VAE

? IntroductionKolors is a large-scale text-to-image generation model based on latent diffusion, developed by the Kuaishou Kolors team. Trained on billions of text-image pairs, ...

FLUX.1 [dev] is a 12 billion parameter rectified flow transformer capable of generating images from text descriptions. For more information, please read our blog post.Key Featu...

FLUX.1 [schnell] is a 12 billion parameter rectified flow transformer capable of generating images from text descriptions. For more information, please read our blog post.Key F...

Made UsaChoMix more polite. (This is without the graffiti style.) Personally, I think I like UsaChoMix better. 描述: 训练词语: ...

Negative prompt: BadDream, SimpleNegativeV2, ng_deepnegative_v1_75t, badv4, bad-hands-5, deformityv6, Low resolution, low-res, worst quality, low quality, average quality, cens...

描述: 训练词语: 名称: autoencoderFixKlF8_v1.pt ...

These are experimental VAE's oriented to generate Anime, maybe they won't be the best but let's see... 描述: First iteration, seems working well.Apparently in...

This Vae is based on the original 8400000 model. To dramatically improve the overall quality of image generatioin.Just place in Vae folder. C:\INSTALL-FOLDER\stable-diffusion-w...

SplatterPUNK PDXL Is a side mix of Virtual Diffusion with FAR too many added Cyberpunk + NEON Loras. We don't RECALL much of them, but we've attempted our best to keep METADATA...

SDXL VAE 描述: 训练词语: 名称: myslrVAE_v10.safetens...

The downloadable binary is a safetensors file for SD1.5, and it's compatible with the SD1.5 license; it's a 16ch VAE Decoder that you already know from SD3.Since the 16ch VAE i...

Model Overview:The Neuro-Swarm: This model was created because my Neuro-Evolution moved too close to reality. I needed something to move it back to the digital realm. I made it...

Merged one of my favorite SD1.5 models Av4/CSR with DarkSushi around a year ago.On it's own it's nothing super special but if you cram it with even more lora's it can produce s...

Latest best VAEexcluded NAI leak(animevae)best of best than past all-things(vae, except for apricots series)soft edges(eliminate roughness, similar fake denoize)calm color with...

Setting Info (USE SD UI PARSING):VAE: Blessed2DPM++ SDE KARRASSteps 40CFG 6Upscaler SwinIR, 4x UltraSharp or Latent (for low steps, don't use Latent if Steps 40)Clip Skip 2Test...

Type H: Hentai Anime StyleThis model is built on ZenTypeM with enhanced hentai attributes and produces very high quality images. No detail or quality enhancing loras needed to ...

チビっ子大好き可愛いモデル,約100枚の画像を学習させました,使われるだけで光栄！感謝！感謝！A cute model who loves little kids. Approximately 100 images were trained. It's an hono...

Type K: Kawaii Anime StyleThis model has a very cute anime style and produces very high quality images. No detail or quality enhancing loras needed to achieve high quality imag...

Use CPU as RNG in Automatic1111/Forge or just ComfyUI. ComfyUI is best choice. All the img in preview are created in ComfyUI.Model Overview:The Neuro-Evolution:I've been dedica...

SharpSpectrumVAE but XL WOO!Someone (cough FallenIncursio cough) asked if I could make it for SDXL and I did.Since the original SDXL VAE is already pretty good, you won't see m...