Motion control workflow: 84% lower cost than premium video AI

The business problem

A digital content agency needed to produce video at industrial scale: hundreds and thousands of clips a month. They were looking at premium video AI services (Kling 2.6 and similar) for motion control video generation, where the movements of a source video are transferred onto a target character.

The economics were harsh:

• Premium services charge $0.21–$1.20 per generation for motion control (3.5–20 credits at ~$0.06–0.08 per credit)
• At their volume (1000+ videos a month, with a target of 100 videos an hour during production sprints) this came to thousands of dollars a month for AI generation alone
• Credit limits kept them from scaling to real output needs
• Premium services' content policies restricted what could be generated at all

They needed industrial-scale video generation that was radically cheaper and operationally flexible.

What made it hard

Motion control is not trivial to reproduce. The technology requires:

• Skeleton and pose detection from the source video
• Character segmentation that works precisely with complex movement
• Motion transfer that preserves both the action and visual coherence
• Background and context handling so the result looks natural

Most premium services (Kling, Hailuo, RunwayML) made motion control a proprietary feature and charge accordingly. Open-source analogs existed, but they were either broken, hard to find, or required deep ComfyUI expertise to bring to production.

Our approach

After broad research and testing, we determined that Wan 2.2, an older but underrated open-source model, can match premium motion control quality with the right ComfyUI workflow architecture.

The difficulty: the available workflows were either broken or required manual segmentation (marking by hand where the character is in every frame, completely impractical at scale).

Iteration 1

We got a broken workflow stuffed with unclear models and unused LoRAs. We cut it down to the working components, but segmentation still needed manual frame-by-frame marking. At production scale, not workable.

Iteration 2

After more research we found a better workflow with automatic segmentation models. We customized it and stabilized it for production. This became the production version.

Ongoing improvements

• Built in a video upscaling sub-workflow to raise quality
• Added frame interpolation (smooth output, 30fps → 60fps)
• Built around the RunningHub API with parallel processing across several keys
• Handled edge cases (mismatches with objects between the source motion and the target character)

Production architecture

• ComfyUI workflow on RunningHub GPU compute
• An RTX 5080-class GPU is enough for the load (premium hardware isn't needed)
• 5 parallel jobs per API key, with a multi-key scheme to scale past a single account's limits
• Generation time: ~20 minutes of compute per video on the standard plan
• Built into a broader content pipeline (as a module inside a large automated content production system)
• Available through several interfaces: Telegram bots, a web interface, or ComfyUI directly for advanced users

Capability comparison: not just cheaper, different capabilities

Beyond cost, premium services have hard technical limits that restrict commercial use:

Premium service limits (Kling 2.6 Motion Control)

• A maximum of 30 seconds per continuous generation
• Credit consumption grows with duration (longer = exponentially more expensive)
• Content policy restrictions on a range of commercial scenarios

Our implementation

• No hard duration limit: video length is bounded only by available GPU time
• You can generate videos of 1, 2, 10+ minutes in a single continuous generation
• The same cost-per-second economics, scaling linearly with duration
• No content policy friction for legitimate commercial work

Cost engineering: the math

RunningHub pricing structure

• $0.0004 per coin
• 24 coins per minute of GPU time
• ~$0.01 per minute of compute

Cost per video for a typical 30-second clip

20 minutes of GPU time per video → 480 coins → ~$0.19 per video

Comparison with Kling 2.6 motion control (the same 30-second video)

15–20 credits per generation × $0.06–0.08 per credit → ~$0.90–$1.60 per video (midpoint ~$1.20)

At the client's real production volume

The lower cost per video is the headline, but the total value comes from three factors that reinforce each other: 84% lower cost, no duration limits, which opens content formats competitors can't make, and operational flexibility from parallel processing across several keys.

Quality comparison

The honest answer: quality is on par with Kling for the production scenario, and in places better.

Where premium services win slightly: edge cases with unusual objects (for example, the person in the source video holds a box and the target character doesn't have one; both systems can produce artifacts here, solved by editing the source image beforehand).

Where our implementation matches or beats them: standard motion transfer scenarios, which are 95%+ of production volume.

Both sometimes hallucinate. This is expected for the current generation of video AI; neither premium nor open source is free of hallucinations.

Expertise gained

On this project we built deep expertise in:

• ComfyUI workflow architecture, including debugging, library management, and the ComfyUI Manager ecosystem
• Open-source video model capabilities, in particular the strengths and weaknesses of Wan 2.2 (excellent for motion transfer, weaker for generation from scratch)
• GPU resource optimization: getting production quality on consumer GPUs instead of enterprise hardware
• Video post-processing integration: upscaling and frame interpolation built into the main generation workflow
• Production stabilization: dealing with the inevitable breakages when custom node maintainers change repositories, model versions go stale, and so on

Result

• 4–5 months of continuous production use at 2 commercial clients in active content production
• Industrial output: supports a target throughput of 100+ videos an hour
• Capabilities beyond premium services: no 30-second cap on video length
• An integrated foundation for a broader automated content production pipeline
• Operational flexibility: no content policy restrictions and no credit limits beyond infrastructure capacity

Tech stack

What this demonstrates

• Deep open-source AI expertise: finding, debugging, and bringing to production workflows that are undocumented and little known
• Cost arbitrage thinking: seeing when premium services charge dollars for capabilities that open source delivers for cents
• Finding capability gaps: finding business value in what premium services don't offer at all (long-form motion control)
• Production engineering: turning broken or impractical workflows into industrial-grade reliable ones
• Workflow architecture: linking several processing stages (motion control + segmentation + upscaling + interpolation) into coherent production pipelines
• GPU compute optimization: production results on consumer-tier hardware