RDBT [Anima]
This page contains the "original LoRA files" for advanced users.
See the main page for more info. (ckpt version, usage, faq, int8 ckpt...)
See this page for update log and version info.
Sharing merges using this model is not allowed. This "restriction" won't affect anyone. It's only aimed at those who steal others' models to sell.
Description
FAQ
Comments (7)
Sorry, I don't quite understand this LoRA. Is it the same as a Turbo LoRA, or is it more of a Style LoRA?
I noticed you compared it to Turbo, but you also mentioned style, so I'm a bit confused. I'm still new to this, so I'm not very familiar with the differences yet.
_ Is it the same as a Turbo LoRA
If you tried the official anime turbo LoRA you will notice it kills diversity and styles. That's the price you must pay if you want to generate an image in just 4 steps. There is no free lunch.
Think of the RDBT as a less aggressive turbo LoRA, which needs 16 steps but you get the diversity and styles back. And finetuned for better aesthetic.
It's not a style LoRA though, it doesn't give you a fixed style.
@reakaakasky I see. Thank you for clarifying.
@reakaakasky why not specify this in the description? Also, how would people ever find this model given the random acronym name? What's the point of putting effort into training a model if you're gonna make it so difficult for ppl to find, offer little to no description of what it does, or how to use it? Sorry if this sounds bitchy, I actually really like the lora, it's just frustrating that creators are kinda clueless when it comes to presenting and marketing their stuff.
The Lora version could also be used on different checkpoint, but only trained checkpoint and not merged one. Like for example I'm using Raehoshi for its newer character dataset but I don't like its base style and the checkpoint is very stiff on that regard. RDBT lora helped a lot to make that checkpoint more responsive to style prompts.
I don't understand. Should these LoRAs be paired with the base Anima, or with your checkpoint? If so, I don't understand which goes which.
Lora gets paired with anima base. The full checkpoint he also offers is just a merge of this lora with anima base 1.0 for simplicity.