Anima Aesthetic 1.1 in ComfyUI: 10 Workflows from Text-to-Image to Tiled Upscale, ControlNet and Face/Hand Repair
Anima Aesthetic 1.1 is an anime-oriented image model. This episode ships ten ComfyUI graphs around it: two baselines, two ControlNet routes, two face/hand repair routes, a 4-step speed run, a three-stage render, a tiled upscale and an Anima-Edit LoRA test bench.
Every number below is read from the saved JSON files. No timing or quality benchmark was run for this release.
The ten workflows at a glance
| # | Workflow | Nodes |
|---|---|---|
| 01 | 1.5x Upscale | 20 |
| 02 | ControlNet Precision Control | 32 |
| 03 | DMDX 4-Step Speed Run | 19 |
| 04 | Three-Stage High-Quality Render | 26 |
| 05 | Tiled Upscale | 45 |
| 06 | Image-to-Image + ControlNet + Face/Hand Repair | 52 |
| 07 | Image-to-Image | 19 |
| 08 | Text-to-Image + ControlNet + Face/Hand Repair | 60 |
| 09 | Text-to-Image Baseline | 17 |
| 10 | Anima-Edit LoRA and IC Method Tests | 52 |
Workflow details
01. 1.5x Upscale
Two ClownsharK Beta passes: 18 steps at CFG 4.5 (er_sde, simple) then 34 steps at CFG 4.5; LatentUpscaleBy between passes; Qwen-Image HDR VAE (qwen_image_HDR_vae_fp32_comfy) and qwen_3_06b_base text encoder.
Run it on RunningHub (invite code rh-v1111): https://www.runninghub.ai/post/2078374859184365569/?inviteCode=rh-v1111
02. ControlNet Precision Control
anima-base-202605-v1.0 UNET with anima-highres-aesthetic-boost and anima-turbo-lora-v0.2; euler sampler, 30 steps, two passes with denoise 1.0 then 0.55; pose/depth/edge/line-art control branches in one graph.
Run it on RunningHub (invite code rh-v1111): https://www.runninghub.ai/post/2078701779059036161/?inviteCode=rh-v1111
03. DMDX 4-Step Speed Run
anima-base-202605-v1.0 + anima-turbo-lora-v0.2 + dmdx_student_final_comfy, ModelSamplingAuraFlow, 4 steps at CFG 1 with er_sde.
Run it on RunningHub (invite code rh-v1111): https://www.runninghub.ai/post/2078701772113272834/?inviteCode=rh-v1111
04. Three-Stage High-Quality Render
Three KSampler stages with LatentUpscaleBy between them: 34 steps at CFG 3, then 18 steps at CFG 1.8 with denoise 0.35.
Run it on RunningHub (invite code rh-v1111): https://www.runninghub.ai/post/2078374835163582465/?inviteCode=rh-v1111
05. Tiled Upscale
TTP_Image_Tile_Batch / TTP_Image_Assemble tiling, 4x_NMKD-Siax_200k upscale model, anima-base + turbo LoRA, plus a Gundam_RX78_Flux LoRA at 0.85 in the sample graph; 20 steps at CFG 8.
Run it on RunningHub (invite code rh-v1111): https://www.runninghub.ai/post/2078701745957588993/?inviteCode=rh-v1111
06. Image-to-Image + ControlNet + Face/Hand Repair
waiANIMA_v10 and anima-base UNETs, six UltralyticsDetectorProvider nodes, four FaceDetailer passes, SAMLoader; 30 steps euler at denoise 0.84.
Run it on RunningHub (invite code rh-v1111): https://www.runninghub.ai/post/2078701753071136769/?inviteCode=rh-v1111
07. Image-to-Image
Qwen3VLProcessor captioning, image_scale_pixel_v2 resize, 31 steps er_sde at denoise 0.7 with the turbo LoRA.
Run it on RunningHub (invite code rh-v1111): https://www.runninghub.ai/post/2078374866838974465/?inviteCode=rh-v1111
08. Text-to-Image + ControlNet + Face/Hand Repair
60 nodes: four VAEDecode, four FaceDetailer passes, two LatentUpscaleBy stages, 30 steps euler at CFG 3 then 2.88 with denoise 0.55.
Run it on RunningHub (invite code rh-v1111): https://www.runninghub.ai/post/2078701786382290946/?inviteCode=rh-v1111
09. Text-to-Image Baseline
17 nodes: UNETLoader, qwen_3_06b_base CLIP, Qwen-Image HDR VAE, turbo LoRA, 30 steps er_sde at CFG 4.
Run it on RunningHub (invite code rh-v1111): https://www.runninghub.ai/post/2078701764664184833/?inviteCode=rh-v1111
10. Anima-Edit LoRA and IC Method Tests
Three KSampler passes at 20 steps dpmpp_2m_sde_gpu CFG 5; lora_edit_ZeroTwo and AnimaEditV1 LoRAs; AddLabel nodes for side-by-side comparison.
Run it on RunningHub (invite code rh-v1111): https://www.runninghub.ai/post/2078374813801996290/?inviteCode=rh-v1111
Promo and episode links
- RunningHub with creator invite code
rh-v1111: https://www.runninghub.ai/?inviteCode=rh-v1111 - Creator workflow page: https://www.runninghub.ai/user-center/1846601088051884033/webapp?inviteCode=rh-v1111
- Video tutorial (YouTube): https://www.youtube.com/watch?v=ShiAjBa8EXc
- Chinese original (Bilibili): https://www.bilibili.com/video/BV1xJKB6vEfp
Limits
- Model files, LoRAs and detector weights are not bundled; every JSON references files that must exist in your environment.
- Third-party LoRAs keep their own licenses.
- Results depend on prompt, seed, sampler settings and the base model version.
Collections and mirrors
- Hugging Face collection (10 workflows + English covers): https://huggingface.co/AI-KSK/anima-aesthetic-1-1-workflows-279
- ModelScope collection (中文说明): https://modelscope.cn/models/Sunjian520/anima-aesthetic-1-1-workflows-279
- Civitai items (one per workflow): 01 · 02 · 03 · 04 · 05 · 06 · 07 · 08 · 09 · 10
- Pinterest pins (one per workflow): 01 · 02 · 03 · 04 · 05 · 06 · 07 · 08 · 09 · 10
All workflow entries above use the verified RunningHub international pages with invite code rh-v1111.
Top comments (0)