Replies: 4 comments 8 replies
-
As the number of files grows, the next request may be to categorize them with tags and ratings, etc. Eventually, they will be classified in some way. |
Beta Was this translation helpful? Give feedback.
-
What I really wish for is an option to either favorite them or sort by when they were last used. I tend to end up using the same ones over and over. (You can somewhat do tags by putting them in subfolders...) |
Beta Was this translation helpful? Give feedback.
-
while lora/locon/loha/lycoris/lokr/ia3 (yes, some may not even be that well known) are different types, they all belong to same family of networks - lora. what i have done in the latest update is to hide extra networks tab that do not have any extra networks. so for example, if you don't have or use hypernetworks, there is no reason why it should clutter the ui only to say "no networks found". |
Beta Was this translation helpful? Give feedback.
-
I feel an "easy solution" would be an 'all' tab that is defaulted by settings if implemented. Could fill in the 'Extra networks tab order' box, or a tooltip with the network names so it's more apparent to end users on what it is and how to use it. Speaking of tooltip, there is a typo in the tooltip for 'Extra networks tab order' right now lsited instead of listed. Left-most would be the first "default" tab that appears first. Would also make it easy for users to remove tabs they just don't want a tab for. |
Beta Was this translation helpful? Give feedback.
-
I've always thought this distinction in the UI seemed pointless. What benefit do I get from having these in separate tabs? They all modify the output in some way and I'm just left with the extra burden of switching tabs to find the right models. A model either goes in positive or negative prompt, so what type of model it is doesn't really concern me.
I appreciated the merge between LyCORIS and Lora as it made it easier to not have to switch tabs to find the right model. On the filesystem level it also made it easier to organize instead of having 2 separate folders like
lora/styles/somestyle.safetensors
andlycoris/styles/someotherstyle.safetensors
So I would personally take this a step further and merge hypernetworks and embeddings into a single tab and even the same folder on the filesystem.
Technically it should be possible to distinguish between these models but it would require reading the file. It should also be possible to just introduce a new directory where you can just place everything so you wouldn't break existing setups.
Beta Was this translation helpful? Give feedback.
All reactions