Configure your edit — paste command into Supercomputer chat to run
First time? Paste this into your Supercomputer chat first
Copy and send to Supercomputer before using commands — only needed once per session.
## Higgsfield Video Editor V9 — Command Protocol
--- BRIDGE ---
run bridge
https://higgsfield.ai/asset/video/CLIP1-JOB-ID
https://higgsfield.ai/asset/video/CLIP2-JOB-ID
duration:<s> resolution:<720p|1080p> aspect_ratio:auto audio:<on|off> [librosa:on] [trim1_start:X.XXX] [trim1_end:X.XXX] [trim2_start:X.XXX] [trim2_end:X.XXX]
prompt:<text>
Pipeline:
1. higgsfield_job_status both IDs → CDN URLs
2. Download both videos
3. If trim flags: ffmpeg -ss trim_start -to trim_end -i input.mp4 -c copy trimmed.mp4 (before anything else)
4. ffprobe aspect ratio Clip1 → map to seedance ratio (16:9/9:16/1:1/4:3/3:4/21:9)
5. Extract last frame Clip1 + first frame Clip2
6. If audio:on → extract last 1s audio Clip1 + first 1s audio Clip2 as WAV
7. If librosa:on → BPM+key both tails → auto-append tempo direction to prompt
8. Upload frames + audio
9. Generate seedance_2_0: duration, resolution, aspect_ratio, generate_audio, medias=[last_frame,first_frame], input_audio=audio1
10. Lossless concat Clip1+Bridge+Clip2 → upload → return URL
--- LOOP ---
run loop
https://higgsfield.ai/asset/video/JOB-ID
duration:SECONDS resolution:720p|1080p aspect_ratio:auto audio:on|off [librosa:on] [trim_start:X.XXX] [trim_end:X.XXX]
prompt:TEXT
Pipeline: get URL → download → trim if flags → aspect ratio → extract LAST frame (start) + FIRST frame (end) → audio → librosa → generate bridge (last→first) → stitch original+bridge → upload → URL
--- EXTEND ---
run extend
https://higgsfield.ai/asset/video/JOB-ID
duration:SECONDS resolution:720p|1080p aspect_ratio:auto audio:on|off [librosa:on] [trim_start:X.XXX] [trim_end:X.XXX]
prompt:TEXT
Pipeline: get URL → download → trim if flags → aspect ratio → last frame → audio → librosa → generate extension → stitch original+extension → upload → URL
--- STITCH ---
run stitch
https://higgsfield.ai/asset/video/JOB-ID [trim_start:X.XXX] [trim_end:X.XXX]
https://higgsfield.ai/asset/video/JOB-ID [trim_start:X.XXX] [trim_end:X.XXX]
Pipeline: get all CDN URLs → download → trim each if flags → lossless concat → upload → URL
RULES:
- Trim runs BEFORE frame extraction: ffmpeg -ss -to -i in.mp4 -c copy out.mp4
- aspect_ratio:auto = ffprobe Clip1, map to nearest seedance ratio
- seedance_2_0 min 4s duration, min 720p (map 480p→720p)
- librosa:on only when clips contain music
- Reuse cached /tmp/ downloads when available
How to use
1First time — copy setup block and paste into Supercomputer chat
2Paste your Higgsfield video URL(s) into the clip slots
3Click elements to insert @Name into the prompt, or type directly
4Adjust settings, then Generate → Copy → paste into Supercomputer chat
Generates a Seedance 2.0 bridge between two clips — extracts last frame + audio tail of Clip 1 and first frame + audio head of Clip 2. Aspect ratio auto-detected from Clip 1.
