Watch the full video first if you want to understand how this SCAIL-2 single-person biological long-video driving workflow works in practice. The video shows how one reference character can follow the movement of a single-person driving video, while the workflow keeps identity, pose flow, limb stability, silhouette clarity, and long-video continuity more consistent.
This ComfyUI workflow is designed for SCAIL-2 single-person biological motion driving. Its main purpose is to animate one reference character by using a single-person driving video as the motion source. This is not a local replacement workflow. It does not aim to replace only a small masked region in the original video. Instead, it uses skeleton guidance and reference identity control to generate a new character video based on the movement of the driving subject.
The workflow is built around wan2.1_14B_SCAIL_2_fp8_scaled.safetensors as the main SCAIL-2 model. It also uses WAN VAE, UMT5 WAN text encoding, CLIP Vision, SAM3 tracking, SCAIL2ColoredMask, WanSCAILToVideo, SamplerCustom, VAEDecode, ForLoop continuation, overlap-frame trimming, ColorTransfer, final video combining, and original audio restoration. A multi-LoRA enhancement chain is also included, using modules such as LightX2V, WanAnimate relight, Wan2.2 Lightning I2V, FastWan 480p, Wan21 PusaV1, Wan2.2 Fun InP, and stage-based enhancement LoRAs.
The key setting in this workflow is replacement_mode=false. This means the workflow focuses on single-person skeleton-guided animation rather than direct person replacement. The reference image provides the target character identity, while the driving video provides the pose, timing, body movement, and rhythm. The positive prompt is kept simple and low-interference, focusing on a full-body single character following one-person pose guidance, natural motion, stable identity, stable limbs, clean silhouette, coherent video, and smooth temporal consistency.
The workflow uses a strict 512×896 alignment rule. Both the reference image and the driving video are resized to the same canvas before entering SAM3, CLIPVision, and SCAIL. This reduces pose mismatch, mask instability, identity drift, and body deformation caused by inconsistent input sizes.
SAM3 is configured with max_objects=1. SCAIL2ColoredMask uses object_indices=0 and sort_by=area, which tells the workflow to focus on the main tracked person in the driving video. CLIP Vision encodes the reference image, helping the generated character keep a more stable identity across the sequence.
The long-video structure is one of the main strengths of this workflow. The first segment is 65 frames and establishes the character identity, pose relationship, mask structure, and motion direction. The continuation segment is 81 frames. Each continuation loop removes 5 overlapping frames, so every loop effectively adds 76 new frames. The loop count is calculated as max(1, ceil((F - 65) / 76)), where F is the loaded frame count of the driving video.
The workflow also includes segment-level continuity processing. After each continuation segment is decoded, the first 5 overlapping frames are removed. The previous segment’s final frame is used as a reference for ColorTransfer, helping the next segment match the tone and reduce visible transitions. The generated frames are then accumulated into the final frame sequence. The final video combines the generated images with the original driving video audio and a unified 24fps frame rate.
Main features:
SCAIL-2 single-person biological long-video workflow
One reference character follows one driving video
Single-person full-body skeleton guidance
replacement_mode=false for motion driving
512×896 unified input alignment
SAM3 max_objects=1 subject tracking
SCAIL2ColoredMask single-subject mask control
object_indices=0 and sort_by=area
CLIP Vision reference identity encoding
WanSCAILToVideo first-segment generation
65-frame initial segment
81-frame continuation segment
5-frame overlap trimming
ForLoop long-video continuation
ColorTransfer segment consistency
Original driving video audio restored
Unified 24fps output control
Multi-LoRA enhancement chain
Suggested workflow:
Prepare one clear reference character image and one clean single-person driving video. The reference image should show the face, outfit, full-body structure, and silhouette clearly. The driving video should contain one main subject with readable movement, stable framing, and limited occlusion. Start with the default 512×896 setting. Check that SAM3 tracks only one person correctly. If the character identity drifts, use a cleaner reference image and simplify the prompt. If the motion becomes unstable, use a driving video with less camera shake and fewer extreme occlusions. Run a short test first, then use the long-video loop after the base motion and identity direction are stable.
⚙️ RunningHub Workflow
Try the workflow online right now — no installation required.
👉 Workflow: https://www.runninghub.ai/post/2067141217971953665?inviteCode=rh-v1111
If the results meet your expectations, you can later deploy it locally for customization.
🎁 Fan Benefits: Register to get 1000 points + daily login 100 points — enjoy 4090 performance and 48 GB super power!
📺 Bilibili Updates (Mainland China & Asia-Pacific)
If you’re in the Asia-Pacific region, you can watch the video below to see the workflow demonstration and creative breakdown.
📺 Bilibili Video: https://www.bilibili.com/video/BV1jWL96nEpw/
☕ Support Me on Ko-fi
If you find my content helpful and want to support future creations, you can buy me a coffee ☕.
Every bit of support helps me keep creating — just like a spark that can ignite a blazing flame.
👉 Ko-fi: https://ko-fi.com/aiksk
💼 Business Contact
For collaboration or inquiries, please contact aiksk95 on WeChat.
⚙️打开下方链接即可在线体验，无需安装。
👉 工作流： https://www.runninghub.ai/post/2067141217971953665?inviteCode=rh-v1111
如果觉得效果理想，你也可以在本地进行自定义部署。
🎁 粉丝福利： 注册即送 1000 积分，每日登录 100 积分，畅玩 4090 体验 48 G 超级性能！
📺 Bilibili 更新（中国大陆及南亚太地区）
如果你在中国大陆或南亚太地区，可以通过下方视频查看该工作流的实测效果与构思讲解。
📺 B站视频： https://www.bilibili.com/video/BV1jWL96nEpw/
我会在 夸克网盘 持续更新模型资源：
👉 https://pan.quark.cn/s/20c6f6f8d87b
这些资源主要面向本地用户，方便进行创作与学习。