AI creators tools

VACE

VACE by Alibaba is a wild all-in-one AI tool for generating and editing videos using just text images or clips, based on Wan.

Visit This Site

Overview

VACE is an AI video tool from Alibaba’s Tongyi Lab and it’s a straight-up powerhouse. You can make videos from text clips or images, edit existing footage or mash it all together with masks and style references. It’s still not up for sale but already pulling off stuff others can’t.

The folks behind it also made Wanx and LTX-Video so they’ve been busy with AI image and video stuff. With VACE they’ve gone for a one-stop setup. You don’t juggle different tools anymore. Just plug everything into this thing called VCU and go. Text video masks clips whatever. VACE takes it all in and spits out finished results.

Add styl,e change poses, adjust depth.
Masked Editing: Want to erase or add just one thing this does that too.
Stacking Tasks: Mix features like style+pose+object to build wild stuff.

It runs on something called DiT which helps it keep quality high. It's made to scale up and tune without retraining everything from scratch.

Tags

Freeware Apache License 2.0 PC-based #Video & Animation

Educators and Trainers Creative Professionals Content Creators Media and Film Makers Marketing and Branding Specialists Developers and Tech Creators Nonprofit and Advocacy Creators Small Business Owners Entertainment and Performance Artists Professional Content Creators

This tool is free to use when installed locally and is offered under Apache License 2.0.

Prompt:
Surreal cinematic scene of a woman in red and black outfit doing gymnastics on a large rooftop of a skyscraper, composed of large reflective glass or metal panels arranged in a clean, grid-like pattern. The surface is smooth and glossy. Her poses defy gravity. She wears black combat boots. Thick misty clouds clouds moving low to the ground and pastel-blue sky.

Generated on May 29, 2025:

Pose transfer from Matrix movie. Wow.
Prompt:
A young woman in black pants and crop top doing yoga, behind her a graffiti wall

Generated on May 27, 2025:

Pose transfer with background as reference.
Prompt:
A hungry young woman with glowing green eyes is holding a black bowl of spaghetti, camera pushes in as she aggressively twirls a golden fork through spaghetti and devours it slightly comically, light reflections playing off her futuristic emerald-scaled outfit and spaghetti as she keeps gobbling them.

Generated on May 19, 2025:

Vace takes 'glowing green eyes' literally - kind of expected it coul, but fork goung through the head is funny. Clearly, not yeat ready to pass the famous pasts eating test. But it's getting there.
Prompt:
A confident young woman in a metallic green snakeskin-patterned outfit lifts a pair of bold, angular green-and-gold sunglasses and smoothly slides them on. She turns her head slightly left, then right, showcasing the geometric gold detailing under soft, high-contrast lighting. With her face now centered again, she lifts one hand and gives a sharp, playful thumbs-up to the camera, her lips curling into a subtle smirk. The luxe vibe radiates attitude and approval in a single smooth sequence.

Generated on May 18, 2025:

Uploaded 2 images: woman in a shimmering green snakeskin-patterned outfit + fancy shades. Her eye is not healthy looking. The prompt wasn't followed perfectly but caught the most important part, guess there wouldn't be enough time for the full set of actions anyway. Don't quite fancy her smile here.
Prompt:
A striking young woman in a shimmering green snakeskin-patterned outfit holds a vibrant "AI Creators Tools" can at shoulder height beside her face. She smiles warmly, then turns her head slightly to admire the can with a soft, approving gaze. The studio lighting glides across the gold accents on her collar and the vivid can design. As she continues holding it steady, the camera begins a slow, deliberate zoom toward the can, framing it as the star of the scene.

Generated on May 18, 2025:

Uploaded 2 images: woman in a shimmering green snakeskin-patterned outfit and my branded soda can, featured here on the website many times. Text is tricky to replicate in a video, otherwise quite good.
Rating:
Favorite

Latest VACE News

May 15, 2025

Wan2.1-VACE is here, a new video creation and editing model available in 1.3B and 14B sizes under the Apache-2.0 license. The model supports reference-to-video generation, video-to-video editing, and masked video editing. It uses multimodal inputs to improve control and allows combining features for more complex video tasks.

Useful Links
VID 2 VID workflow for the 14b

Workflow

A.I.Warper shared the WAN 2.1 VACE video-to-video workflow for the 14bn model in their Patreon (free). It’s said to handle 360° turns and fast motion better than the 1.3bn. Sage Attention is off, SDPA is on by default.

Added on: May 23, 2025
Quantized VACE 14B

Version

This is a direct GGUF conversion of Wan-AI/Wan2.1-VACE-14B

Added on: May 21, 2025

This page was last updated on August 11, 2025 at 3:09 AM