Refine Spectrum guidance and standard workflow

This commit is contained in:
ukr8b3g-cmyk
2026-08-11 03:12:14 +09:00
parent 46a1b11dd3
commit b0949c3000
5 changed files with 328 additions and 233 deletions
+9 -2
View File
@@ -13,12 +13,19 @@ Setup:
Default profile:
- 3 chunks x 5.0 seconds
- Timeline prompt with three sections
- Scale Image to Total Pixels: area, 0.60 MP, 32-step rounding
- Scale Image to Total Pixels: area, 0.50 MP, 32-step rounding
- Continuity: Balanced, 22 frames
- Seam correction: Auto
- Report Detail: Basic
Quality note:
- In the current local A/B sample, Seam Correction Off produced the least noticeable visual boundary.
- Compare Auto and Off with the same seed; keep Off when Native Continuity already provides a clean join.
Reference test environment:
- NVIDIA GeForce RTX 5060 Ti 16GB
- 64GB system RAM
- Python 3.13.12, PyTorch 2.13.0+cu130, CUDA 13.0
- SageAttention 2.2.0+cu130, KJNodes FP8 CUDA++, compile disabled
The 0.60 MP profile is a starting point. If VRAM is insufficient, reduce the image scale to 0.30 MP.
The 0.50 MP profile is a starting point. If VRAM is insufficient, reduce the image scale to 0.30 MP.