Allow prompting through url endpoint

[Logan-Mart]

I think it'd be useful if instead of having to visit the site, you can streamline the process by enabling users to directly add their prompt to the URL endpoint, so you don't have to alt tab out of your IDE and grab the session id. Please add this feature, and let me know if collaborative work is up for discussion.

[vehm]

This is a good idea, however there would be limitations due to how it is being stored in the backend.

Currently, when a user submits the form on the website it queues three image generations and links those two a single session ID. If a user were to only supply the prompt, we would be relying on the uniqueness of the provided prompt since they wouldn't have an ID to link their particular request to. This would be pretty unavoidable, as far as I can tell, and I'd probably make it to where it simply grabs the most recent submission where the prompt matches the provided prompt.

For instance, if a user were to prompt for "red phone" a day ago and I prompted for "red phone" right now, they would pull my generated images since our prompt was the same. Otherwise, we'd need some sort of unique identifier which I don't see how we could accomplish through only the URL params.

Another option would be to generate new images each request, but this would make my generation costs skyrocket.

[secretgspot]

came here to say this. instead of dealing with session id it'd be nice to be able to prompt in the url string itself. limit prompt to lets say 100 characters or less, also since this is placeholder service then it shouldn't store images in your db since that waste space. instead it should generate and cache image for a day or so after which it would generate again when site is visited. most of the time services like that are used for web dev with low or no traffic, so having new images generate for same prompt is not an issue. you'd encode it in url text=Red+Phone

[vehm]

also since this is placeholder service then it shouldn't store images in your db since that waste space. instead it should generate and cache image for a day or so after which it would generate again when site is visited. most of the time services like that are used for web dev with low or no traffic, so having new images generate for same prompt is not an issue. you'd encode it in url text=Red+Phone

The main issue I see with this is potential repeated generations if a placeholder would be re-requested consistently. Purely for reasons of cost, I'd prefer storing them and either returning a random one that matches the requested prompt + generating a few new ones.

I've unfortunately let this project fade into the background, but I do have another version of this that is better, albeit incomplete. I'll start working on the new version again and implement this feature, likely deprecating this version. My goal with the new version is also to make this easily self-hostable. I'm not currently a fan of how this version "hides" everything behind AWS.