Language

MiniMax H3 text-to-video workflow

Imported from ModelTrends

MiniMax H3 turns a prompt into a video with its sound generated in the same pass: describe the scene, the motion and what is heard, and the clip comes back with audio. It is the most used video workflow on Graydient, with a large library of fine-tuned styles and motions that drop in as LoRAs. Open weights, hosted and ready to call; no setup.

Compatible / Available in Graydient-powered Apps

Sample renders

Made with the models this workflow runs. Prompts are credited on BunchaPrompts; the model opens on ModelTrends. Pick a thumbnail to see the render large with the prompt that made it.

Sample render 1

Prompt by Moonbear_AIArt on BunchaPrompts · Grandline - Legendary Stylized Illustration

Prompt

a stylized illustration of a warrior lunging forward through smoke and sparks, cloak whipping in the wind, embers trailing behind, intense forward motion, dynamic perspective, cinematic lighting

Prompts from BunchaPrompts. Workflows and models on ModelTrends.

Options

OptionMaps toDefault
(prompt)CLIP Text Encode → text—
/guidanceKSampler → cfg4
/lengthEmpty MiniMax Latent → length (frames)125
/seedKSampler → seedrandom
/sizeEmpty MiniMax Latent → width, height640 × 640
/stepsKSampler → steps20

Models used

MiniMax H3 Video Model imported by shortcode /run:minimax-h3 from ModelTrends See it on ModelTrends

For Developers (optional)

Example API snippet. You can build your own app with this workflow for personal use. Contact us for high availability servers when you're ready to deploy to the world. See the full rendering API documentation.

curl https://cloud.graydient.ai/api/v3/render/ \
  -X POST \
  -H "Content-Type: application/vnd.api+json" \
  -H "Accept: application/vnd.api+json" \
  -H "Authorization: Bearer $GRAYDIENT_TOKEN" \
  -d '{
    "prompt": "/workflow /run:minimax-h3 /size:640x640 /length:125 a parrot in a pirate hat dances on the deck of a ship at sunset while a sea shanty plays",
    "callback_url": "https://example.com/webhook/render"
  }'

This is how it works behind the scenes. This is already configured for you. The parameters in the table above plug into these slots, allowing you to change how it works and which models get loaded. You don't have to understand or modify this - it just works, already set up.

Load Diffusion Model MODEL to KSampler; Load CLIP CLIP to CLIP Text Encode (Prompt); CLIP Text Encode (Prompt) CONDITIONING to KSampler; Empty MiniMax Latent LATENT to KSampler; KSampler LATENT to VAE Decode; KSampler LATENT to VAE Decode Audio; Load VAE VAE to VAE Decode; Load VAE (audio) VAE to VAE Decode Audio; VAE Decode IMAGE to Create Video; VAE Decode Audio AUDIO to Create Video; Create Video VIDEO to Save Video
Load Diffusion Model#1
MODEL
unet_nameMiniMax H3 (dynamic c…
weight_dtypedefault
Load CLIP#2
CLIP
clip_nameMiniMax H3 text encod…
typeminimax
Load VAE#3
VAE
vae_nameMiniMax H3 video VAE
Load VAE (audio)#4
VAE
vae_nameMiniMax H3 audio VAE
CLIP Text Encode (Prompt)#5
clipCONDITIONING
texta parrot in a pirate …(prompt)
Empty MiniMax Latent#6
LATENT
width640/size
height640/size
length125/length
frame_rate25
KSampler#7
modelLATENT
positive
latent_image
seedrandom/seed
steps20/steps
cfg4/guidance
VAE Decode#8
samplesIMAGE
vae
VAE Decode Audio#9
samplesAUDIO
vae
Create Video#10
imagesVIDEO
audio
fps25
Save Video#11
video
filename_prefixGraydient
formatmp4

Read-only view. The highlighted pills are the options a call can set. Describe the sound as well as the picture; the fine-tunes change the style or the motion.

All the workflows: Powerful Workflows

See plans and pricing