A mesh node's result is a file the gate can fetch
CI / test (push) Successful in 8m59s

Stable Fast 3D's own save node hands back the GLB as base64 in its UI
output, which the gate does not read. LLMGWSaveMesh writes each mesh to
output/ as a GLB and names it, as a picture's save node does. The
prepare node also gives the object's mask now, which SF3D's sampler
takes.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
This commit is contained in:
Ilker ManapandClaude Opus 5.5 committed 2026-10-10 22:29:34 +02:00
1 parent b1490dbf30
commit 9b8ec89247
2 files changed
+63 -6

No files matched your search

+19 -1
View File
@@ -1126,7 +1126,8 @@ EU-safe textured path — its ComfyUI node and VRAM need are to be
checked), **Direct3D-S2** (MIT, DreamTech; high-detail shape, has a
ComfyUI node) if TRELLIS.2's detail falls short. Not in: TripoSR and
InstantMesh (permissive but superseded), Stable Fast 3D (Stability's
community licence, revenue-capped), SAM 3D Objects (Meta's own licence,
community licence, revenue-capped; served all the same since 2026-10-10
at the user's request, see G3), SAM 3D Objects (Meta's own licence,
to be read before it is considered). Steam's disclosure form (changed
2026-01-16) lists pre-generated 3D models as content that must be
declared on the store page: the licence notice G3 writes into the GLB is
@@ -1201,6 +1202,23 @@ newer). TripoSR (MIT) proves the chain on cuda01: a .glb from a picture
in about 30 s. Left: background removal before TripoSR (BiRefNet, MIT,
as a node), TRELLIS.2 or Direct3D-S2 on an Ampere card with an ungated
encoder, a missing node refusing the replica by name.
Added 2026-10-10 at the user's request, beside the order above: **Stable
Fast 3D** as `sf3d` (Stability AI Community License: free under US$1 M
annual revenue, an enterprise licence above, "Powered by Stability AI"),
the textured mesh that runs on Turing: about 1 s a mesh warm in fp16,
UV-mapped GLB with its texture, 18 GB of the card at peak. Its weights are
gated, so they were fetched once with the user's token into
/srv/llmgw-comfy/stable-fast-3d and linked into the engine user's
Hugging Face cache, which its loader node reads by repository name; no
token is on the machine at runtime. Its own save node returns the GLB as
base64, which the gate cannot fetch, so `LLMGWSaveMesh` writes the file;
`LLMGWPrepareForMesh` now also gives the object's mask. Left: its two
CUDA extensions (texture_baker, uv_unwrapper) are not on PyPI, so they
were built on cuda01 and installed into the engine's venv by hand; a
custom node's requirements need a way to name a local build, or the
engine install must carry them, before a customer site gets this. And
rembg's default session now fetches BRIA RMBG-2.0 (non-commercial):
every workflow must name its rembg model (ours say u2net).
- Catalog: a model makes `pictures` or `meshes` (`MakesImages` becomes an
output kind; `ServesComfyUI` stays). 3D weight folders in the gate's
`modelFolders`; the workflow's placeholders gain `image` (the input
+44 -5
View File
@@ -55,7 +55,8 @@ class LLMGWPrepareForMesh:
"optional": {"mask": ("MASK",)},
}
RETURN_TYPES = ("IMAGE",)
RETURN_TYPES = ("IMAGE", "MASK")
RETURN_NAMES = ("image", "mask")
FUNCTION = "prepare"
CATEGORY = "llmgw"
@@ -80,14 +81,52 @@ class LLMGWPrepareForMesh:
sq = _fit(rgba, foreground_ratio).astype(np.float32) / 255.0
a = sq[..., 3:4]
img = sq[..., :3] * a + (1.0 - a) * background
out.append(torch.from_numpy(img))
out.append(torch.from_numpy(np.dstack([img, a])))
# One size for the batch: the first picture's.
side = out[0].shape[0]
out = [o if o.shape[0] == side else torch.from_numpy(
np.array(Image.fromarray((o.numpy() * 255).astype(np.uint8)).resize((side, side)))).float() / 255.0
for o in out]
return (torch.stack(out),)
both = torch.stack(out)
# The picture on its background, and the object's alpha as a mask
# (1 where the object is) for a model that wants it, as Stable Fast
# 3D's sampler does.
return (both[..., :3].contiguous(), both[..., 3].contiguous())
NODE_CLASS_MAPPINGS = {"LLMGWPrepareForMesh": LLMGWPrepareForMesh}
NODE_DISPLAY_NAME_MAPPINGS = {"LLMGWPrepareForMesh": "Prepare a picture for a 3D model (llmgw)"}
class LLMGWSaveMesh:
"""Writes each mesh a 3D node made to output/ as a GLB and names the
files, which is how the gate finds a result. A model's own save node
may return the mesh as base64 (Stable Fast 3D's does) or as an OBJ
without its texture, and the gate reads neither."""
@classmethod
def INPUT_TYPES(cls):
return {"required": {"mesh": ("MESH",), "filename_prefix": ("STRING", {"default": "mesh"})}}
RETURN_TYPES = ()
OUTPUT_NODE = True
FUNCTION = "save"
CATEGORY = "llmgw"
def save(self, mesh, filename_prefix="mesh"):
import os
import folder_paths
import trimesh
from trimesh.exchange import gltf
folder, name, counter, subfolder, _ = folder_paths.get_save_image_path(
filename_prefix, folder_paths.get_output_directory())
saved = []
for i, m in enumerate(mesh if isinstance(mesh, (list, tuple)) else [mesh]):
file = f"{name.replace('%batch_num%', str(i))}_{counter + i:05}_.glb"
scene = m if isinstance(m, trimesh.Scene) else trimesh.Scene(m)
with open(os.path.join(folder, file), "wb") as f:
f.write(gltf.export_glb(scene, include_normals=True))
saved.append({"filename": file, "type": "output", "subfolder": subfolder})
return {"ui": {"mesh": saved}}
NODE_CLASS_MAPPINGS = {"LLMGWPrepareForMesh": LLMGWPrepareForMesh, "LLMGWSaveMesh": LLMGWSaveMesh}
NODE_DISPLAY_NAME_MAPPINGS = {"LLMGWPrepareForMesh": "Prepare a picture for a 3D model (llmgw)",
"LLMGWSaveMesh": "Save a mesh as GLB (llmgw)"}