wan / v2.6 / fal / text-to-video

Wan 2.6 text-to-video, running closed on fal's fleet through Civitai's Orchestration API. Prompt in, native VIDEO out, billed in Buzz - the whole generation happens in the cloud as part of the official civitai-comfy-nodes pack.

Straight talk about this one: Wan 2.6 was the December 2025 release where the community concluded Alibaba was never shipping open weights again, and the model itself got a rough reception ("worse than 2.2" was a recurring verdict). You're not choosing this node for a quality leap. You're choosing it for 1080p text-to-video, multi-shot output, and background music via URL - the three things that distinguish 2.6 from the open Wan nodes in this pack.

The inputs that matter

prompt and cfg_scale (default 4) are required. Then:

• resolution - 720p or 1080p (default 1080p).
• aspect_ratio - 16:9, 9:16, 1:1, 4:3, 3:4 (default 16:9).
• multi_shots - off by default. A single run producing multiple shots. The 2.6 headline feature - and the biggest lever on the meter. Start off, budget when on.
• audio_url - URL to an audio file (3–30 seconds, up to 15 MB) for background music, supported for text-to-video and image-to-video. One of the few places this pack's video nodes can bring sound into the clip.
• enable_prompt_expansion - on by default; fal's prompt rewrite. Off for exact prompts.
• negative_prompt - steering.
• enable_safety_checker - off by default.
• loras - CIVITAI_LORAS socket; trigger words go in prompt.
• source_image exists but is optional here - for pure text-to-video leave it empty.
• duration 1–30s (default 5), frame_rate 24, steps 10–50, seed, api_config.

Outputs

Standard trio: video (VIDEO) into a Save Video / preview node, workflow_id (STRING) for sidebar tracking, raw_json (STRING) for response and cost.

Install and auth

One-time, pack-wide. ComfyUI Manager → search Civitai Comfy Nodes → install, restart:

cd ComfyUI/custom_nodes
git clone https://github.com/civitai/civitai-comfy-nodes.git
pip install -r civitai-comfy-nodes/requirements.txt   # just `requests`

Auth: Civitai Auth node, CIVITAI_API_TOKEN (headless), or sidebar sign-in.

Troubleshooting and takes

• Set expectations before you spend. 2.6's reception was hostile; it's a features play, not a quality play. If you just want good Wan video, the open 2.2 fal nodes are cheaper and better-loved.
• Audio URL must be public. Fal's pipeline fetches it - localhost and private URLs fail silently as a dead clip.
• Multi-shot = multi-bill. One toggle, bigger meter, longer queue.
• Text-to-video drift. No anchor frame means the prompt does everything; write motion and camera explicitly, or accept static scenes.
• Closed model, fixed filter. Alibaba's moderation applies server-side; no local patch exists.
• Slow / timeout. Pack launched June 2026 during Civitai's GPU crunch; 1080p jobs queue long. Default timeout 30 min - raise via Civitai Auth timeout_minutes or CIVITAI_COMFY_TIMEOUT.
• Early preview. Generated nodes "may change without notice."

Reach for this when you want Wan 2.6's multi-shot or background-music text-to-video at 1080p and understand what you're paying for. Have a subject that must stay consistent across clips? That's the reference-to-video sibling's job.