CREATION MODES

Text to Video Video to Video URL to Video

MODELS

Runway Luma Topaz Upscale Lip Sync Gemini Omni

Guest trial

Attempts left 3 Available

My Library

· 0 videos

Your next video starts here

Finished and in-progress generations will stay organized in this Account Library.

Write a prompt →

AI video model workspace

OmniHuman talking video generator for portrait-led performances

Use OmniHuman 1.5 on ReelPush with a required source image and driving audio reference for one to 15 second talking-person video assets.

No credit card Account-owned Library

17available models

4creation modes

1Account Library

An AI video maker for teams that publish often

DTC brands Creative agencies In-house teams Independent creators

See what reelpush.ai creates

A precision product reveal on a reflective studio surface

reelpush.ai

A vertical skincare routine with clean morning light

reelpush.ai

Aerial movement over a quiet coastline at sunrise

reelpush.ai

Macro coffee pour with warm, tactile detail

reelpush.ai

A bold sneaker launch built for a 9:16 feed

reelpush.ai

A cinematic night market with a slow camera push

reelpush.ai

Soft fabric movement for a seasonal apparel drop

reelpush.ai

A compact unboxing sequence with crisp transitions

reelpush.ai

OmniHuman talking video generator

OmniHuman on ReelPush

OmniHuman 1.5 is ByteDance's human-animation model. The ReelPush workflow starts from one source image and audio reference, with 720p or 1080p output options.

ByteDance

OmniHuman capabilities available now

1 available variants, image input, up to 1080p, 1 to 15 seconds, source image, and up to 1 reference audio

Treat the portrait as casting

The source image defines the person and composition, so use a clear authorized portrait near the intended output framing.

Drive timing with final audio

The audio reference supplies speech timing. Prepare the final read before generating visual performance.

Choose 720p or 1080p

Resolution is visible in the request, along with a one-to-15-second duration and the active Credit quote.

ReelPush workflow

How to use OmniHuman on ReelPush

Prepare an authorized portrait

Choose one clear image with a visible face and composition suitable for the intended short-form frame.

Add the final audio reference

Upload a clean spoken track within the supported duration, then choose resolution and review the quote.

Check performance and consent

Inspect lip timing, face identity, hands, body motion, audio, disclosure, and usage rights before sharing.

Short-form video uses

Three jobs suited to OmniHuman

Founder message prototype

Test an approved portrait and script read for an internal creative review.

Multilingual spokesperson draft

Pair authorized imagery with a localized audio track and check disclosure and consent requirements.

Talking product educator

Create a short explanatory performance for later editing with captions and supporting footage.

Choose by task

When to compare OmniHuman with another model

OmniHuman or InfiniteTalk

Both begin with portrait and audio. OmniHuman offers 720p and 1080p; InfiniteTalk offers 480p and 720p in the current catalog.

InfiniteTalk Video Generator

OmniHuman or source-video lip sync

OmniHuman creates the performance from a still. Lip sync keeps an existing recorded performance and changes mouth timing.

Volcengine Lip Sync

Evidence

Product facts and independent material

Capability facts come from ReelPush's live model catalog. External links provide workflow context and are not endorsements or customer testimonials.

ReelPush OmniHuman capability catalog Independent OmniHuman workflow video

Product boundaries

Generate a video asset, then complete the publishing work

ReelPush generates and stores video assets from the selected model's supported inputs. It does not automatically add final captions, verify claims, clear likeness or music rights, schedule posts, or publish to a channel. Review every output for identity, objects, movement, audio, brand rules, consent, and platform requirements before use.

Model questions

OmniHuman questions, answered

What input does OmniHuman 1.5 require?

The ReelPush workflow requires one source image and at least one audio reference. The image establishes the person and frame, while the audio drives speech timing. Use only material you are authorized to process and review likeness, voice, and disclosure obligations.

What OmniHuman resolutions are available?

The current OmniHuman 1.5 integration lists 720p and 1080p. Select resolution before generation and check the live Credit quote. Higher resolution does not remove the need to inspect facial identity, mouth motion, hands, clothing, background, and audio synchronization.

How long can an OmniHuman clip be?

The active ReelPush catalog supports one to 15 seconds. Prepare audio within that range and keep the performance focused on one short statement. Longer scripts should be divided into deliberate segments that can be reviewed for timing, continuity, and visual consistency.

Can I create a video of any person?

Use a person's image and voice only when you have the necessary rights and informed consent for the intended context. You remain responsible for authorization, disclosure, claims, and platform rules. Review the output carefully and do not present generated behavior as a real event.

Unlock this generation

My Library

Your next video starts here

Generating video