Layer Frame Preloading

[annehaley]

For large layers like the flood simulations, we should create an option for the user to preload all of its frames in the background so that switching between frames is more seamless.

[BryonLewis]

There are a couple of different research options to improve the speed at which items are being loaded.

Attempt to combined multiple tif files into a single tif file with multiple layers:

• Prevent IO access to multiple files and hitting different tile server endpoints, overwhelming the server
• This could show a performance increase by filtering by layer within the TIF instead of different files.
• It may make the tif file larger than what we want

NETCDF/GeoSpatial Image Sequence Image Mapping:

• If the data is low res enough (8k x 8k or less) you could create an image for each tif file and store it in minIO and then serve it directly
• The images can then be bound to their corner endpoints
• This means that the client is just accessing minIO/S3 directly and doesn't have a server impact
• You can design a caching system on the client to precache images in the sequence making it quicker to scrub through them at a higher rate
• The downside is that you don't have the level of detail loading that you get with tile servers and zoom levels. This is beneficial each each TIF is smaller and doesn't have a high resolution.

Hybrid Approach:

• Create a Video for the TIF images at a decent resolution on the back-end
• Use this image and bind it to the Video layer in maplibre so that when playing the animation, you have this video directly playing.
• When the user pauses it could swap to the tile server endpoint so the user can zoom around.

[BryonLewis]

Making some updated notes about this task.

Possibly use the django-larg-image thumbnail capability for a raster image.

• This raster image sequence could generate thumbnails for each view
• When sliding across frames it could transfer between thumbnails images mapped to the surface and when settled on a frame for more than 1-2 seconds it swaps to the underlying image

Technical Information:

• Thumbnails are included as a new model that has the MinIO/S3 reference and the geospatial coordinates for the mapping
• Have it so when a layer contains multiframes it auto generates the thumbnails
• Updates thumbnails on the styling changes because
• When loaded in the interface, precache around the current frame some number before/after frames of the thumbnail images
• For styling quickly possible to use svg transforms