https://github.com/nej-dot/loralist
This is a python based toolkit to parse long strings of loras to a format that either xyz plot or "prompt from file" can understand, (comma and line separted). It has a bunch of other features as well, and im always wiling to add more or change if anyone has suggestions.
currently it also has a find and replace, and a parse list function.
If you leave the input box for these two functions empty it will default to work on the text in the workspace box.
one feature is a bit tricky, if you want to use several loras in a plot, but also want several values for each, ie <dkjsdfk:0.4> and <dkjsdfk:0.8>, first run a find and replace for ":1" to ":x.x", and then set up all the values for your list, start stop etc, and then fill in blankspace for your separator, and press the parse and generate list button.
Description
FAQ
Comments (5)
This looks useful but I still dont really understand it. Maybe a video of it in use?
yeah, probably need to rework the ui, it doesnt quite make sense atm.
Yeah, could be useful, but I don't fully understand how exactly it works..
The simplest way to interact with it is to just treat each boxed frame as a separate script/function. Right click on things to get info
Think it would be helpful to explain the purpose of "Prompting From File" and "Xyz Plotting"
I'm just getting into SD/AI so everything is a bit touch and go, but...
from what I understand, the Tokens (blocks of the prompt if I observe correctly) have a softcap of 75, where you are able to go infinitely beyond, but, your vector (strength of tokens) will permanently be nerfed from 0.5 to -0.5 after ~75(or some point exponentially close to 75), therefore prompts require a certain degree of efficiency to be effective and not wasteful or have null effects on the AI's generation....
If I understand all that correctly, I'm assuming your goal here is to bypass that limit, if Prompt From File, or XYZ Plotting, don't have such limitations? If so, I'd say this could be a cornerstone project seeing as I don't understand PFF, or XYZ, so a tool to help newbies unlock more prompting power would go far in the community overall, imo; and if I'm assuming everything correctly lol.
only other thing i can think of is it would help prevent the artificing issue? As I've noticed if you don't change the weights on Loras, then it literally saturates the stackflow and you end up with either minor artifacts, or a complete green/purple washed and warped screen. Knowing a bit more about the Loras you're using, like which ones effect this scenario the most, would be handy as well.