digital-garbageandClaude Opus 5.5 05fde80caf feat: H3 decode logs where its time goes: picture vs sound seconds, VAE precision, and whether the VAE was fully on the GPU ...

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

2026-10-07 15:26:31 +03:00

..

conditioning

test: a bad-composition dislike teaches shot camera moves, styles and views

2026-10-07 14:44:03 +03:00

latent

feat: LTX 2.x model module — recognised from the checkpoint header, joint video+audio latent sized from the audio VAE, decode, text/image-to-video pipelines

2026-10-03 10:24:43 +03:00

loaders

fix: v4 regressions before the rental: taste judge trains under inference mode; LoRA keys unwrapped, misfits dropped, no-match said; unsupported dtype said; missed run finish read from history; prompt nudge no longer holds the model

2026-10-06 03:08:29 +03:00

media

feat: assign category to every remaining module

2026-09-15 09:46:58 +03:00

models

feat: H3 decode logs where its time goes: picture vs sound seconds, VAE precision, and whether the VAE was fully on the GPU

2026-10-07 15:26:31 +03:00

output

revert: H3 decode in one piece showed a tile grid (the decoder is trained on 256px tiles); back to ComfyUI's own tiling, tooltip says so

2026-10-07 15:01:07 +03:00

sampling

fix: Taste steering is named "(REINS)" and says it moves toward liked AND away from disliked, only after 2+ of each

2026-10-07 14:46:44 +03:00

system

fix: values typed during a node change survive its answer; preset/import releases the run wait; a group move outlives a queued save; corrupt captures don't block a new Generate

2026-10-06 10:14:33 +03:00

__init__.py

core+shell: the bridge, and the first module through it

2026-08-28 19:58:06 +03:00

_core.py

fix: LTX — tiled decode asked the way comfy asks it, picture found for ALG/DynaShift/guidance (shared av_video_stream), odd sizes said

2026-10-03 10:30:37 +03:00

.gitkeep

core: serve the app, with the module allowlist as the wall

2026-08-28 14:18:55 +03:00