Nodes/ComfyUI/Luma Text to Video
ComfyUI Node Runs on cloud

Luma Text to Video

The dependable prompt-to-clip node

By Comfy-Org·Created 4 years ago·Updated about an hour ago· 129,820
Luma Text to Video
  • luma_concepts
  • VIDEO
prompt
model
aspect_ratio16:9
resolution540p
duration
loopfalse
seed0

The Luma Text to Video node is the plain-Jane entry point into Luma's video models from inside ComfyUI core: type a prompt, get a video. It's the one to learn first in the Luma family because everything else (image-to-video, concepts, the Ray 3.2 line) builds on the same account, billing, and wait-the-poll-out rhythm.

It's an API node through and through. Your prompt goes to Luma's servers via your Comfy account, the generation runs in the cloud, and the node polls until it's done - Luma text-to-video averages around 105 seconds of wall time, so the node sitting on "polling" for a couple of minutes is normal, not a hang. No local model, no downloads, billed per generation to Comfy account credits.

The inputs that matter:

  • prompt - the whole creative payload. Minimum 3 characters, so it needs real content. Luma responds to vivid, motion-heavy descriptions.
  • model - ray-2, ray-flash-2, or ray-1-6. Ray 2 is the current quality line, flash is the speed tier. The old ray-1-6 is still there for matching older work - but note that the source quietly ignores resolution and duration when ray-1-6 is selected, since that model only does its one thing.
  • aspect_ratio - 1:1, 16:9, or 9:16.
  • resolution - 540p (default), 720p, 1080p, or 4k.
  • duration - 5s or 9s.
  • loop - off by default; on requests a seamless looping clip.
  • seed - the Luma stub: it only decides whether the node re-runs; actual results are nondeterministic either way.
  • luma_concepts - optional LUMA_CONCEPTS input from the Luma Concepts node, if you want explicit camera direction without gambling on prompt wording.

Single VIDEO output into Save Video or Preview Video.

The two things people trip on. First, the seed isn't a seed - don't treat it as reproducibility; Luma results are nondeterministic by design, and the input exists mainly so the node can decide when to re-run. Second, the model choice quietly disables knobs on ray-1-6: you'll set resolution and duration and they'll do nothing, which confuses a lot of people. If you want those controls to mean something, stay on ray-2.

House rules for the family: be logged into your Comfy account or the node refuses to run, and if partner nodes clutter your search, --disable-api-nodes in the launch args hides them all.

Categorypartner/video/Luma

Inputs (8)

NameTypeDefaultDescription
promptSTRINGPrompt for the video generation
modelCOMBO3 options: ray-2, ray-flash-2, ray-1-6
aspect_ratioCOMBO16:97 options: 1:1, 16:9, 9:16, 4:3, 3:4, 21:9, +1
resolutionCOMBO540p4 options: 540p, 720p, 1080p, 4k
durationCOMBO2 options: 5s, 9s
loopBOOLEANfalse
seedINT00–18446744073709550000Seed to determine if node should re-run; actual results are nondeterministic regardless of seed.
luma_conceptsoptLUMA_CONCEPTSOptional Camera Concepts to dictate camera motion via the Luma Concepts node.

Outputs (1)

NameTypeDescription
VIDEOVIDEO