This is a LoRA trained on various images from the artist "Rick Griffin".
It has a very decent understanding of prompts and a pretty good composition if you promt it right.
I recommend using it at around 0.41 to 0.5 strength for optimal results.
Description
FAQ
Comments (3)
How did you generate outlines this good in the example images? Did you render at Full HD straight away or used some upscaler or highres fix? I could never achieve good outlines with prompting on various models, tried YiffyMix and EasyFluff.
Hi and thank you for using this LoRA!
I'm sorry to hear that your outlines didn't come out quite as neat.
I've purrsonally been using the DPM++2M SDE Karras ampler with a hires.fix with 40 steps, denoising of 0.65 and an upscale by 2.5 with the Ultramix_Smooth upscaler.
Furthermore, i have been using the Furception VAE, if that helps as well.
It's also important to use CFG Rescale with the EasyFluff model but i'm sure you knew that already.
Hopefully this'll help, otherwise feel free to ask!
@FuOx may I ask you, do you have nice outlines with Hires. Fix disabled? Unfortunately I have a very old videocard with low amount of memory and I can't use Hires. Fix because with any settings it needs more VRAM than I have. I wonder if you are getting sharp outlines without it. I also tried lowering the CFG Scale down to 7 (usually I crank it up to 10 or 12) and using the model and all LoRAs you've used in your prompt (I fed this picture to PNG Info https://civitai.com/images/10193566 ) but the outlines are still meh. It's like I'm missing something but I don't understand what exactly. What's interesting is when I decreased the <lora:fluffyrock-quality-tags-v2.1.1-vpred:1>, weight down to -0.3 the outlines got noticeably better but the overall quality has dropped significantly. Rick Griffin's LoRA suffers from this issue much more than other LoRAs you've made - usually I get good outlines with Jishinu (testing v2 as I'm writing this and it's kinda fine too)
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.





