SYSONLINE
RELv3.5.0
RACK9 PIPELINES
VRAM4.0-23.8 GB
VISION STUDIO v3.5.0 · FLUX.1 DEV + SCHNELL · SD 3.5 LARGE + MEDIUM · SDXL · SD 1.5 · LTX VIDEO · ANIMATEDIFF V1-5-2 · STABLE VIDEO DIFFUSION XT · LORA + CONTROLNET + IP-ADAPTER · CUDA 13.2 · METAL · LOCAL · FIRST
UTC10:25:38
S/N · VX-00001
MOD · FEATURES · 01S/N · VX-FEAT-0001 · v3.5.0INDEXED
PWR+12V+5V+3.3V-12VGND
EVERY · FEATURE · LINKS · TO · ITS · SOURCE
FEATURE REFERENCE·v3.5.0·OPEN SOURCE

Every control, labeled.

What Vision Studio does, grouped the way you work: generate, edit, animate, organize, source models, run graphs, tune the engine, and choose where each job runs. Each entry names the files in the public repository that implement it - open any SRC chip to read the code behind the claim.

VX-FEAT · REFERENCE
S/N · VX-FEAT-0001 · v3.5.0
MIT · SOURCE · LINKED
MOD · GENERATE · 028 FEATURESONLINE

Generate

Image and video models that run on your own hardware, and the controls that shape a render: sampler, steps, guidance, seed, framing, LoRAs, ControlNet and reference images.

Image models

IMAGE

FLUX.1 [dev] and [schnell], Stable Diffusion 3.5 Large and Medium, Stable Diffusion XL and Stable Diffusion 1.5, each run through its Hugging Face Diffusers pipeline.

  • VRAM listed in the model catalog: about 24 GB for FLUX.1, 12 GB for SD 3.5 Large, 8 GB for SDXL, 6 GB for SD 3.5 Medium and 4 GB for SD 1.5.
  • A model loads on first use and stays in memory for the session, so switching back to it does not reload it.

Video models

VIDEO

LTX Video and AnimateDiff generate a clip from a prompt; Stable Video Diffusion animates a still image you supply. Clips are written as MP4.

  • Clip length is frame rate times duration: 12 to 60 fps, 1 to 10 seconds.
  • AnimateDiff runs a motion adapter over Stable Diffusion 1.5.

Sampling controls

IMAGE

Prompt and negative prompt, eight samplers (Euler, Euler a, DPM++ 2M, DPM++ 2M Karras, DPM++ SDE, DPM++ SDE Karras, DDIM, UniPC), 1 to 50 steps, CFG from 1 to 20 and a fixed or random seed.

  • Every model starts from the same defaults, 25 steps at CFG 7.5; nothing is tuned behind your back.
  • Eight aspect ratios at three resolution tiers (512, 768 or 1024 px on the long edge), or a custom size from 256 to 2048 px a side.

ControlNet

IMAGE

Guide composition with edge, depth, pose, scribble and normal-map layers placed on the canvas.

  • SD 1.5: canny, depth, OpenPose, scribble and normal.
  • SDXL: canny, depth and OpenPose, plus scribble and normal through the Union model.
  • FLUX.1 [dev]: canny, depth and OpenPose through the Union model.
  • SD 3.5 Large: canny and depth.
MOD · EDIT · 037 FEATURESONLINE

Edit

Model-backed tools that change the pixels, run by the local engine on the image you are working on, plus the canvas layers that steer the next render.

Guided AI tools

EDIT

Background removal (U2-Net), AI upscale (Real-ESRGAN, general or anime, 2x or 4x), face restoration (GFPGAN v1.4), style transfer, generative fill, object removal, AI expand and background replace.

  • Fill, removal, expand, background replace and style transfer run as guided passes through the checkpoint you have selected.
  • Style transfer has six presets: Van Gogh, Monet, Ukiyo-e, Comic, Watercolor and Pencil Sketch.

Clone Stamp and Heal

EDIT

Retouch on the canvas, stroke by stroke. Clone Stamp copies one part of the image over another from a source you Alt+click; Heal blends texture into the colour and light around each stroke, and with no source it finds a matching patch nearby by itself.

  • Heal solves a Poisson blend over the stroke, so the patch has no visible seam.
  • Size, hardness and opacity per stroke; [ and ] change the brush size.
  • Strokes stay editable history: each one is an undo step, Before/After hides them, and they are burned into the PNG you export.

Adjustments, filters, undo and export

EDIT

Fifteen adjustment sliders and the look filters are applied to the real pixels, so the canvas shows exactly what you export. Every edit is undoable, and Export saves the flattened result as a new PNG in your library.

  • Up to 100 undo steps across layers, strokes, retouching, adjustments and filters.
  • Before/After shows the original without your edits.
MOD · STORY · 043 FEATURESONLINE

Storyboard and timeline

Plan a sequence as scenes, then cut generated clips together and export a finished MP4.

MOD · LIBRARY · 054 FEATURESONLINE

Library

Everything you generate lands in a local library you can search, branch from and export.

MOD · FOUNDRY · 065 FEATURESONLINE

Model Foundry

Find, vet and install models without leaving the app, with the risky formats locked behind an explicit consent step.

Download manager

LOCAL

Downloads run two at a time with live status. Direct links are accepted only over HTTPS from CivitAI and GitHub and are checked against a SHA-256 hash.

MOD · WORKFLOWS · 072 FEATURESONLINE

Workflows

Bring ComfyUI graphs in, or let a ComfyUI server you already run do the rendering.

External ComfyUI server

LOCAL

If a ComfyUI server answers at 127.0.0.1:8188, or at the address in the COMFYUI_URL environment variable, plain image jobs and image-to-video jobs are sent to it instead of the built-in engine. Text-to-video always runs on the built-in engine.

  • The hand-off uses standard checkpoint file names and does not pass LoRAs. Image-to-video on that path becomes the ComfyUI SVD-XT workflow, which does not receive the model, prompt or clip length.
  • The engine reconnects in the background when ComfyUI starts or drops, and a job ComfyUI does not accept runs on the built-in engine instead. To keep every job on the built-in engine, run ComfyUI on another address.
MOD · ENGINE · 084 FEATURESONLINE

Engine and performance

A local Python service does the generating on your hardware, and a Performance panel decides how hard it pushes.

Local generation engine

LOCAL

A bundled FastAPI service runs PyTorch 2.14.1 and Hugging Face Diffusers on 127.0.0.1:8000, reachable only from your own machine. The app and the engine prove each other with an HMAC handshake at launch, and every HTTP and WebSocket call carries the session token, so another local program cannot drive your GPU or read your outputs. It uses the first NVIDIA GPU (Turing or newer, through CUDA 13.2), Apple Silicon through Metal (MPS), or the CPU.

  • The engine ships inside every installer as a hash-locked bundle that checks itself at startup; Python is not needed on your machine.

Performance panel

LOCAL

Auto, On or Off for SDPA attention, channels-last memory, torch.compile, quantization, attention slicing and TensorRT, behind one master switch. Auto decides per model and GPU, and the panel shows what was applied, skipped or fell back.

  • Quantization (optimum-quanto) ships in every build. TensorRT ships in the Windows and Linux builds, but Auto leaves it off until a model family passes verification, and the panel says so; you can still switch it On.

Memory fallback

LOCAL

A model that only fits with offload loads with CPU offload and VAE tiling, and an out-of-memory error steps down a fallback ladder instead of failing outright. Precision follows the hardware: bf16 where the GPU supports it, otherwise fp16 (fp32 for FLUX and SD 3.5, which are not run in fp16), and fp32 on the CPU.

Jobs that survive a crash

LOCAL

Generation jobs are written to the local database, so an engine restart keeps their history and results, and a job it interrupted is marked as such rather than lost. If the engine crashes, the app restarts it after 2 seconds, then 4; a third crash within five minutes stops the restarts and the app says so, and you start the engine again from Settings once the cause is fixed.

MOD · ROUTING · 095 FEATURESONLINE

Privacy and provider routing

Local by default. A cloud provider only when you add your own key and choose it, one job type at a time.

Local by default

LOCAL

Prompt tools, still images and video all start on the Local route. The interface is only allowed to connect to the engine on your own machine, and the code contains no analytics, telemetry or crash reporting.

  • The interface can reach files only inside your output folders and the files you pick: every file call resolves real paths and refuses traversal and symlink escapes, every channel to the app core is typed and validated, and a strict Content Security Policy blocks inline scripts.

Network use

LOCAL

With every route on Local, the app's code goes online for model searches and downloads you start and for the update check, which asks updates.vision-studio-x.com for a newer release 15 seconds after launch and every four hours on Windows and Linux. A saved OpenRouter key also lets Settings fetch that account's model list.

  • A found update is never downloaded on its own: Download update in Settings > About starts it. On Windows the updater installs only builds signed by the expected publisher.
  • Set the environment variable VISION_STUDIO_DISABLE_UPDATES=1 to switch the update check off.
MOD · WORKSPACE · 103 FEATURESONLINE

Workspace

Eight panels, reachable from the keyboard, with desktop notifications when work finishes.

DOWNLOAD · FREE
VX-FEAT · REFERENCE
S/N · VX-FEAT-1000
MIT · SOURCE · LINKED