The dataset is mainly from sakuemonq.
It's a Pony Diffusion V6 Based model.
You don't need to use "pixel art" in the prompt.
This lora almost generates only 8x larger pixels, so you can use sd-webui-pixelart (in Auto1111's Webui ) for 8x Downscale, to get a more pixel-perfect style.
Like this:
to this:
If generating NSFW, you may need to put monochrome in the negative prompts to keep results colorful.
Hope you enjoy my model!
Description
FAQ
Comments (5)
images in description broken
This works well with palletize, which I used in my image post. However, I notice that it seems to be cutting a lot of the legs off on about half the pictures I generate - tried removing all other LORAs as well to confirm it was a result of this one.
I know this is older but obviously if adding "cropped" or "letterboxed" to your negatives doesn't help, I've also used "Cowboy Shot" which is the shot that is almost full body but cuts off at the thigh.
Any chance of making a Illustrious version of this lora?
Honestly, even though this pixel LoRA is based on the Pony, it surpasses even the more advanced Illustrious LoRAs—this one truly stands out as the best.
I do hope the creator will eventually port this LoRA into a proper IL version, but honestly, even as it is now, it’s more than good enough.
Trust me—this is the best.
I've tried many different pixel LoRAs, and nothing else compares.
Details
Files
Available On (3 platforms)
Same model published on other platforms. May have additional downloads or version variants.