![Eleven v3 — official cover. Audio tags like [whispers] [laughs] [excited] live inline in the script](https://eleven-public-cdn.elevenlabs.io/payloadcms/cc4b5qn8p7t-v3-cover.webp)
HCinema Studio 4.0Higgsfield▶Video
generate
Every other canvas sells you nodes. This one shows you what each one can do.
01Every node is a frontier model
Pick by what it is good at — the canvas shows what each one takes and makes, never a vendor logo you have to decode.
02Ports are typed
Text is gold, image ember-light, video ember, audio mint. Click a port and you only see nodes that fit. Wrong joins can't be made.
03Check flow is free
A dry run that walks the order, quotes, estimates time and lists problems with one-click fixes. Every Run checks first, silently.
04Every setting a model has
Standard holds the controls and anything priced. Advanced holds the rest — 3 settings on one model, 21 on another — with plain labels and a Reset.
The models
Every node is a frontier model.
Here is who's inside.
Only the latest version of each, always the vendor's own release. Scroll — the samples on the right are theirs, from their sites.
ByteDance Seed · VideoSeedance 2.5
One take, thirty seconds, with the sound already in it.
The reference-native video model: up to 30 images, 10 clips and 10 audio files in a single pass, sound generated with the picture, and timestamp-level edits after. It plans shots — setup, turn, payoff — instead of stretching one moment.
30 s in one passnative audiomulti-round extend30 img · 10 vid · 10 audio refstimestamp editinggreen-screen edit
In Control RoomSits in the Video node. Takes a prompt, a start or end image, references, and an audio reference; hands a clip downstream.
Node · Video
1 / 9
Seedance 2.5ByteDance Seed
Kling O3Kuaishou
H3 · H3 MaxMiniMax
Wan 3.0 PrimeAlibaba Tongyi
Nano Banana ProGoogle DeepMind
GPT Image 2OpenAI
Eleven v3ElevenLabs
LyriaGoogle DeepMind
The WritersOpenAI · Anthropic · xAI · Google
Model names and marks belong to their owners; samples are theirs and link to their pages. Prices are Motionbar credits, quoted live in the studio.
How a flow runs
Check. Run the diff.
Keep what worked.
There is no paid test run and no overlay. The canvas is the progress bar.
01Wire itPick a node, click a port, pick what fits. Or describe the flow and the assistant proposes ghost nodes with the cost to run them.
02Check flowReady to run · about 6 min · 7 nodes — or a list of problems, each naming its node with a Fix button.
03Run all — the diffOnly out-of-date nodes run. Waiting, running, done and cached show on the nodes themselves; Stop cancels what hasn't started.
04Failures fail aloneA content check or a model error fails one node and refunds it. Everything else keeps what it made.
05Save it. Run it again.Swap the Upload for tomorrow's product and press Run. Or mark inputs and call it from the API.
Every model in the room
55 models. 20 labs.
One canvas.
Pick by name in any node. Every model shows its own settings and its own price on the node before you run. The list grows as models ship.
⌕
MH3MiniMax▶Video
image-to-videotext-to-videoreference-to-video
MH3 MaxMiniMax▶Video
image-to-videotext-to-videoreference-to-video
MH3 Max TurboMiniMax▶Video
text-to-videoimage-to-video
KKling 3.0Kuaishou▶Video
text-to-videoimage-to-video
KKling O3Kuaishou▶Video
text-to-videoimage-to-videoreference-to-video
PPixVerse V6PixVerse▶Video
text-to-videoimage-to-videotransition
SdSeedance 2.0ByteDance▶Video
text-to-videoimage-to-videoreference-to-video
SdSeedance 2.0 FastByteDance▶Video
text-to-videoimage-to-videoreference-to-video
SdSeedance 2.0 MiniByteDance▶Video
text-to-videoimage-to-videoreference-to-video
SdSeedance 2.5ByteDance▶Video
text-to-videoimage-to-videoreference-to-video
WWan 2.7Alibaba▶Video
text-to-videoimage-to-videoreference-to-video
WWan 3.0Alibaba▶Video
text-to-videoimage-to-videoreference-to-video
WWan 3.0 PrimeAlibaba▶Video
text-to-videoimage-to-videoreference-to-video
FFLUX.2 ProBlack Forest Labs▣Image
edit
GGPT Image 2OpenAI▣Image
edit
IdIdeogram V4Ideogram▣Image
image-to-image
GNano Banana 2Google▣Image
edit
GNano Banana ProGoogle▣Image
edit
WQwen Image 3Alibaba▣Image
text-to-imageedit
RRecraft V4.1Recraft▣Image
text-to-image
SdSeedream 5.0 ProByteDance▣Image
text-to-imageedit
AClaude Fable 5.1AnthropicTWriter
generate
AClaude Haiku 4.5AnthropicTWriter
generate
AClaude Opus 5.5AnthropicTWriter
generate
AClaude Sonnet 5AnthropicTWriter
generate
DDeepSeek V4 ProDeepSeekTWriter
generate
DDeepSeek V4.1 FlashDeepSeekTWriter
generate
GGemini 3.1 ProGoogleTWriter
generate
GGemini 3.8 FlashGoogleTWriter
generate
GGemma 4GoogleTWriter
generate
ZGLM 5.3ZhipuTWriter
generate
GGPT 6 AstraOpenAITWriter
generate
GGPT 6 LunaOpenAITWriter
generate
GGPT 6 SolOpenAITWriter
generate
XGrok 4.7xAITWriter
generate
KKimi K3MoonshotTWriter
generate
WQwen 3.8 FlashAlibabaTWriter
generate
WQwen 3.8 MaxAlibabaTWriter
generate
WQwen 3.8 Omni FlashAlibabaTWriter
generate
IIEleven Multilingual v2ElevenLabs∿Voiceover
tts
IIEleven v3ElevenLabs∿Voiceover
tts
GGemini 3.8 Flash TTSGoogle∿Voiceover
tts
WQwen Audio 3Alibaba∿Voiceover
tts
MSpeech 2.8 HDMiniMax∿Voiceover
generate
MSpeech 2.8 TurboMiniMax∿Voiceover
generate
XxAI TTSxAI∿Voiceover
tts
RsChatterbox HDResemble AI∿Voice
speech-to-speech
IIEleven Voice ChangerElevenLabs∿Voice
voice-changer
IIEleven DubbingElevenLabs⇄Translate
dubbing
HgHeyGen TranslateHeyGen⇄Translate
translate
▮Audio mixMotionbar⚙Steps
amix
▮Audio onto videoMotionbar⚙Steps
merge
SdVideo upscaleByteDance⚙Steps
upscale
▮Voice separationMotionbar⚙Steps
demucs
Model names and marks belong to their owners. The room offers a model while at least one of its endpoints is on; credits are quoted live on the node and on Check flow.
Flows as endpoints
Mark the inputs.
The flow becomes an API.
Mark a Prompt or an Upload as an Input and the flow gets a simple form, a REST endpoint and an MCP tool. Same credits as the canvas. A free /check beside every run.
curl -X POST https://api.motionbar.ai/api/flows/flw_8c1e2b/runs \
-H "Authorization: Bearer mb_live_…" \
-F idea="A 20-second sneaker ad, Spanish VO" \
-F product_photo=@sneaker-01.jpg
→ { "run_id": "run_3f9a", "status": "queued", "quote": "…" }
# free dry run · same body
POST /api/flows/flw_8c1e2b/check → { "ready": true, "quote": "…", "eta_s": 360 } Build it once.
Run it with a new product every morning.
Control Room is included on Studio and Floor, and open on pay-as-you-go.