Would you like some whipped cream on top?
Yes, the answer is always yes. This LoRA attempts to put a splash of white on anything.
Use the trigger word: "WhippedCreamTopStyle" along with the LoRA to use it.
For some extra flavor, you can use the trigger word: "WhippedCreamTopStyle NSFW" if you got the NSFW version.
Most of these were accidentally horny images. Just in case you were wondering what's up with the fish...
It is a bit overtrained, so it's recommended to use weights between 0.2 - 0.6.
If you use a full 1.0 weight, it looks almost like the edges are a bit burnt or colored, so more like meringue.
Undesired results? If you use a low strength value, things that the used model is heavily trained on and things that are commonly white will not work well. For example a car, shoes or a lamp. Try to increase the weights!
If it doesn't understand what to do with your prompt, it will likely produce cream in a metal bowl.
Like it? Feel free to do the coffee thing:
Description
Would you like some whipped cream on top?
Yes, the answer is always yes. This LoRA attempts to put a splash of white on anything.
Use the trigger word: "WhippedCreamTopStyle" along with the LoRA to use it.
For some extra flavor, you can use the trigger word: "WhippedCreamTopStyle NSFW" if you got the NSFW version.
It is a bit overtrained, so it's recommended to use weights between 0.2 - 0.6.
If you use a full 1.0 weight, it looks almost like the edges are a bit burnt or colored, so more like meringue.
Undesired results? If you use a low strength value, things that the used model is heavily trained on and things that are commonly white will not work well. For example a car, shoes or a lamp. Try to increase the weights!
If it doesn't understand what to do with your prompt, it will likely produce cream in a metal bowl.
Like it? Feel free to do the coffee thing:
FAQ
Comments (12)
The cream is
Capybara is understandable. Friends.
don't understand WOMAN and GIRL. Enemies.
We need capybaras for world peace. Isn't that right?
I love that somebody else is doing the konyconi thing, and with the same level of quality too! ❤️
I also think it's amazing you're obsessed with desserts like they are with coffee 🤣
Looks like this version had some NSFW training, e.g. "Laptop <lora:WhippedCreamTopStyle:1> whippedcreamtopstyle" yields a mostly nude girl.
The SFW version had no NSFW training data at all on it. Here's what it was trained on:
https://imgur.com/a/vkRM0Q0
I think that this concept is a bit too generic, that the checkpoint models likely has some training on it already.
I'm having any trouble getting the SFW version of this so so much as load in Automatic1111. I can't figure out what the deal is but it doesn't even show up. It looks cool though.
Hmm, I can't think of anything special about this one.
I uploaded it here in case you want to try another download source: https://temp-file.org/7Ed09YUxaddLTxK/file (link alive for 1 week).
It should be just fine. I assume you know how to use LoRAs and trigger them? If not, try copying the prompt and settings from one of the example images.
Write here if you're still getting issues from it =/
can you create a LoRA to let any character gets messy with cake?
Yes I can!
But I probably won't. It's a bunch of work.
But you can do it too :D
@mnemic thanks, but i HOPE myself to can do such thing, because i don't understand nothing about creating a lora.
@Podededecol It's not very hard actually.
https://www.youtube.com/watch?v=N4_-fB62Hwk
I think this was the first video I watched.
Steps are basically:
Gather images showing what you want to train (generate them with Bing or search for images online or take photos yourself)
Write a .txt-file with the same name as each image, describing the image
Use some good values for your training, and start the training script.
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.