Sonny Anime Fixed

🚀 Sonny Anime Fixed (Flux Dev)

This is a text - to - image model based on Stable Diffusion and LoRA technology. It can generate various anime - style images according to the input text prompts.

🚀 Quick Start

You can test this model through the following examples:

• Example 1: Input prompt "a chibi astronaut, nm22 minimalist style", and you can get corresponding output images.
• Example 2: Input prompt "a pink frog, nm22 minimalist style", and you can get corresponding output images.

✨ Features

• Style Adherence: This is one of two Flux LoRAs on this style, and this particular version is more fixed. It closely adheres to the anime style and doesn't perform well with many anime - style prompt additions.
• Artifact Reduction: By removing additional stylization words, the linework can be fixed, and fewer artifacts are produced when not adding too many stylizations like "anime" or "illustration".
• Specific Word Suitability: Words like "pastelcore", "lofi", and "minimalist" work well with this model.

📚 Documentation

Model description

This is one of two Flux LoRAs on this style, this particular version being more fixed. This means it doesn't do as well with a lot of anime style prompt additions, and tends to stick to anime style really closely. If you want a more flexible version, check out the Flex LoRA.

Typically if the linework seems like it's falling apart a bit, it can be fixed by removing any additional stylization words you've added. I also find it produces fewer artifacts if you don't add too many stylizations like "anime" or "illustration".

I have included some examples with and without Anime prompts to show some of the difference.

Some words I like to use with the model:

• pastelcore
• lofi
• minimalist

Warning: This model seems to hallucinate fake art signatures from time to time. This is less present in this version vs. the Flex version, and I did not include any images with that kind of artifact in them. It seems it may have more to do with the underlying model, apologies if it causes a challenge for you!

Also, a big thank you to Fal for sponsoring credits for me to do some test trainings on the platform!

Trigger words

You should use nm22 style to trigger the image generation.

Download model

Weights for this model are available in Safetensors format.

Download them in the Files & versions tab.

📄 License

This model is licensed under creativeml - openrail - m.