Closed amiantos closed 1 year ago
V over on discord says I should probably just use dry_run
... I'll debounce it a bit to keep unnecessary requests down
Implemented today with a short debounce, so far so good. Sure, there's a half second delay between when you change settings and the kudos updates, but that time is probably spent scrolling back up to hit generate...
Looks like every client estimates kudos costs locally, probably to reduce pings to the remote server ('cause I think there is a kudos cost estimator endpoint, but hitting that constantly is no bueno).
https://github.com/aqualxx/stable-ui/blob/abe38b5ec1703fbd212941d686ab4b85f88d9181/src/stores/generator.ts#L361 https://github.com/daveschumaker/artbot-for-stable-diffusion/blob/a88a9aaf36b96e8f3eb8f2eb6dd6f2c3f193a541/utils/kudosCost.ts#L22
I'm undecided on the UX for kudos cost estimation. I want to be a little Apple-y and hide the kudos costs until they are truly relevant (like if they get high / you hit the pre-pay limit, etc). But then there's the question of where to show that temporarily if it isn't always on screen.