Bernini Unveils High-Resolution Tile Generation Solution
Emotional_Example_12 · reddit · 2026-07-16
This post introduces **Bernini**'s high-resolution tile generation solution, focusing on a complete redesign of the tile workflow for video and image processing. Key highlights include: - **Performance**: Generating 39 frames at 1920×1080 takes 325 seconds. - **New 3-Node Workflow**: Replaces the complex Tile Settings subgraph with **Tile Split / Tile Select / Tile Merge**. - **Tile Split**: Automatically splits based on the tile grid and aligns to the multiple of 16 required by Wan. - **Tile Select**: Extracts a specific tile by index while retaining all video frames. - **Tile Merge**: Seamlessly stitches tiles back together using feathered overlap and auto-detects dimension changes post-upsampling. - **Auto Padding/Cropping**: For instance, 1080 is padded to 1088 (the nearest multiple of 16) and cropped back post-merge, ensuring a complete round trip without pixel loss. Further optimizations include: - **Pixel-based Tiling**: Runs the full pipeline on each tile to prevent positional misalignment. - **Live Stitching**: Feeds generated results from adjacent tiles into the overlap area to reduce drift between tiles. - Updates to the installer, INT8 models, LoRA, CUDA auto-detection, Bernini paper features, and a prompt guide for 22 tasks.
Related event: Bernini Releases High-Resolution Tile Video Generation Workflow(2 posts)→
More from Multimodal
- ElevenLabs raises ElevenMusic free usage to 400 tracks a month — lukeharries · 2026-07-21
- Google Gemini now watermarks every AI video it generates — Sure_Belt9076 · 2026-07-21
- GPT Image 2 Prompt Turns Product Shots into Surreal Reality-Bending Ads — aziz4ai · 2026-07-21
- LTX 2.3 LoRA demo changes a video’s camera angle — CQDSN · 2026-07-21
- Reddit user shares a surreal ChatGPT-generated poster — Creamy-Sundae-9991 · 2026-07-21
- A cinematic SEEDANCE 2 prompt turns an empty sunrise city into a memory-driven video — LudovicCreator · 2026-07-21