Nodes/ComfyUI Beeble AI/Beeble SwitchX
ComfyUI Node

Beeble SwitchX

Run a Beeble SwitchX compositing job: re-generate everything outside a masked subject while preserving the original pixels inside it, and relight the subject to match. Wire a 'source' (image or video frame batch), an optional 'reference_image' for look guidance, and (for select / custom alpha modes) an 'alpha' mask. Set generation_type to 'image' or 'video', pick an alpha_mode, write a prompt, and queue. The node uploads inputs, submits the job, polls until done with a live progress bar, then returns the rendered output + alpha as IMAGE tensors. Note: at least one of prompt or reference_image must be provided. Beeble bills per generation in USD; check the Beeble Account Info node to monitor spend.

By am-pipeline-prod·Created 3 months ago·Updated 3 months ago· 3
Beeble SwitchX
  • source
  • reference_image
  • alpha
  • source_video
  • alpha_video
  • output
  • output_alpha
  • job_id
  • render_url
  • raw_json
  • video
generation_typevideo
alpha_modeauto
max_resolution1080
fps24.0
prompt
source_uri
reference_image_uri
alpha_uri
poll_interval_seconds15
poll_timeout_seconds1800
CategoryBeeble AI

Inputs (15)

NameTypeDefaultDescription
generation_typeCOMBOvideo'image': single still in, single still out. 'video': multi-frame in, multi-frame out (max 240 frames). The wired source must match this setting.
alpha_modeCOMBOautoauto: API auto-detects the foreground subject. select: Provide first-frame alpha; API propagates. fill: Keep everything as-is, no masking. custom: Provide a full per-frame alpha sequence.
max_resolutionCOMBO1080Output resolution cap (longest side). 1080 is the API default; 720 is faster and cheaper.
fpsFLOAT24.01–240FPS used when encoding a wired source/alpha frame batch as MP4 for upload (video mode only). Ignored when generation_type='image' or when you supply URIs directly.
promptSTRINGText description guiding the new look. Recommended even when a reference image is provided. Either prompt or reference_image (or both) is required by the API.
sourceoptIMAGESource media. Single image for image jobs, multi-frame batch (e.g. from Load Video) for video jobs. Overrides source_uri when wired.
reference_imageoptIMAGESingle still image used as a style/look reference. The Beeble API hardcodes this to one image -- no array, no per-frame targeting. Overrides reference_image_uri when wired.
alphaoptIMAGEAlpha mask. Single image for image jobs and 'select' mode (first-frame alpha). Multi-frame batch for 'custom' mode in video jobs. Overrides alpha_uri when wired.
source_urioptSTRINGbeeble:// or https:// URL. Used only when 'source' is not wired.
reference_image_urioptSTRINGbeeble:// or https:// URL. Used only when 'reference_image' is not wired.
alpha_urioptSTRINGbeeble:// or https:// URL. Used only when 'alpha' is not wired.
poll_interval_secondsoptFLOAT155–120Seconds between status polls. Read endpoints are capped at 5 RPM per account, so 12+ s is safe.
poll_timeout_secondsoptFLOAT180060–86400Maximum wall-clock time to wait per job (seconds).
source_videooptVIDEOOptional VIDEO source. Forces generation_type = video, uploads the underlying file directly (no IMAGE re-encode). Takes precedence over `source` and `source_uri`. Wire from Topaz, AM Read Video → VIDEO output, or a lazy AM transform chain.
alpha_videooptVIDEOOptional VIDEO alpha (per-frame mask sequence). Same upload semantics as source_video. Takes precedence over `alpha` and `alpha_uri`. Use with alpha_mode='custom' for per-frame masks.

Outputs (6)

NameTypeDescription
outputIMAGE
output_alphaIMAGE
job_idSTRING
render_urlSTRING
raw_jsonSTRING
videoVIDEO