AAM XL Anime Mix
Anime Screencap Style Model
Do you like what I do? Consider supporting me on Patreon đ
żď¸ or feel free to buy me a coffee â.
A â¤ď¸, a kind comment or a review is greatly appreciated.
Join my Discord Server
This is essentially derived from AAM AnyLoRA Anime Mix, but it's based on SDXL.
It won't work with loras and embeddings for SD1.5 even if they're trained on the original AAM or AnyLoRA. "Mix" as in mix of anime styles.
I suggest you use CFG 5-7 (not higher than 8), 20-30 steps with Euler a.
Upscalers suggestions are None (bicubic upscaling in comfyui), any good GAN for anime or Latent (only if you know what you're doing).
For Turbo I suggest you use CFG 3-4 and 8 steps Euler a or 15 steps LCM.
Unlike with DreamShaper Turbo, I think base AAM XL is better than AAM XL Turbo most of the time. However the latter is MUCH faster.
Purpose of this model
Make amazing anime style artworks on its own.
Train character loras.
Use anime styles.
Generate anime art and stylized art.
It can generate hentai on its own, but you might need loras and embeddings for specific stuff.
ComfyUI Workflow: https://pastebin.com/BHCEzc6T
Finetuned over "DreamShaper Anime", which is a mix of Anime Art Diffusion, Hassaku and DreamShaper XL. Finetuned using the AAM dataset. Fair AI Public License 1.0-SD
Description
FAQ
Comments (48)
Talk about coincidence, 2h ago I just trained my first SDXL manga character and here you come with perfect checkpoint for it, can't want to try it. Thanks you for your work :)
thank you!
No thank you, I mean it, and for the other one to; the 1.5 version :)
Helo can unlock generated AAM XL Anime Mix
I'll ask the boss
@Lykon can request artsyle lora without buy the money?
@angelike14Â I closed commissions, I don't have the time at the moment :(
@Lykon what about Hiroyuki Kitazume artsyle?
@angelike14Â not gonna touch artists, sorry
the model is up for generation
@Lykon thank
@Lykon Hey thanks for not train on artist. But are you sure about that? I can generate artwork similar to hews when i type hews in prompt, "by hews, 1 girl, solo, school uniform"
@angelike14Â send dataset on wetrasnfer ill do it
@airindo by not touching artist Lykon meant to not make a specific mini model (called Lora here) on people or artist in particular. altho he have done that in the past for science (sam does art) this model does not target anyone in particular.
However, due to the images collected in the process, some would have "hews" as common descritive tag. when it happen, the "likeness" of each picture under "hews" description get trained and you would be able to get some aspect of his style. it is a side effect and one that is core to any Stable diffusion model.
train them but don't tag artist's name so you don't sacrifice model quality
@qoop272215Â just send with wetransfer no zip file i wont touch that
This model is just amazing. best XL model ever.
you're amazing!
did you use animagineXL 3 as your model for merge ?
THIS, Glad I'm not the only one who think this is a merge of caliostro's model
hassaku mixed with anime art xl, dreamshaper xl -> this became an intermediate model called "dreamshaper xl anime". I finetuned over it.
hassaku is based on animagine (I didn't notice at first), so that's why only hassaku is cited in the model description
so yes, but indirectly and before finetuning over it.
im very new to this, everything is so confusing, i cant understand anyting, i miss the free midjourney when you would just put the prompt and get good images
@SantidomKÂ you miss paying a multi million dollar company?
most gans work fine on anime. For screencap-like stuff I suggest you use None in auto1111 / bicubic in comfyUi
is there any plan to make a diffuser format from this model?, so ppl can train a lora with this model with free tier colab, it would be great.
diffuser converter program I use Colab after conversion using the converter program. You can find the converter program by doing a google search
I'll make it on my HF today
@Lykon thankyou
@adc25a137678477 i found one using colab, but i don't wan to pay their subs, better to sub runpod instead.
Will there be a dedicated inpainting model at some point?
Is that even a thing with SDXL?
Solutions for inpainting will come out for XL.
Currently only FooocusV2 has good inpainting. But you can inpaint with base model, works fine but not pixel perfect. Enough to fix faces etc. I found out that you can "fix" tiling with giving image couple iterations in img2img with low denoise, but it can also ruin detail
hello ! , can we add the anime character without lora like on Animagine XL v3 ? for example adding yae miko from genshin or asuna from blue archive without using lora. thanks for the models by the way. im new trying SDXL đ
I've uploaded a workflow with suggested settings and eye/hands detailing: https://pastebin.com/BHCEzc6T
Thanks for the workflow, awesome !
How to use it?
@kuroi_mato_o comfyui, just save it and drag and drop it. Youi might need some community nodes, get comfyui manager to download them automatically.
Thanks Lykon
Masterpiece.
This and Animagine (yes I know they're kind of related) are my go-to XL anime models.
Well, they're totally different. I made a comparison a couple of weeks ago. This is made by mixing Hassaku and DreamShaper, then finetuned on top of it. Not sure how much Animagine has been used in Hassaku, but with the finetuning on top, this doesn't even resemble Hassaku anymore.
I love this checkpoint.
Details
Files
aamXLAnimeMix_v10.safetensors
Mirrors
aamXLAnimeMix_v10.safetensors
aamXLAnimeMix_v10.safetensors
aamXLAnimeMix_v10.safetensors
aamXLAnimeMix_v10.safetensors
AAM.safetensors
aamXLAnimeMix_v10.safetensors
aamXLanimemixV_10.safetensors
AAM_XL_Anime_Mix.safetensors
AAM_XL_Anime_Mix.safetensors
AAM_XL_Anime_Mix.safetensors
aamXLAnimeMix_v10.safetensors
aamXLAnimeMix_v10.safetensors
aamXLAnimeMix_v10.safetensors
aamXLAnimeMix_v10.safetensors
aamXLAnimeMix_v10.safetensors
aamXLAnimeMix_v10.safetensors
ăSDXLăaamXLAnimeMix_v10.safetensors
aamXLAnimeMix_v10.safetensors
aamXLAnimeMix_v10.safetensors
aamXLAnimeMix_v10.safetensors
Available On (2 platforms)
Same model published on other platforms. May have additional downloads or version variants.