MountaintopLotus / braintrust

A Dockerized platform for running Stable Diffusion, on AWS (for now)
Apache License 2.0
1 stars 2 forks source link

Training dataset #18

Open JohnTigue opened 1 year ago

JohnTigue commented 1 year ago

If we're going to be running a bunch of experiments, then we should have a well defined training set. We'll want to have art in multiple folders by topic. Then we'll want to preprocess/clean the data.

The VM will want the data on EBS but long term storage should maybe be on S3?

JohnTigue commented 1 year ago

PNG vs JPG format for the highest quality?

JohnTigue commented 1 year ago

I guess a spreadsheet of Verses lore based prompts IS indeed part of a reproducible research training data.

JohnTigue commented 1 year ago

Get Mike's 3D assets into the training dataset: https://discord.com/channels/800449043414515803/801731426613198898/1062470174927761408

JohnTigue commented 1 year ago

Lots of potential to play with in The Wasteland Express.

And what about all the rest? Foundataion Edition NFTs