Jian AI Lab · 中文EN

NINE FIELDS · 06

Video, Image & Audio

Compare editing, subtitles, voice, avatars, image generation, and video models with their actual compute costs.

How to use this pageStart with the job you need to finish. Open a research file, then check capability, requirements, license, and actual cost.

CURATED FOR THIS FIELD

29 GitHub projects worth investigating

All projects →

Each project is classified by the value it adds, its research status, requirements, and real operating costs.

Cut costsResearch queue

OpenCut-app/OpenCut

OpenCut

IN ONE SENTENCE

Keep everyday timeline editing, media organization, and export inside an open-source video workspace.

OpenCut is catalogued as open-source project. Read the research file and current GitHub README before using it in production.

Project type
Open-source project
Requirements
Web / 桌面路线
Real cost
MIT; models, APIs, compute, storage, hosting, and maintenance may add real costs. Verify the repository license and current service terms.
Add capabilityResearch queue

meituan-longcat/LongCat-Video

LongCat-Video

IN ONE SENTENCE

Make a person, animated character, or animal speak naturally from one image and an audio track.

LongCat-Video is catalogued as open-source video tool. Read the research file and current GitHub README before using it in production.

Project type
Open-source video tool
Requirements
Python / CUDA GPU / Docker
Real cost
MIT; models, APIs, compute, storage, hosting, and maintenance may add real costs. Verify the repository license and current service terms.
Cut costsTesting

lipku/LiveTalking

LiveTalking

IN ONE SENTENCE

Drive a real-time talking avatar from text or audio and stream it to the web, live video, or a virtual camera.

LiveTalking is catalogued as open-source project. Read the research file and current GitHub README before using it in production.

Project type
Open-source project
Requirements
云 GPU candidate
Real cost
Open-source; models, APIs, compute, storage, hosting, and maintenance may add real costs. Verify the repository license and current service terms.
Add capabilityVerified

KlingAIResearch/LivePortrait

LivePortrait

IN ONE SENTENCE

Animate a portrait from one image and a driving video with efficient motion and expression control.

LivePortrait is catalogued as open-source model. Read the research file and current GitHub README before using it in production.

Project type
Open-source model
Requirements
Windows / Linux / Apple Silicon Mac
Real cost
Open-source; models, APIs, compute, storage, hosting, and maintenance may add real costs. Verify the repository license and current service terms.
Add capabilityNew discovery

debpalash/VoiceStudio

VoiceStudio

IN ONE SENTENCE

Clone voices, dub video, transcribe speech, and build audiobooks locally without a required subscription.

VoiceStudio is catalogued as open-source speech tool. Read the research file and current GitHub README before using it in production.

Project type
Open-source speech tool
Requirements
macOS Apple Silicon / Windows / Linux / Docker
Real cost
AGPL-3.0; models, APIs, compute, storage, hosting, and maintenance may add real costs. Verify the repository license and current service terms.
Cut costsNew discovery

QuentinFuxa/WhisperLiveKit

WhisperLiveKit

IN ONE SENTENCE

Convert live speech into low-latency captions with speaker separation, translation, and an API.

WhisperLiveKit is catalogued as open-source tool. Read the research file and current GitHub README before using it in production.

Project type
Open-source tool
Requirements
Requirements to be verified
Real cost
Actual cost still needs verification.
Add capabilityNew discovery

ItusiAI/Open-Magiviz

Open-Magiviz

IN ONE SENTENCE

Move from a story idea through characters and storyboards to generated scenes and a complete video.

Open-Magiviz is catalogued as open-source video tool. Read the research file and current GitHub README before using it in production.

Project type
Open-source video tool
Requirements
Web / Next.js / PostgreSQL / self-hosted
Real cost
Open-source; models, APIs, compute, storage, hosting, and maintenance may add real costs. Verify the repository license and current service terms.
Add capabilityNew discovery

ATH-MaaS/Pixelle-Video

Pixelle-Video

IN ONE SENTENCE

Turn a topic into an automated pipeline for script, visuals, voice, music, and final video assembly.

Pixelle-Video is catalogued as open-source model. Read the research file and current GitHub README before using it in production.

Project type
Open-source model
Requirements
Requirements to be verified
Real cost
Actual cost still needs verification.
Automate workTesting

calesthio/OpenMontage

OpenMontage

IN ONE SENTENCE

Turn a brief into research, script, media, voice, subtitles, editing, rendering, and a finished video.

OpenMontage is catalogued as open-source video tool. Read the research file and current GitHub README before using it in production.

Project type
Open-source video tool
Requirements
Python / Node.js / FFmpeg / AI 编程 Agent
Real cost
AGPL-3.0; models, APIs, compute, storage, hosting, and maintenance may add real costs. Verify the repository license and current service terms.
Cut costsResearch queue

Comfy-Org/ComfyUI

ComfyUI

IN ONE SENTENCE

Build generation workflows like blocks, save them once, and reuse them for images, video, audio, and 3D.

ComfyUI is catalogued as workflow automation platform. Read the research file and current GitHub README before using it in production.

Project type
Workflow automation platform
Requirements
Windows / macOS / Linux / Localorcloud
Real cost
GPL-3.0; models, APIs, compute, storage, hosting, and maintenance may add real costs. Verify the repository license and current service terms.
Cut costsResearch queue

Wan-Video/Wan2.1

Wan2.1

IN ONE SENTENCE

Use one open model family for text-to-video, image-to-video, frame control, editing, and audio tasks.

Wan2.1 is catalogued as open-source video tool. Read the research file and current GitHub README before using it in production.

Project type
Open-source video tool
Requirements
Python / NVIDIA GPU; 部分模型可用消费级显卡
Real cost
Apache-2.0; models, APIs, compute, storage, hosting, and maintenance may add real costs. Verify the repository license and current service terms.
Cut costsResearch queue

Tencent-Hunyuan/HunyuanVideo-1.5

HunyuanVideo 1.5

IN ONE SENTENCE

Run Tencent's open video-generation models through local GPU, Diffusers, or ComfyUI workflows.

HunyuanVideo 1.5 is catalogued as open-source video tool. Read the research file and current GitHub README before using it in production.

Project type
Open-source video tool
Requirements
Linux / Python 3.10+ / NVIDIA GPU; 模型卸载时官方minimum约 14GB VRAM
Real cost
Open-source; models, APIs, compute, storage, hosting, and maintenance may add real costs. Verify the repository license and current service terms.
Cut costsResearch queue

Lightricks/LTX-Video

LTX-Video

IN ONE SENTENCE

Control video development with keyframes, extend clips forward or backward, and generate picture with sound.

LTX-Video is catalogued as open-source video tool. Read the research file and current GitHub README before using it in production.

Project type
Open-source video tool
Requirements
Python / NVIDIA GPU / ComfyUI / online demo
Real cost
Apache-2.0; models, APIs, compute, storage, hosting, and maintenance may add real costs. Verify the repository license and current service terms.
Add capabilityNew discovery

MartinDelophy/ai-video-editor

Timeline Studio

IN ONE SENTENCE

Edit multitrack video, subtitles, dubbing, AI enhancement, and offline exports directly in the browser.

Timeline Studio is catalogued as open-source video tool. Read the research file and current GitHub README before using it in production.

Project type
Open-source video tool
Requirements
桌面Browser / PWA / WebGPU
Real cost
MIT; models, APIs, compute, storage, hosting, and maintenance may add real costs. Verify the repository license and current service terms.
Add capabilityResearch queue

invoke-ai/InvokeAI

InvokeAI

IN ONE SENTENCE

Combine generation, inpainting, a creative canvas, asset management, and node workflows locally.

InvokeAI is catalogued as open-source image tool. Read the research file and current GitHub README before using it in production.

Project type
Open-source image tool
Requirements
Local Web UI / GPU
Real cost
Apache-2.0; models, APIs, compute, storage, hosting, and maintenance may add real costs. Verify the repository license and current service terms.
Cut costsResearch queue

mcmonkeyprojects/SwarmUI

SwarmUI

IN ONE SENTENCE

Use a simple generation screen and advanced Comfy workflows from the same local image interface.

SwarmUI is catalogued as open-source image tool. Read the research file and current GitHub README before using it in production.

Project type
Open-source image tool
Requirements
Windows / macOS / Linux; Localor云 GPU
Real cost
MIT; models, APIs, compute, storage, hosting, and maintenance may add real costs. Verify the repository license and current service terms.
Cut costsResearch queue

ByteDance-Seed/BAGEL

BAGEL

IN ONE SENTENCE

Understand, generate, and edit images with one open multimodal model from ByteDance Seed.

BAGEL is catalogued as open-source model. Read the research file and current GitHub README before using it in production.

Project type
Open-source model
Requirements
Python / Gradio / CUDA GPU
Real cost
Apache-2.0; models, APIs, compute, storage, hosting, and maintenance may add real costs. Verify the repository license and current service terms.
Cut costsResearch queue

lllyasviel/Fooocus

Fooocus

IN ONE SENTENCE

Focus on prompts instead of parameters with a streamlined local image-generation experience.

Fooocus is catalogued as open-source image tool. Read the research file and current GitHub README before using it in production.

Project type
Open-source image tool
Requirements
Windows / Linux / Colab; NVIDIA GPU preferred
Real cost
GPL-3.0; models, APIs, compute, storage, hosting, and maintenance may add real costs. Verify the repository license and current service terms.
Add capabilityResearch queue

Acly/krita-ai-diffusion

Krita AI Diffusion

IN ONE SENTENCE

Generate, inpaint, upscale, and use references without leaving Krita's real editing workflow.

Krita AI Diffusion is catalogued as open-source project. Read the research file and current GitHub README before using it in production.

Project type
Open-source project
Requirements
Krita 5.2+; Windows / macOS / Linux
Real cost
GPL-3.0; models, APIs, compute, storage, hosting, and maintenance may add real costs. Verify the repository license and current service terms.
Cut costsResearch queue

lllyasviel/stable-diffusion-webui-forge

Stable Diffusion WebUI Forge

IN ONE SENTENCE

Manage models, LoRAs, and ControlNet with more efficient local Stable Diffusion inference.

Stable Diffusion WebUI Forge is catalogued as open-source image tool. Read the research file and current GitHub README before using it in production.

Project type
Open-source image tool
Requirements
Windows 一键包 / Python self-hosted
Real cost
AGPL-3.0; models, APIs, compute, storage, hosting, and maintenance may add real costs. Verify the repository license and current service terms.
Cut costsResearch queue

Kwai-Kolors/Kolors

Kolors

IN ONE SENTENCE

Generate images with stronger Chinese and English text rendering for posters and visual content.

Kolors is catalogued as open-source image tool. Read the research file and current GitHub README before using it in production.

Project type
Open-source image tool
Requirements
Python / Gradio / ComfyUI / NVIDIA CUDA
Real cost
Apache-2.0; models, APIs, compute, storage, hosting, and maintenance may add real costs. Verify the repository license and current service terms.
Add capabilityResearch queue

Stability-AI/generative-models

Stability Generative Models

IN ONE SENTENCE

Start from Stability AI's official code, inference examples, and model-license boundaries.

Stability Generative Models is catalogued as open-source model. Read the research file and current GitHub README before using it in production.

Project type
Open-source model
Requirements
Python 3.10 / CUDA / Streamlit
Real cost
MIT; models, APIs, compute, storage, hosting, and maintenance may add real costs. Verify the repository license and current service terms.
Cut costsResearch queue

KDE/kdenlive

Kdenlive

IN ONE SENTENCE

Keep professional multitrack editing, transitions, and audio work inside a mature open-source editor.

Kdenlive is catalogued as open-source video tool. Read the research file and current GitHub README before using it in production.

Project type
Open-source video tool
Requirements
Windows / macOS / Linux
Real cost
GPL-3.0; models, APIs, compute, storage, hosting, and maintenance may add real costs. Verify the repository license and current service terms.
Cut costsResearch queue

lllyasviel/FramePack

FramePack

IN ONE SENTENCE

Generate longer image-to-video sequences in segments on comparatively modest local GPU hardware.

FramePack is catalogued as open-source video tool. Read the research file and current GitHub README before using it in production.

Project type
Open-source video tool
Requirements
Windows / Linux / NVIDIA RTX 30、40、50 系; minimum 6GB VRAM
Real cost
Apache-2.0; models, APIs, compute, storage, hosting, and maintenance may add real costs. Verify the repository license and current service terms.
Cut costsResearch queue

zai-org/CogVideo

CogVideo / CogVideoX

IN ONE SENTENCE

Bring text, reference images, and existing clips into an open, self-deployable video-generation workflow.

CogVideo / CogVideoX is catalogued as open-source video tool. Read the research file and current GitHub README before using it in production.

Project type
Open-source video tool
Requirements
Python / NVIDIA GPU / Colab
Real cost
Open-source; models, APIs, compute, storage, hosting, and maintenance may add real costs. Verify the repository license and current service terms.
Cut costsResearch queue

ace-step/ACE-Step-1.5

ACE-Step 1.5

IN ONE SENTENCE

Generate complete songs, covers, edits, and separated stems from a description or reference audio.

ACE-Step 1.5 is catalogued as open-source model. Read the research file and current GitHub README before using it in production.

Project type
Open-source model
Requirements
Windows / macOS / Linux; CUDA、Apple Silicon、ROCm、Intel XPU or CPU
Real cost
MIT; models, APIs, compute, storage, hosting, and maintenance may add real costs. Verify the repository license and current service terms.
Cut costsResearch queue

QwenAudio/CosyVoice

CosyVoice

IN ONE SENTENCE

Build multilingual, Chinese-dialect, cross-lingual cloning, and streaming speech services from one model family.

CosyVoice is catalogued as open-source speech tool. Read the research file and current GitHub README before using it in production.

Project type
Open-source speech tool
Requirements
Python / Linux / Docker / NVIDIA GPU
Real cost
Apache-2.0; models, APIs, compute, storage, hosting, and maintenance may add real costs. Verify the repository license and current service terms.
Cut costsResearch queue

hexgrad/kokoro

Kokoro

IN ONE SENTENCE

Create lightweight multilingual text-to-speech that is fast enough to integrate into real content workflows.

Kokoro is catalogued as open-source speech tool. Read the research file and current GitHub README before using it in production.

Project type
Open-source speech tool
Requirements
Python 3.10–3.13 / Localor Colab
Real cost
Apache-2.0; models, APIs, compute, storage, hosting, and maintenance may add real costs. Verify the repository license and current service terms.
Cut costsResearch queue

modelscope/ClearerVoice-Studio

ClearerVoice-Studio

IN ONE SENTENCE

Repair noisy, reverberant, or unclear recordings with enhancement, separation, and target-speaker extraction.

ClearerVoice-Studio is catalogued as open-source speech tool. Read the research file and current GitHub README before using it in production.

Project type
Open-source speech tool
Requirements
Python / Linux preferred / Localor GPU
Real cost
Apache-2.0; models, APIs, compute, storage, hosting, and maintenance may add real costs. Verify the repository license and current service terms.