Inputs
Says what it needs
A picture, a 3D model, a part, a film, sound, or words. A tool is only offered work that matches what it says it needs.
What's inside
— tools. — ways to use them.
This is the studio's public list of its tools. For each one it says what you give it, what it makes for you, and whether it is ready. The studio checks every tool before any work starts, so a tool that isn't ready says so here instead of failing halfway through your film. Where a tool can run in two ways, both ways are listed.
| Engine | Flow | Runtime | Configs | State |
|---|---|---|---|---|
| env_probe probe | metadata → metadata | linux-process | 1 · 1 bound | probe |
| glbcoder_text_to_glb | text → mesh | llama.cpp build-vk (Vulkan), llama-server --jinja, one stream, plus headless Chromium (ANGLE/SwiftShader) to press the page's Export GLB button; loopback only | 1 · 1 bound | bound |
| skintokens_tokenrig_autoregressive | mesh → mesh | python 3.11 + torch 2.14.1+rocm7.2 (gfx1151) + transformers 4.57.1 + bpy 4.2, SDPA instead of flash-attn | 1 · 1 bound | bound |
| trellis_cpp_image_to_3d_cuda_remote | image → mesh | trellis.cpp 0.8.1 CUDA on appdev1 (RTX 4090, 24 GB) driven over ssh by scripts/trellis_remote.py; trellis2-fp16 weights sha256-identical to this box (10/10 files, 16 GB), verified 2026-11-07 | 1 · 1 bound | bound |
| trellis_cpp_image_to_3d | image → mesh | trellis.cpp 0.8.1 prebuilt Vulkan (ggml 0.24.0), 10 GGUF weights 16 GB | 1 · 0 bound | unbound |
| trellis_cpp_image_to_3d_rocm | image → mesh | trellis.cpp 0.8.1 prebuilt ROCm (ggml 0.24.0 hip), 10 GGUF weights 16 GB | 3 · 0 bound | unbound |
| zero_to_cad_qwen3vl_2b | image → cad | python3.13 + torch 2.15 cpu + transformers 5.22 | 1 · 0 bound | unbound |
| frames_to_video | frames → video | ffmpeg · no model in the loop | tool — no adapter | ready |
| text_to_music | words → music, voices & sound | ACE-Step (music) · Dia-1.6B + Qwen3-TTS (voices) · the studio's own + CC0 sound effects — all Apache-2.0 / CC0 | tool — built in | ready |
Checked at: — · — ways
set up, — ready to use · tools present:
—/—.
This list is made by checking the studio's computers, not typed by hand: a way
that isn't ready stays listed and simply won't be given work, rather than
failing halfway through. Ready means it can run right now.
Two ways of making things are built-in tools — turning pictures into film
(ffmpeg) and making sound (ACE-Step for music, Dia and Qwen3-TTS for voices) —
and the first row is just a health check, not a way of making things. Making
pictures from words is handled separately, by ComfyUI (FLUX.2 klein 4B), and
is not counted above.
Our promise
What a tool must tell us before it can make things for you.
A tool only gets work once all three are known. Anything missing, and it waits.
Runtime
Written down exactly
The exact program, version, and settings are written down for every tool, so what you made can be made again the same way, months later.
State
Ready — or it says why not
If a tool can't run, it says so when we check — never quietly in the middle of your work. The list above shows it, and your work is never sent to a tool that isn't ready.
Want something made for the way you work?
If part of your work eats hours — shaping models, making characters move, opening pictures, sound — that is exactly the shape of thing this studio is for.