Track exact stock-loader calls across equal strengths, disjoint targets and MODEL/CLIP-only slots. Preserve captured CLIP biases and emit stock-loadable CLIP export aliases.
Remove the Legacy optimizer registration, both WanVideoWrapper-specific nodes, obsolete bridge UI and bundled legacy example. Document replacements and add stock-loader/file-stack/export parity checks to CI. Existing workflows using removed nodes require migration.
Validate the staged tree: 1064 Python tests and 69 subtests passed; 3 CUDA-only skipped; JS migrations and real ComfyUI round trips passed. Keep the rejected H3 search prototype and unrelated plans out of this release.
- Additional notes (Optimizer): rewrote strategy_set and merge_refinement
tips to explain they're different toolboxes, not quality rankings
- Merge Formula note (782): added formula syntax explanation
- Merge Formula note (783): added description of what the node does
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Update Tips note to reference patch_compression, strategy_set, and
additive instead of old names. Clear stale cached output.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Rewrite the three main note nodes to be beginner-friendly:
focus on what the node does, when to use it, what to connect,
and good starting settings rather than technical internals.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add descriptions and tips to the 6 note nodes in the example
workflow covering: LoRA Optimizer standalone, AutoTuner standalone,
and the AutoTuner + Optimizer bridge workflow.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Saves TUNER_DATA to models/tuner_data/ as JSON so users can reload
past tuning results into Merge Selector without re-running the AutoTuner.
LoadTunerData uses a combo dropdown that refreshes like standard model loaders.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Ethanfel