fix: video RoPE clock uses the render fps (hardcoded 24 at 25fps renders = 4%/s AV rope skew = the ~10s lip-sync drift)

credits: JD Joy Future Academy, Lightricks, TenStrip (strong-mode sigma ladder), RealRebelAI, Comfy-Org, Google

hires v2: per-shot shared noise fields (kills the window-boundary texture switch) + tenstrip 4-step upscale ladder as 'strong' mode

fix: cast float32 stragglers under quantization policies (plain-distilled-1.1-derived files store F32 tables; fp8_scaled_mm load crashed with Float x BFloat16) + sync drifted files

audio memory default ON: cross-shot voice consistency is the point of multishot; sync/voice tradeoff documented in tooltip

enable_audio_memory default OFF (every perfect-sync render ran without it)

head_trim_frames default 14 (production-proven; shot-start sync fix)

v1.4 final: operator node layout

v1.4: LoRA Stack node (dropdown+strength UI, chainable), shipped wired into the loader

v1.4: multiline lora_path box; operator's node layout; widget repair

v1.4: multi-LoRA loading; shipped Reference Image node; instructions updated

Full step-by-step instructions (reference image, LoRAs, troubleshooting)

Add demo video (rendered with this build)

v1.3: VHS glitch reaches the master (worker port + transition sidecar); v2a_grad_scale note corrected

v1.3: VHS glitch reaches the master (worker port + transition sidecar); v2a_grad_scale note corrected

v1.3: VHS glitch reaches the master (worker port + transition sidecar); v2a_grad_scale note corrected

RefPicker: honor a scene's JSON refs block without requiring the folder name in the prose (v1.2)

Changelog: split v1.2 (api_key local endpoints, refs force-inject) from v1.1 (workflow release)

LLMEnhance: local endpoints no longer require an api_key (placeholder sent automatically); clearer error + tooltip; document the field in the workflow README

Lead with the workflow: retitle card, workflow section first, patch as 'under the hood'; add workflow/video tags

v1.1: A/V sync fix, AutoFinish nodes, gemma_file dropdown, manual_path, framing/movement doctrine, ready-to-run multishot workflow

Lazy text-encoder load: build only on conditioning-cache MISS (saves 21-33GB host RAM + load time on cache-HIT runs)

hires-fix: 33f windows + zero resident blocks during the pass (VAE encoder fatal abort at 65f/1920x1088 with 12GB pinned)

fix: _ref_sched unbound on no-reference runs (no-character scripts)

encoder_fp8: strip unused vision tower + absolute-headroom fits check -> GPU encode engages on 24GB cards

conditioning disk cache (encode once per script per machine) + CPU-fp8 encoder warning

README: document memory-trim fix (critical), fp8_scaled_mm, resident_blocks, encoder_fp8, hires-fix, ref scheduling, robustness; full file list + no-cherry-pick warning

per-device capability probe: pre-Ada GPUs auto-fall back to upcast path

fp8_scaled_mm: clear sm_89+ gate (RTX 30-series lacks fp8 cores)

harden pipeline dtype derivation against fp8 params (2 sites)

harden pipeline dtype derivation against fp8 params

hires-fix second pass: per-shot latent re-encode + DMD tail re-denoise at target res (real detail synthesis; 65f windows, 3090-safe)

fix: scaled-mm swap ran on META model (0 linears matched); branch on weight dtype at call time

guard: fp8-compute toggles require bf16 source file (fp8 files load all-fp8 with cast skipped -> randn crash)

encoder_fp8: halve the Gemma wrapper footprint (fp8 storage, upcast at encode; enables GPU encode pass on 32GB)

torch._scaled_mm fp8 Linear forward (measured x2.8 kernel / x1.5 e2e vs bf16 on RTX 5090)

QuantizationPolicy.fp8_scaled_mm_torch (torch-native, no tensorrt_llm)

fp8_scaled_mm mode (native fp8 matmul, no upcast tax) + resident_blocks partial offload

FIX: memory bank trim was a no-op when max_size<=num_fix_frames (tail[-0:] returns whole list) -> bank grew unbounded

RefPicker: re-entry ref scheduling (3+ shot absence) + script-carried ref pinning via refs map

Generate: schedule-aware ref dedup, cap 6 scheduled entries

TextEncode: hot-swap encoder to GPU for the encode pass (fits-check + OOM fallback)

README: document fp8 gemma scale-key layout fix

fp8 gemma: accept .weight_scale layout + zero-match warning

per-character reference scheduling: RefPicker emits ref_shots (first-appearance shot per character), Generate injects each ref at its character's entry shot with the long trim - multi-character single-run videos get every character properly seeded

load order: DiT (quantize/fuse peak) first, VAEs, text encoder LAST + gc between stages - encoder-first stacked 24GB under the fuse peak and caused an mmap access violation at VAE load on a 96GB box

references seed shot 1 only: from shot 2 the bank carries identity+staging; re-injecting a wide full-scene ref every shot made the subject relocate between shots

RefPicker auto mode: emit ONE ref per character named in prose (batch output) - solo scene 1 ref, two-character 2 refs, none 0; dialogue mentions still excluded; single picker now covers every queue case with zero mode switching

RefPicker prompt scan: parse JSON script payloads before scanning - on raw JSON the plain-quote stripper deleted the entire prompt body, so auto mode never matched a character on TXT-wrapped JSON blocks

RefPicker IS_CHANGED: guard None from unresolved wired inputs (per-item WARNING spam + lost caching)