Shift+F5 would do. There should be more than just the "Navigate neurons" icon; there's also the "Query neurons by annotations" next to it.
2. Renaming a neuron:
   a) shift+F2 always worked.
   b) In the "Navigate neurons", select "Active neuron" and see several buttons, such as "Rename neuron", right there at the top.
3. On the "Object Tree" being gone: the new system of annotations fully replaces it. Think of an annotation as a label in gmail. Every former folder in the "Object Tree" is now an annotation to each of the neurons contained under it, recursively. To explore the old hierarchy, open the "Neuron Navigator", search for one of your folders, then expand its co-annotations, search again, etc, until having a sequence of "folders" (annotations, now) that restricts the search to the list of neurons that used to be included in a nested folder of the now gone "Object Tree".

[SRez]

Hi Albert,
Thank you, that is clear now about The Navigation and Query and Renaming.
I tried 'Shift+F5', as you suggested, but I don't see any difference in the way nodes appear in the layers as I place them. About 3 or 4 layers from the parent node, there is a message that "node is one section beyond it's parent node"; I continue tracing and when I scroll back down (z-stack) toward the parent node I see that the nodes I placed have disappeared! Not all, but every two or three layers.

[acardona]

SRez,
there must be a floating-point error. What is your Z resolution? Nodes are
retrieved from the treenode table with an equality over their Z. If
javascript and postgres round the numbers differently, they won't match,
and your nodes will not be retrieved by the client even though they are
inserted into the table.

2014-04-16 7:02 GMT-04:00 SRez notifications@github.com:

Hi Albert,
Thank you, that is clear now about The Navigation and Query and Renaming.
I tried 'Shift+F5', as you suggested, but I don't see any difference in
the way nodes appear in the layers as I place them. About 3 or 4 layers
from the parent node, there is a message that "node is one section beyond
it's parent node"; I continue tracing and when I scroll back down (z-stack)
toward the parent node I see that the nodes I placed have disappeared! Not
all, but every two or three layers.

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/606#issuecomment-40586700
.

http://albert.rierol.net
http://www.ini.uzh.ch/~acardona/

[SRez]

Yes! Great, we changed the Z resolution to a non-floating point and now the nodes appear in every layer. However, Now the previous traces are gone! But that's not a big problem - we can do them again.
But I do need to ask you one more time about 'Navigate Neuron' - if I want to create a completely new neuron to trace, is it the Annotate button that is suppose to open up a small window that asks for a new neuron name? Because at the moment Catmaid seem to want to associate everything to the previously traced neurons in the list..sorry that I am asking this question again!

[acardona]

Glad to hear the Z resolution was the issue. One line of code would enable
floating-point Z resolutions, but at the cost of 10x to 20x slowdown in the
retrieval of nodes.

To recover the prior skeletons: these are still in the treenode table. You
will have to run an SQL script that transforms the prior Z coordinates to
the new ones.

About creating a new neuron: make sure no skeletons are selected in the
canvas, and click anywhere: will add a new, single treenode. Then, open the
"Neuron Navigator" and go to "Active Neuron", and push "Rename". Or
directly push shift+F2.

2014-04-16 11:32 GMT-04:00 SRez notifications@github.com:

Yes! Great, we changed the Z resolution to a non-floating point and now
the nodes appear in every layer. However, Now the previous traces are gone!
But that's not a big problem - we can do them again.
But I do need to ask you one more time about 'Navigate Neuron' - if I want
to create a completely new neuron to trace, is it the Annotate button that
is suppose to open up a small window that asks for a new neuron name?
Because at the moment Catmaid seem to want to associate everything to the
previously traced neurons in the list..sorry that I am asking this question
again!

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/606#issuecomment-40613422
.

http://albert.rierol.net
http://www.ini.uzh.ch/~acardona/

[SRez]

Fantastic! Thank you so much for you and Tom's help on all of this. I'm a happy camper!

[aschampion]

Looks like everything has been resolved.