LyCORIS of John Oliver
Welcome to my corner!
I'm creating Dreambooths, LyCORIS, and LORAs. If you want to know how I do those, here is the guide: https://civarchive.com/articles/7/dreambooth-lycoris-lora-guide
I upload 2-3 times a week, however, I write down suggestions all the time. If you want to see someone made into a model just leave a comment :)
Suggestions left on my ko-fi page ( https://ko-fi.com/dreambooths ) are taken with the highest priority and you will be able to enjoy them way before they get released here. (I accept also custom requests)
The reason is that I can't keep up with uploading and the models are being made pretty much all the time :-)
As mentioned, via the ko-fi page I accept custom requests, so feel free to engage with me there, I will answer all your questions :)
Description
FAQ
Comments (23)
Special request from @JustMaier :-)
Best likeness yet, I think.
I am happy this is the bizarre future we are in, and i look forward to John Oliver finder this.
not the lora we deserve, but the lora we need
How many images are used for training usually?
in my case - I try to go for 22 and usually end up between 20 and 25.
this is my current sweet spot, but it is possible of course to do with less or more.
the most important part is quality over quantity - sometimes it is best to remove some images if they are problematic (low quality, weird looking face, etc)
@malcolmrey Thanks 🙏, and steps is 3000 yea
@Astro9 indeed, 3000 :)
@malcolmrey i tried to make a similar realistic checkpoint model like yours, following your steps, but i used Google colab notebook of ShivamShrirao Repo, when i use MODEL_NAME: runwayml/stable-diffusion-v1-5, everything works fine but whenever i change it to SG161222/Realistic_Vision_V2.0 i end up with this error https://sharetext.me/ouj5qpf0sn is there any way to fix this, Thanks
@Astro9 ok so I see you're trying to use this model from huggingface directly: https://huggingface.co/SG161222/Realistic_Vision_V2.0/tree/main
it tries to fetch this file https://huggingface.co/SG161222/Realistic_Vision_V2.0/resolve/fp16/tokenizer/vocab.json but it does not exist
not sure why it tries to check for it, because that file is absent in my local version as well, could you show what are the other parameters in your training script file?
btw, since it downloads the model from huggingface anyway, perhaps you could use the one I provide in the "base model as diffusers"? put it on google drive and provide the absolute path to it ("/home/whatever/.../your_folder_with_the_model")
@malcolmrey this were parameters i used https://pasteboard.co/tKmniLKI46u3.png, i couldn't add
export MODEL_NAME="/home/malcolmrey/models/realistic2" export LD_LIBRARY_PATH=/usr/lib/wsl/lib:$LD_LIBRARY_PATH export CLASS_DIR="/home/malcolmrey/dreambooth/data/person" export INSTANCE_DIR="/home/malcolmrey/github/data/morgan" export OUTPUT_DIR="/home/malcolmrey/dreambooth/models/morgan"
because the upper cell already made those directories
@Astro9 no no, do not copy it directly, the WSL is only for WSL, and the /home/malcolmrey is my path, you need to provide yours
in the google colab you can either put some values in the input fields or switch to the advanced view where you can manually edit it what you want
can you make a screen where you put the SG161222/Realistic_Vision_V2.0 ?
@malcolmrey yea i know i puted my directories path, https://pasteboard.co/VPCDryCquW7K.png
@malcolmrey sorry i'm kinda noob and asking many question, base model as diffuser, i couldn't find those and i don't know what's WSL directory, there's nothing like that in colab
@malcolmrey here's i even tried downloading the realistic model ckpt https://pasteboard.co/IonPRoU1mFNk.png but i get same error some kind of tekonizer https://pasteboard.co/KhnLSDwTGBw7.png - https://pasteboard.co/dU9zxg2xnMBy.png
Untitled 2023-06-24 11:07:34 - Share Text
@Astro9 no worries!
so first things first, here is the base model in diffusers format: https://civitai.com/models/45539?modelVersionId=81468
"i don't know what's WSL directory, there's nothing like that in colab"
Indeed, you do not need to worry about WSL, it is only if you want to run it on Windows
@Astro9 okay so looking at your config there are two things,
first - you can try with the SG161222/Realistic_Vision_V2.0 again BUT remove those two lines that have the 'fp16' - it is probably why you may have that error
so fingers crossed that it works,
BUT if not, then the other issue is in: pretrained_model_name_or_path,
you can put there 'SG161222/Realistic_Vision_V2.0' and see if it works, but if you want to use my model locally then you cannot use the ckpt file, you need the model that i linked in a comment above (the base models in diffusers), you need to download that into colab, unpack it and point to that directory
@malcolmrey Thank u so much, i'll try that, sorry for irritating so much
@Astro9 no reason to be sorry :) good luck!
John Oliver is the type of person who would mention this on his show. Just sayin' ;-)
I hope he will :)
Bro make a model for Hailee steinfeld (current look)
added to the list :)
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.