Skip to content

MRG, ENH: Add brain movies to rendered examples#8265

Merged
agramfort merged 7 commits intomne-tools:masterfrom
larsoner:video
Sep 23, 2020
Merged

MRG, ENH: Add brain movies to rendered examples#8265
agramfort merged 7 commits intomne-tools:masterfrom
larsoner:video

Conversation

@larsoner
Copy link
Copy Markdown
Member

@larsoner larsoner commented Sep 16, 2020 •
edited
Loading

Eventually it would be nice to have animations play for some of our source modeling examples. I plan to:

  • Add time_viewer=False to brain.screenshot to allow getting the brain+traces instead of just the traces
  • Add movie embedding by using matplotlib's javascript (maybe just by subclassing their animation class?) and sphinx-gallery's animation code (maybe by making this public? not sure) turning brain outputs into matplotlib figures and using matplotlib_scraper(...) directly
  • Add # sphinx_gallery_brain_movie code parsing that does brain.save_movie(...) then uses the javascript to embed a movie
  • Add tests

Not 100% sure it will work or produce movies of a reasonable size, but it would be pretty cool if it worked.

@larsoner larsoner added this to the 0.22 milestone Sep 16, 2020
@larsoner
Copy link
Copy Markdown
Member Author

larsoner commented Sep 22, 2020

Okay seems to be working:

  • Landing page (400 kB GIF; static image is 288 kB)
  • MNE (7.7 MB; 55 sec instead of 35 sec on master)
  • LCMV (3.8 MB; Also takes ~6 minutes to generate rather than ~3 on master because software rendering of volumes is painfully slow)

HTML sizes can be reduced by reducing the _Brain dimensions, reducing framerates, going to static GIF files, or (maybe; much harder) changing how matplotlib embeds these things. I've already limited the time intervals to a pretty small range (50-150 ms). But maybe these are small enough already to be tolerable.

Let me know if people think it's worthwhile continuing and if so I'll add some tests.

@agramfort
Copy link
Copy Markdown
Member

agramfort commented Sep 23, 2020 via email

@larsoner
Copy link
Copy Markdown
Member Author

larsoner commented Sep 23, 2020

At least for dSPM on circle it's only an extra 20 sec. For LCMV volume I agree it's too much

@agramfort
Copy link
Copy Markdown
Member

agramfort commented Sep 23, 2020 via email

@larsoner
Copy link
Copy Markdown
Member Author

larsoner commented Sep 23, 2020

Yep, you have to tell it which examples to turn into movies via comments. So I can just deactivate the LCMV one for now. Then the only one that's a movie will be mne_dspm.

@agramfort
Copy link
Copy Markdown
Member

agramfort commented Sep 23, 2020 via email

larsoner
larsoner commented Sep 23, 2020
View reviewed changes
tutorials/source-modeling/plot_mne_dspm_source_localization.py
Comment on lines +139 to +141
# The documentation website's movie is generated with:
# brain.save_movie(..., tmin=0.05, tmax=0.15, interpolation='linear',
# time_dilation=20, framerate=10, time_viewer=True)
Copy link
Copy Markdown
Member Author

@larsoner larsoner Sep 23, 2020

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@agramfort the scraper looks for a line that is # brain.save_movie( and if it's there actually makes one using the call (equivalently, anyway). So it allows/forces us to keep our narrative doc descriptive while also allowing us to specify parameters.

@agramfort
Copy link
Copy Markdown
Member

agramfort commented Sep 23, 2020 via email

@larsoner
Copy link
Copy Markdown
Member Author

larsoner commented Sep 23, 2020

Then the next question is: are there any other examples that you'd like to see turned into movies? I'd like a vector source estimate one, I can look for that.

@agramfort
Copy link
Copy Markdown
Member

agramfort commented Sep 23, 2020 via email

drammock
drammock approved these changes Sep 23, 2020
View reviewed changes
Copy link
Copy Markdown
Member

@drammock drammock left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM pending removal of the LCMV movie for now

@larsoner larsoner changed the title WIP: Add brain movies to rendered examples MRG, ENH: Add brain movies to rendered examples Sep 23, 2020
@larsoner
Copy link
Copy Markdown
Member Author

larsoner commented Sep 23, 2020

Pushed a commit to fix the documentation of _Brain and add a couple more videos. Will post a link once it's done.

Should we make _Brain public? I feel like some parts are maybe not ready, but it would be nice for people to know about the API. I can do it in this PR and we can backport...

@larsoner
Copy link
Copy Markdown
Member Author

larsoner commented Sep 23, 2020

@larsoner
Copy link
Copy Markdown
Member Author

larsoner commented Sep 23, 2020

Missed visualize_stc image:

@agramfort agramfort merged commit 724f9ae into mne-tools:master Sep 23, 2020
@agramfort
Copy link
Copy Markdown
Member

agramfort commented Sep 23, 2020

thx @larsoner !

@larsoner
Copy link
Copy Markdown
Member Author

larsoner commented Sep 24, 2020

Should I backport? It does have some nice doc updates in addition to the candy. It would be nice to make _Brain public, maybe with a warning about the API bring incomplete...

@larsoner larsoner deleted the video branch September 24, 2020 00:17
larsoner added a commit that referenced this pull request Sep 24, 2020
@larsoner
MRG, ENH: Add brain movies to rendered examples (#8265)
6f60f44
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@agramfort agramfort agramfort approved these changes

@drammock drammock drammock approved these changes

Assignees

No one assigned

Labels

backported

Projects

None yet

Milestone

0.22

Development

Successfully merging this pull request may close these issues.

3 participants

@larsoner @agramfort @drammock