Anima: SCAIL单人三维姿态驱动 图像参考的长时序稳定生成 — 7 Workflows, Read From the Graphs

This release ships 7 workflow files built around Anima. Together they total 308 nodes. Rather than summarise the video, this write-up reads the graphs directly — what each one loads, how big its latents are, and which sampler it runs.

The workflows at a glance

Workflow Nodes Links Sampler Latent
Wan 2.2 姿态引导下的 SVI 2.0 视频到视频无限循环延长与高一致性生成工作流 116 184 — —
SCAIL单人三维姿态驱动+图像参考的长时序稳定生成 45 59 — —
kj新Pose Detection + Wan Animate 44 49 — —
kj新Pose Detection +SteadyDancer 33 45 — —
Qwen Image Edit 2511姿态迁移AnyPose 33 45 — 928x1664 1f
Wan 2.2 Fun VACE 姿态视频引导加参考 28 39 — —
HyMotion 文本生成 FBX 骨骼动作 9 8 — —

The node count range is wide: HyMotion 文本生成 FBX 骨骼动作 at 9 nodes against Wan 2.2 姿态引导下的 SVI 2.0 视频到视频无限循环延长与高一致性生 at 116. That gap is usually the difference between a single-pass preview graph and one that adds conditioning, upscale stages, or a second sampling pass.

What the graphs load

Model file Loader node
Wan22Animate/Wan2_2-Animate-14B_fp8_scaled_e4m3fn_KJ_v2.safetensors WanVideoModelLoader
Wan2_1_VAE_bf16.safetensors WanVideoVAELoader
clip_vision_h.safetensors CLIPVisionLoader
umt5-xxl-enc-bf16.safetensors WanVideoTextEncodeCached
clip_vision_vit_h.safetensors CLIPVisionLoader
Wan21_I2V_SteadyDancer_fp16.safetensors WanVideoModelLoader
qwen_2.5_vl_7b_fp8_scaled.safetensors CLIPLoader
qwen_image_edit_2511_bf16.safetensors UNETLoader
qwen_image_vae.safetensors VAELoader
https://huggingface.co/Kijai/WanVideo_comfy_fp8_scaled/blob/main/OneToAllAnimation/Wan21-OneToAllAnimation_fp8_e4m3fn_scaled_KJ.safetensors Note
SCAIL/Wan21-14B-SCAIL-preview_comfy_bf16.safetensors WanVideoModelLoader
Wan21_Uni3C_controlnet_fp16.safetensors WanVideoUni3C_ControlnetLoader
umt5_xxl_fp8_e4m3fn_scaled.safetensors CLIPLoader
wan_2.1_vae.safetensors VAELoader

LoRAs in play

LoRA Weight
lightx2v_elite_it2v_animate_face.safetensors 1
lightx2v_I2V_14B_480p_cfg_step_distill_rank128_bf16.safetensors 1
Qwen-Image-Edit-2511-Lightning-4steps-V1.0-bf16.safetensors 1
2511-AnyPose-base-000006250.safetensors 0.7
2511-AnyPose-helper-00006000.safetensors 0.7
lightx2v_T2V_14B_cfg_step_distill_v2_lora_rank128_bf16.safetensors 1
tifa_000002250_low_noise.safetensors 1
tifa_000002250_high_noise.safetensors 1
Wan2.2-Fun-A14B-InP-HIGH-MPS_resized_dynamic_avg_rank_21_bf16.safetensors 1
Wan2.2-Fun-A14B-InP-LOW-HPS2.1_resized_dynamic_avg_rank_15_bf16.safetensors 1
Wan22_PusaV1_lora_HIGH_resized_dynamic_avg_rank_98_bf16.safetensors 1.0000000000000002
SVI_Wan2.2-I2V-A14B_low_noise_lora_v2.0.safetensors 1
SVI_Wan2.2-I2V-A14B_high_noise_lora_v2.0.safetensors 1
Wan22_PusaV1_lora_LOW_resized_dynamic_avg_rank_98_bf16.safetensors 1

Not every LoRA runs at 1.0 — the non-default weights here are [0.7]. A weight above 1.0 amplifies the LoRA's effect past what its author trained for; below 1.0 backs it off. Both are deliberate tuning choices, not leftovers.

Latent geometry

Distinct latent configurations across the set: 928x1664 1f.

The frame count matters more than the pixel dimensions here. A graph at 97 frames is built for roughly a four-second unit; one at 121 frames is closer to five. Sigma schedules do not transfer cleanly between them, so treat each geometry as its own tuning target.

Where this set is weak

  1. Node count is not quality. The largest graph is not automatically the best one; a 40-node graph with a good prompt beats a 200-node graph with a bad one.
  2. No measured comparison in the files. The workflows carry settings, not benchmarks. Any quality claim needs your own A/B run.
  3. Model files are not included. You supply the weights; the graphs only reference filenames.
  4. Version drift. A graph pinned to a specific checkpoint revision will run on a newer one but may produce different output.

Practical order

Open the smallest graph first. It has the fewest moving parts and will confirm your models load before you spend time on the elaborate ones.

Workflows

7 graphs published on RunningHub with invite code rh-v1111:

  • HyMotion 文本生成 FBX 骨骼动作
  • kj新Pose Detection + Wan Animate
  • kj新Pose Detection +SteadyDancer
  • Qwen Image Edit 2511姿态迁移AnyPose
  • SCAIL单人三维姿态驱动+图像参考的长时序稳定生成
  • Wan 2.2 Fun VACE 姿态视频引导加参考
  • Wan 2.2 姿态引导下的 SVI 2.0 视频到视频无限循环延长与高一致性生成工作流

Files

https://huggingface.co/AI-KSK/ksk-video-153-comfyui-workflows

Video walkthrough: https://youtu.be/03xW3gEQr2A
Bilibili: https://www.bilibili.com/video/BV1kbqPBzEKL/

Notes

  • Workflow files only. Model weights are not included and each keeps its own licence.
  • Node counts, geometries and loader filenames above are read from the published JSON files.
  • Independent AI-KSK release. Not affiliated with the model authors or RunningHub.

Optional API platform - 88API (affiliate link)

Explore 88API for access to multiple AI model APIs. Available models, routes and prices are listed on the platform; Alipay top-ups are supported.

https://88api.ai/sign-up?aff=usx3

Disclosure: I may earn a commission from eligible purchases after you register through this referral link.