How to add a new set of images to Galaxy Zoo

A checklist of steps to take when adding new subjects to Galaxy Zoo (on Ouroboros):

Making the images

Select the new images of galaxies that will be input into the system.
Decide on the color balance and scaling for the new images. Beta test on scientists (and volunteers if necessary).
Create 424x424 JPG or PNG color composite images (standard) from the FITS files. Here are a couple of example codes:

Create inverted and thumbnail versions for each standard. Both can be batch-scripted with ImageMagick using the -negate flag (for black/white inversion) and -thumbnail (set the size at 150x150).

Make the data manifest - this is a flat file with the metadata that will be associated with each GZ subject. Important for making sure the classifications can be matched to the astronomical objects, and for displaying information about the subject in Talk, Examine, and Navigator.

If there is a new workflow/decision tree being used, create a new Coffeescript file to be put in app/lib/. Examples of existing decision trees are already present and can be used as templates.
Help text for the workflow also needs to be added to app/lib/en.coffee.
Select 1-4 examples of every possible morphological category in the workflow for display in the "Help" section. Critical that they come from the actual survey being classified, if possible. Can either be done relying on classifications of a smaller subset or (more usually) having someone manually go through images until they find examples of all the categories.
Add example images to public/images/examples. They can be named whatever they want, but the most consistent format for filenames is survey-taskno_a-ansno-exampleno.jpg (eg, sloan-0_a-0_0.jpg). Double check whether tasks and responses are zero-indexed or one-indexed.
Add the list of help images to css/common.styl.

(dev): Put all three directories with the images on AWS.
(dev): Put the raw FITS images (with separate bands) on AWS for access in Examine and the FITS viewer.
(dev): Ingest images into Ouroboros as subjects.
Add the group id and workflowId to https://github.com/zooniverse/Galaxy-Zoo/blob/master/app/lib/config.coffee
Update https://github.com/zooniverse/Galaxy-Zoo/blob/master/app/models/subject.coffee to draw images from the new survey

Have at least some of the images loaded into staging for Ouroboros for testing.
Test every response in the workflow for the new images. Make sure that each path in the tree goes to the correct subsequent task.
Test the new images in Talk. Check to see if the metadata (if present) is being properly displayed.
Test the new images on the Examine page. Click through all external links to make sure they go to the correct location. Check the metadata.
Test the individual bands and all stretches in the FITS viewer on Examine.
Test the new images in Navigator. Test both a defined collection (Favorites/Recents) and selecting randomly from staging. Make sure data being plotted looks sensible.

Change the call to action, if necessary
Add text to the "Story" section on the Galaxy Zoo site (app/lib/en.coffee) explaining the current data/images.
Add any new paragraphs or elements to app/views/story.eco so they appear.
Alert translators that new text is available. Strongly suggest re-using code wherever possible so they don't have to repeat work. Changes in the workflow and help text are typically the most important.

Alert the GZ moderators (Els, Julianne, Christine / ElisabethB, Capella05, Budgieye) that the new images will be posted. Give them details on why the images are new (so they can answer questions) and advance notice of at least a couple of days.
Alert Murray Cumming (@murraycu), who is the developer and maintainer of the GZ Android app. He needs some advance notice so that the app will work with any new subject sets and/or workflows.
Send a short email to the GZ science team.

(dev): retire any active images, if necessary
(dev): activate the new images
Write a post on Talk about the new images
Write a post for the Galaxy Zoo blog
Be present on Talk for the next couple of days. Answer questions and make an FAQ if necessary.
Get aggregated data dumps (within a couple of days, if not a week) and see if the annotations make sense.