In development · Free · Open source planned

Forge VFX

Motion graphics and compositing, running entirely on your own GPU.

A multi-layer timeline, node-based effects, 3D layers and cameras, and local AI generation of video, images and audio. Written from the ground up in C++ on its own Vulkan and OpenGL renderer.

See it generate
Forge VFX editor: a composited night city scene in the viewport, a multi-layer timeline, and the inspector showing a gradient effect and an audio spectrum
51.7klines of first-party C++
2render backends, Vulkan and OpenGL
16built-in effects, node-graph ready
51.8fpssustained 4K export on an Apple M2

01 · Generate

Describe the shot.
Your GPU renders it.

Draw a region on a still, write a prompt, and Forge generates the clip on your own hardware, then drops it onto a new layer ready to composite. No uploads, no queue, no credits.

Generate Local
Type
Video
Model
Wan 2.2 TI2V 5B
Area
Region only, 908 × 642
Frames
65
Steps
20
Seed
7

Prompt

Ready.
Output Waiting for prompt

The demo runs at presentation speed. Generating a clip like this takes minutes on a desktop GPU; Wan 2.2 5B is comfortable at 12 GB of VRAM and runs from 6 GB.

The generated flames composited back into a photograph of a cabin living room with falling snow outside the window

From clip to scene

The generated flames are feathered back into the photograph they came from. A mask keeps them inside the firebox, a Particle System layer adds snow that only falls behind the window glass, and three candle layers animate the mantel. One project, one timeline, one export.

  • VideoWan 2.2, LTX-Video, HunyuanVideo 1.5, CogVideoX
  • ImageFLUX.1 schnell, SDXL, Stable Diffusion 3.5
  • AudioStable Audio Open
  • MotionDepth-driven parallax and orbit from a single still

First run offers a one-click setup that installs a private Python and PyTorch, the CUDA build when an NVIDIA GPU is present, touching nothing else on the system.

02 · Compose

Every layer, every parameter, on one timeline.

Video, image sequences, SVG, text, shapes, solids, audio, nested scenes and glTF models all live on the same frame-accurate timeline.

  • Keyframes on any parameter, with fades and a seeded wiggle layered on top
  • Any layer can go 3D and be seen through a scene camera
  • Normal, Add, Multiply and Screen blending on premultiplied alpha
  • An Audio Spectrum effect that draws from a layer’s sound
Detail of the Forge VFX timeline with repeated clip segments, and the inspector showing a gradient effect with a wiggle animator and an audio spectrum

03 · Effects

A stack when you want it. A graph when you need it.

Every layer owns a node graph. A simple effect stack is just a straight chain through it, so you can start in the inspector and open the graph the moment a shot gets complicated.

BlurToneBrightness & ContrastHue Saturation ValueColor KeyGradientNoisePlotParticle SystemCrowdShape MaskMatte MaskDisplacement MapAudio SpectrumChange PitchRepeat
Forge VFX node editor wiring a particle system into a matte mask over an indoor pool scene

04 · Deliver

Straight to production formats.

The graph above, rendered: twenty thousand rain particles, held behind the glass by a matte.

Final rendered frame of an indoor pool at night with rain falling outside the windows
H.264MP4, MOV
HEVCMP4, MOV
ProRes 422 / 4444MOV on macOS
AAC / PCMmixed scene audio

Hardware encoders are probed per codec through VideoToolbox on macOS and Media Foundation on Windows.

05 · Engine

Built on its own renderer.

No game engine, no UI toolkit. Forge draws its entire interface and every composite through a small render hardware interface with two interchangeable backends.

Language
C++17, Objective-C++ for macOS platform code
Rendering
Handle-based RHI. Vulkan by default (dynamic rendering, synchronization2, VMA), OpenGL as the reference backend
Shaders
59, as GLSL 450 compiled to SPIR-V and GLSL 330, addressed by logical name
Compositing
Offscreen targets per scene, nested compositions, dirty-flag caching, 4× MSAA
Compute
Compute-shader effect path on Vulkan, fragment fallback on OpenGL
Decoding
Multi-cursor video readers that walk forward instead of seeking
Generation
Python sidecar over a line protocol, Hugging Face diffusers, ONNX Runtime depth
Platforms
macOS, Windows, Linux

4K export, Apple M2, sustained

Before
2.2 fps
After
51.8 fps

A profiling pass fixed an unoptimized build, stray validation layers, per-frame buffer reallocation, a presentation path that tied export to the display's refresh rate, and a scalar copy into the encoder that became 3.8× faster with NEON. Output stayed bit-identical throughout.

Free. Open source next.

Made to be shared.

Forge VFX will be free to use, and the source is being prepared for release under the GNU GPL v3, the license behind Blender, Krita and Kdenlive. Every fork stays open.

NowKeyframe easing curves and advanced shape tools
NextPipelined export: render frame N while encoding N−1
ThenPublic repository and first downloadable builds
Follow development

Questions

Is Forge VFX free?

Yes. It will be free to use, and the source code is planned for release under the GPL v3.

Does anything leave my machine?

No. Generation runs on local models. Nothing is uploaded, and there are no accounts or usage credits. A free Hugging Face sign-in is only needed to download models whose publishers gate them.

What hardware do I need?

The editor runs on any GPU with Vulkan or OpenGL support. AI video generation is the demanding part: Wan 2.2 5B is comfortable with 12 GB of VRAM and 16 GB of system memory, and runs from 6 GB of VRAM.

When can I download it?

There is no public build yet. Join the Discord to follow development and hear first when builds are available.