Sometimes changing the weight of the current prompt card using the mouse wheel doesn't work.

Although I had such a bug, I was not able to reproduce it in a later time. I am not sure under what conditions this bug occurs.

Hey, first of all thx for the extension.

I have ran into a little problem, it wont save previews if they have spaces in them.

PromptsBrowser: 18:17:53 -> failed to save preview: file "F:/sd.webui/webui/outputs/txt2img-images/2023-10-02/realisticVisionV51_v51VAE/realisticVisionV51_v51VAE%20-%202023.%20Monday.%2002.%20October%20-%20kl.18.15.51.png" not found

Are there something you can do about it, or do i need to change the way i structure my images?
I like to have it like "realisticVisionV51_v51VAE - 2023. Monday. 02. October - kl.18.15.51"

Unless i'm just not seeing it, there doesn't seem to be a way to rename existing styles or collections. If i want to reorganize or fix a spelling mistake it would be great to just simple be able to rename these things.

Without an internet connection, some browsers may wait too long to download fonts from google services, which causes the extension to expire its loading attempts timeout.

Rising number of attempts or timeout between them can solve this, but this is not a very elegant solution.
Can be changed in the index.js - PromptsBrowser.initPromptBrowser method.

The Issue

The ability to add preview images to each prompt in a collection and each style is super helpful and one of the main reasons i got this extension. However, often time different checkpoints or loras could result in a drastically different look/effect. The most obvious is anime vs live action, if i'm using a live action checkpoint, previews in an anime style may not be as helpful.

The Solution

I propose we add a dropdown that lets the user select a preview source. The user should be able to add and remove new entries to this list. So lets say we have preview source A and B. If source A is selected then all previews for both prompt collections and styles should show previews associated with source A. If the user then selects source B then previews will swap to only show the previews for source B. This allows us quickly see what are prompts would look like in various styles that we set.

Considerations

• I imagine it should be implemented in a way to that allows filtering based on if a preview exists to still work properly depending on what preview source is selected. So if we only want to see prompts without a preview, and we add preview to every prompt but it's only tied to prompt A then when this filter is active no prompts should show. but if we change preview source to B then (which has no previews set for any of the prompts yet) then it should show all prompts.
• When generating new previews the previews should be saved to the currently selected preview source

I made a collection of my negative prompts so that i can just click from a list of them to add. However, it seems like the style won't save unless there is at least one positive prompt. Would be nice if this wasn't necessary

Discussed in #8

The code had become much larger than I originally planned and it is not convenient to continue developing the project in its current form. It's worth switching to Typescript and automatic bundling.

I was saving all my portraits at 1664 x 2432 which is the normal size i typically render at since it often results in some extra details and bigger is always better. We'll except that it seemed like this was causing some massive slowdown when scrolling and hovering over prompts in the prompt collection. I converted them all to 104 x 152 and performance was instantly improved.

Instead of using the original full sized file we should create a thumbnail version of each preview that maintains aspect ratio. The original should still be saved though as the larger sized image is still really nice for the larger hover preview and background previews.

I have spaces and other symbols in path to file, save preview didn`t work, so I have modify "src" in savepreview.py

from urllib.parse import unquote
src = unquote(postJSON.src)

Originally posted by @fin-gal-25 in #10