Trigger: helen parr
Extra prompts:
- 3d (styled like the movie)
- superhero costume (red and black costume)
- mask
- mind control goggles
- alternate color, bodysuit (her new grey costume. If the "bodysuit" tag is giving you trouble you can try removing it)
- bathrobe
- no mask (usually not necessary, but her hero costumes might sometimes gen the mask with it. Use this to remove it)
Tested on Illustrious Personal Merge: https://civarchive.com/models/835655?modelVersionId=1023901
It should work on other models as well (such as NTR Mix and WAI), but may have less of an style impact depending on the model
I suggest Euler A/Karras, CFG 5, 25 to 35 steps, but feel free to experiment and come with better settings.
Strength between 0.85 and 1. You can also use Lora Block Weight on lower steps (1 up to step 5) without much negative impact if the gens are deviating too much from the main model.
ATTENTION: This was trained as a DoRA, if your local install (A111/Forge/Comfy/etc) is a few months old it might not be compatible. In this case just update your install and the model should load properly.
Description
FAQ
Comments (19)
anyone else having trouble using this model after downloading
yep
What trouble are you having? Is is not showing up on the available networks? If so, you can try enabling the "Show all networks" option on A1111/Forge
Alright, I tried downloading the model from civitai and apparently it's corrupted. I had an issue while uploading the lora and apparently it messed it up. I'm uploading a fixed version atm.
Edit: Done. It seems fine now after reuploading, please try downloading it again.
tried copying generation data over but not getting anything like the sample images from the creator
Can you please try redownloading it? When I first uploaded it to Civitai it got corrupted. I have since fixed it but you need to delete the old model.
redownloaded looks like it works fine now thank you!
...Wow! Thank you! 🙇
I know my Forge is probably the problem, but it's one of those Lora's that doesn't show up on the list. Whereas your Violet Lora is displayed. There must be something in there somewhere that makes the XL models in the Forge not see her.....
Ok, "Show all networks" fix that. But unfortunately, another problem came up: it kind of doesn't work. I mean, it draws something that doesn't look like what it should be. I've been using Illustrious XL personal merge and nothing. It's the same thing with Violet's lora, by the way.
@AvelN It's probably not loading the LoRA then, even after you included it in your prompt. The results of the gens you see are probably from the checkpoint's own knowledge. Check Forge's console for any errors. I had an issue similar to this with LoRAs that had double underscore on the name.
Yeah, that's what happened. My version of Forge doesn't want to load this version of Lora. I didn't think of it right away, because I've only had this problem with a couple of Loras out of several dozen. Thanks for the tip!
@AvelN Did you find a fix? Try renaming the lora file to something very simple, like Helen.safetensors, and see if it helps, that's what fixed a similar issue I had with some loras a while ago.
@Nedok No, I just made sure that the console says something like “LoRA version mismatch for SDXL”, and then I ran the latest update of Forge, instead of the one I'm using - and everything works there.
As for renaming it - no, the result, unfortunately, is the same.
@AvelN Gotcha, so in the end you made it work by updating Forge?
@Nedok Weeell, kinda. I mean, technically, on the new version of Forge these loras work (They work great, by the way!), but I don't want to move from the old Forge, familiar to me, to the new one for the sake of a couple of loras.
But in any case, thank you for your involvement and advices)
@AvelN Oh that's some very nice piece of information, I think I figured it out. These LoRAs that I posted were trained as DoRAs, it's a somewhat new way of training them. It is widely supported right now, but if your install is a few months old it might not be compatible with it. I'll add this as a disclaimer on my posts. Thanks!
@Nedok Hmm, yeah? Then maybe I can get them to work on my version, because I think I once saw instructions to install some add-on extensions to make DoRA work on SD versions that don't have them originally. I'll give it a try, thanks!
@Nedok And hell yeah! It worked!
Short instructions: need to find the files “lora.py” and “model_patcher.py” in "webui > ldm_patched"
and replace them with new ones from here:
https://github.com/lllyasviel/stable-diffusion-webui-forge/pull/608/files