Comfyui_TTP_Toolset

TTPlanetPig/Comfyui_TTP_Toolset is ranked 11,923 among these 8 peers; vercel-labs/tersa is ranked 11,920.

**ComfyUI TTP Toolset** ComfyUI TTP Toolset is a collection of tiled upscale, Smart Tile, video sampler, coordinate, and conditioning helper nodes. The current main feature is **Smart Tile 2. 0**, an object-aware tiled img2img workflow with automatic layout, per-tile prompts, loop processing, semantic priority, mask-aware assembly, and final-only saving. Instead of cutting every image into equal squares, it lets you build variable-size tiles around visual structure: face, eyes, hands, text, subject, clothing, foreground details, and

background/context regions. Smart Tile is designed for workflows where small areas need more detail than the full image. You can draw or auto-detect tiles, process one tile at a time through your sampler, then assemble the final image with feathering, masks, semantic priority, color correction, and optional pixel alignment. **What It Solves** Avoids cutting important objects through the middle when a simple grid would split a face, hand, or text. Allows different tile sizes, so a face tile

can be small and high detail while the background stays large. Sends one tile at a time through img2img, so the number of tiles can be 4, 8, 16, or anything produced by the editor. Keeps high-detail focus tiles visible during assembly with priority, layer, and semantic ranking. Supports mask-aware pasteback, soft detail overlay, final-only assembly, and black-base canvas mode. **Visual Tile Editor** is the recommended starting point. It can load an image like ComfyUI's official

, or receive a connected . The editor stores the current layout in a hidden widget, so your workflow keeps the tile plan. Useful editor actions: : replace the whole layout with an even grid. : subdivide the selected tile without rebuilding the whole layout. / : toggle a colored overlay for tile object masks in the editor. / : preview the hardened and feathered replace-paste mask shape without changing saved masks. : turn painted regions into object tiles. : Shift/Cmd/Ctrl-select multiple tiles, then union their

cropped object masks back onto the selected tiles. : Shift/Cmd/Ctrl-select multiple tiles, then replace them with one larger tile; existing masks are unioned into the merged tile when available. : after manually moving sub-tiles from a masked tile, re-crop the inherited masks to the current tile boxes. : add background tiles for uncovered areas. 1 or QwenVL3 detection and write the detected layout back into the editor. 1 detection and keep only detected object tiles without automatic gap

filling. When subdividing a masked tile with , the Mask mode can crop the original object mask into child tiles or skip empty mask children. The child tiles keep a parent mask source, so can update them after manual edits. | | | Preview a tile set as a contact sheet or one selected tile. | | | Load a local QwenVL tagging model from . | | | Generate per-tile prompts before loop processing. | | | Filter selected tiles and replace, prepend, append, or find/replace per-tile prompts before sampling

. | | | Filter selected tiles and set composite mode, replace shape, layer, priority, occlusion priority, blend, and importance before assembly. | | | Put tiles