Create MiniMax H3 talking videos from a single image and an audio file.

Choose the video duration, upload your audio, load the image you want to animate, select the maximum output size, and adjust the prompt if needed. The workflow uses the audio to drive the speaking performance while keeping the uploaded image as the visual reference.

The Load Audio node makes it easy to select exactly the part of the sound file you want. Drag across the waveform to choose the section, or use Start at for precise timing. Because the Duration node is connected to it, the selected audio automatically matches the video length. If the audio is too short, the remaining time can be filled with silence depending on the node settings.

For Longest Side, choose how large you want the longest edge to be. Use Keep to preserve the image's original shape, or select a ratio such as 1:1, 16:9, 9:16, or 2:3 if you want it cropped.

This workflow is ideal for talking characters, avatars, presenters, mascots, dialogue clips, voiceovers, and short social media videos.

Learn more from Pixaroma on YouTube:
https://www.youtube.com/@pixaroma