AI filmmaking on a node canvaswith a built-in diffusion engine
Inline Studio is a free, open-source app for AI filmmaking on a node canvas. Generate locally with the built-in Inline Core engine, add hosted models with API Nodes when you want them, train your own custom styles and consistent characters on your own GPU, and keep every version as a non-destructive take.
Free and open source, GPL-3.0 licensed. Runs on macOS, Windows, and Linux.
A single render is not a film.
It's one layer of many.
A single graph makes one thing well. A film is dozens of them: shots, each with its own graph, layered and wired together until they add up to something greater. Inline Studio is where you compose that. Build a graph, turn it into a shot, feed it into the next, and keep every version that works, so the canvas holds the whole pipeline, not just the last render.
Graph → Shot → Layer → Pipeline
Trained with Inline Studio
Krea 2 LoRA for realistic skin texture
Our first published LoRA brings back real pores, fine hair, and natural texture across skin tones, ages, and lighting. Free to download, with the dataset it trained on.
- Base: Krea 2 RAW, carries over to Turbo
- Rank 16, full scope (attention + feed-forward)
- 1024px, 1500 steps, learning rate 1e-4
- 26 image and caption pairs
- Trained locally on your own GPU
- Trigger: inline-skin-lora
or grab the training dataset. New to LoRAs? Check out the training guide.
Local generation
Generate locally, built in
Inline Core is a from-scratch generation engine built into Inline Studio. No account, no external server, no cloud. Drop a Z-Image Turbo, FLUX.2 or Krea 2 node, wire a prompt, and hit Run. That is one node, with no loader, sampler, or VAE wiring, generating on your own GPU. Krea 2 ships as two models that work together: Turbo for fast 8-step images and RAW for maximum quality, and any LoRA you train applies to both. For video there is MiniMax H3, which generates a clip and its soundtrack in a single pass. Core owns the whole backend: it serves the web UI, runs the models, keeps the project database, and drives the timeline, all as a single Python process. Add a LoRA loader node to bring your own LoRA models.
Made locally with Krea 2
Generated on the Inline Core engine, no cloud. Krea 2 ships as Turbo for fast 8-step images and RAW for maximum quality.
LoRA training
Your character, your style, your look
Keep a character consistent across shots, or lock in a custom visual style, by training your own model (a LoRA) right on the canvas. The Trainer tab turns a folder of images into a trained style without leaving the app. Wire Load Dataset into Caption into Train LoRA, watch the loss curve fall live, and the finished style saves straight to your models/loras folder, ready to drop onto the generation canvas. Read the full LoRA training guide.
- Load Dataset
- Caption
- Train LoRA
- Graph
- Resources
Training is just another graph
No scripts, no notebooks, no config files. Auto-caption a folder of images, wire the nodes, and hit Start. The Graph node plots loss step by step while a Resources monitor tracks CPU, RAM, and VRAM in real time.
Stop and resume, never restart
Stop a run and it flushes a checkpoint with the adapter weights, optimizer and RNG state, and step count all intact, so resuming picks up on the exact step you paused. Interrupted runs recover on their own.
Z-Image and Krea 2
Train on Z-Image Turbo, or on Krea 2 RAW and generate with Krea 2 Turbo, the LoRA carries over. A 4-bit base fits Krea 2 training on a 16GB card. Set a trigger word, and the moment training finishes your LoRA lands in models/loras, ready to prompt.
What fits on your GPU
Measured peak VRAM for a LoRA training run, so you know before you start.
| Model and resolution | Fits 16GB? | Peak VRAM |
|---|---|---|
| Z-Image, 512px | Yes | ~13GB |
| Z-Image, 1024px | No, needs 24GB | ~15GB |
| Krea 2 (4-bit base), 512px | Yes | ~12GB |
| Krea 2 (4-bit base), 1024px | No, needs 32GB | ~28GB |
Peak allocation on a Tesla T4 (16GB) and L40S, rank 16, batch 1, gradient checkpointing on. Leave headroom for your GPU's context. A LoRA trained at 512 still applies at any generation resolution. See the full matrix. Or read the full LoRA training guide.
Community extensions
Bring your own extension packs
Nodes are not just what we ship. Bundle your own nodes into an extension pack and share it with the community, or install a published pack in one click from the registry. Its nodes drop straight onto your canvas.
- Browse & one-click install with Inline Registry
- Security scan built in
- Hot reload enabled
- Zero dependency conflicts
- Model aware
Hosted models, no GPU
API Nodes bring hosted, closed models onto the same canvas: no GPU, no setup, instant creative range. Add a Generate node, pick a model, and bring your own provider key. It stays on your machine, and you pay the provider per render, with a rough price estimate on every node before you run. Mix hosted models with your local renders in the same film.
- Image
- GPT Image 2, Nano Banana, Krea (hosted)
- Video
- LTX, Seedance
- Audio
- Sonilo
Showcase
Projects built on Inline Studio
Real films, put together shot by shot on the canvas, start to finish, without jumping between apps.
View all projectsCircuit Race: Animated Short Film with LTX 2.3 and GPT Image Generation
See how it was madeThe sequence
- 01Garage
- 02Helmet in hand
- 03Into the car
- 04Eyes closed
- 05She appears
- 06Fire in his eyes
- 07Lights out
- 08The win
Made with
- Stills
- GPT2 Text to Image
- Video clips
- LTX 2.3
- Final edit
- Premiere Pro
Storyboard, character, every scene, and the final cut all ran from one canvas, with Inline Studio generating behind each frame.
From the community
Real builders, talking about Inline Studio
Unprompted reactions across Reddit, X, and the ComfyUI community.
“This is the part of open source you can’t fake. Someone wanted a tool that didn’t exist, built it on us, and gave it to everyone.”
“The frame/take model solves a real problem, especially for anyone doing narrative work where you’re juggling multiple versions and need to track what worked. Node graphs get messy fast once you’re thinking in sequences rather than single images.”
“…overall you’ve got a great style here, and the feeling is very tight.”
“It’s a neat concept for a process that you can clearly iterate on to improve. Seems way more useful than something like the Director node because you have way more options to think about the whole project.”
“I love this. The whole comfyui node based never sat with how I think and work. Will definitely explore.”
“A new tool called ‘Inline Studio’ for managing ComfyUI video production on a frame-by-frame basis is gaining attention. It’s well-received as a UI that complements traditional node management.”
“Assets, scenes, characters, cars, storyboards, final shots: organized in a versioned visual pipeline. Not just generation. Production structure.”
On the canvas
Assemble and edit, right on the canvas
Cut your sequence together and trim your clips without leaving Inline Studio. The timeline and the edit both live as nodes, wired into the same pipeline as everything else.
Video Director
A timeline-in-a-node that combines your rendered frames into a single cut. Wire unlimited frame outputs into it to build the video track, layer audio on top (your videos' own audio plus your own music or voiceover) with per-input and per-layer volume, scrub the in-node preview, and export the combined high-res video.
Trim Video/Audio
A simple trim node. Drop in a video or audio clip, drag the in/out handles over its filmstrip or waveform, and pass just the trimmed segment downstream to the director or preview node.
Install
Four commands and it runs
One command sets up the engine, the model runtime, the trainer and the UI. On an NVIDIA machine it pulls the CUDA build of PyTorch for you.
- 01Python 3.11 or newer. No Node.
- 02NVIDIA on Linux or Windows is the tested path. CPU works, slowly.
- 03No downloads first. Grab checkpoints on first run.
- 04Leaves your other environments alone.
$ git clone https://github.com/inlineresearch/Inline-Studio.git$ cd Inline-Studio/core$ ./webui.sh --install --extra all$ ./webui.shOn Windows use .\webui.bat in place of ./webui.sh, or just double-click it.
Frequently asked questions
New to Inline Studio? Read the getting-started guide →
Is Inline Studio free?
Yes. Inline Studio is free and open source under the GPL-3.0 license. There is no paid tier to use it, and nothing to buy.
Do I need a GPU?
Local generation runs on your own machine with the built-in Inline Core engine and Z-Image Turbo, so a GPU makes it much faster. If you would rather not run models locally, API Nodes reach hosted models that run in the cloud and need no GPU at all.
Is there a desktop app to install?
No. Inline Studio runs as a local app: one process serves a web UI that you open in your browser. You start it from the source on your own machine (Python 3.11 or newer), so there is no installer to download.
Can I add my own nodes?
Yes. Install community extensions from the registry in one click, or paste any GitHub repo URL. Every install is security-scanned, and each entry links back to its code. You can build your own with the extension guide.
Can I keep a character or style consistent across shots?
Yes. Train your own model (a LoRA) on a handful of images to lock in a character's face, a custom visual style, or a recurring look, then use it on every shot so they match. It trains on your own GPU on the Trainer canvas, and Krea 2 training fits a 16GB card.
Can I train my own LoRAs?
Yes. Inline Studio has a built-in LoRA trainer on a dedicated Trainer canvas. Create a dataset, auto-caption it locally, set a trigger word, then wire Load Dataset into Caption into Train LoRA to run the job on your own GPU, with a live loss curve and resumable checkpoints. Trained LoRAs save to your models/loras folder, ready to use on the generation canvas. Training needs a GPU and the training extra installed.
Which base models can I train LoRAs for?
Z-Image Turbo and Krea 2. For Z-Image, train with the Turbo training adapter to keep its 8-step speed, or in De-Turbo mode with nothing extra to download. For Krea 2, train on the undistilled RAW base and generate with Krea 2 Turbo, the same LoRA carries over unchanged.
Can I train a Krea 2 LoRA on a 16GB GPU?
Yes, at 512px with the 4-bit base. It peaks around 12GB, measured on a Tesla T4, so it fits a 16GB card. Krea 2 at 1024px needs roughly 32GB, so train at 512 (a LoRA trained at 512 still applies at any generation resolution). Z-Image LoRAs train at 512 in about 13GB.