【Anima】Simple T2I Workflow with Upscale, Detailers and ControlNet — v1.0

This Anima text-to-image workflow takes text, its required model files, and rgthree-comfy to generate images, then supports second-pass upscaling, face and hand detailers, and a bypassable Anima-LLLite ControlNet.

The second pass offers two routes: Ultimate SD Upscale or second-pass resampling.

The workflow includes face and hand detailers, and the Anima-LLLite ControlNet can be bypassed.

Inputs are text plus anima-preview3-base.safetensors, qwen_3_06b_base.safetensors, qwen_image_vae.safetensors, 4x_foolhardy_Remacri.pth, Furry Domain .07-000009.safetensors, and rgthree-comfy; the output is a generated image that can continue through the chosen second-pass route.

The listing identifies version v1.0, publisher SAN0, and base model Anima.

No license is stated; verify the applicable usage terms before using or sharing the workflow.

Vorschlag

The model files are anima-preview3-base.safetensors (UNet), qwen_3_06b_base.safetensors (text encoder), qwen_image_vae.safetensors (VAE), and 4x_foolhardy_Remacri.pth (upscaler).

The workflow also names Furry Domain .07-000009.safetensors (LoRA) and rgthree-comfy (node pack).

The SDU path is described for denoise below 0.4 and supports upscaling beyond 2x (4MP).

For Hires Upscale, try 1.5x (2.25MP) with 0.65 denoise; this path is described for denoise above 0.4.

Vorschlag · nicht geprüft

High denoise can make SDU Upscale unstable; Hires Upscale beyond 2x (>4MP) is very unstable, and low denoise can sometimes create a grainy texture.

Before use, verify that the named model files match the files you intend to use.

Vorschlag · nicht geprüft