floyo logo
Powered by
ThinkDiffusion
⚡️Nano Banana 2 ⚡️ just landed. Start creating now.
floyo logo
Powered by
ThinkDiffusion
⚡️Nano Banana 2 ⚡️ just landed. Start creating now.

ComfyUI-TBG-ETUR

121

Last updated
2026-03-03

TBG Enhanced Tiled Upscaler and Refiner is a sophisticated tool designed for enhancing images within the ComfyUI framework, capable of upscaling images to resolutions as high as 200MP. It combines advanced processing techniques, including dual-model processing for structure and detail, as well as per-pixel denoising to allow for nuanced creative control over image refinement.

  • Dual-Model Processing: Utilizes separate models for structural integrity and detail enhancement, ensuring high-quality outputs.
  • Per-Pixel Denoising: Offers fine-grained control over the denoising process, enabling users to dictate where creative alterations are applied.
  • Generative Tile Fusion: Seamlessly merges tiles, even under high noise conditions, producing cohesive and artifact-free images.

Context

The TBG Enhanced Tiled Upscaler and Refiner serves as an advanced module within ComfyUI aimed at enhancing the quality of tiled image generation and refinement. Its primary purpose is to facilitate high-resolution workflows while allowing users to maintain control over various aspects of the image enhancement process.

Key Features & Benefits

This tool boasts several practical features that significantly improve the image enhancement workflow. Key functionalities include high-resolution image generation, advanced AI algorithms for image quality improvement, and user-friendly interfaces for both community and professional users. These features collectively enhance the efficiency and output quality of image processing tasks.

Advanced Functionalities

The tool introduces several advanced capabilities, such as Neuro-Generative Tile Fusion, which intelligently adapts the image generation process based on previously generated content. This allows for the creation of ultra-high-resolution images without compromising detail or introducing seams. Additionally, users can access per-tile prompts and denoising settings, allowing for targeted adjustments and refinements.

Practical Benefits

Incorporating the TBG Enhanced Tiled Upscaler and Refiner into a workflow streamlines the process of image enhancement, allowing for quicker iterations and higher quality results. The ability to refine specific tiles without regenerating the entire image saves time and resources, while advanced denoising options provide greater control over the final output.

Credits/Acknowledgments

The development of TBG Enhanced Tiled Upscaler and Refiner is credited to the TBG team, with contributions from various community members. The project is open-source and available under a suitable license, encouraging collaboration and ongoing improvement from users and developers alike.

Inner Nodes

ApplyRegionalConds, ConfigureModifiedFlux, CreateRegionalCond, EdgePadNode, FlowEditForwardSampler, FlowEditGuider, FlowEditReverseSampler, FlowEditSampler, FluxAttnOverride, FluxDeGuidance, FluxForwardODESampler, FluxInverseSampler, FluxKontextResolution, FluxNoiseMixer, FluxReverseODESampler, ImageComplexityMap, InFluxFlipSigmas, InFluxModelSamplingPred, IncrementBatch, MaskConditionalPaint, MaskGrayValueScaler, MaskMultiplyAdvanced, MaskMultiplyZeroWins, MaskToDenoiseInterpolator, MaskUpdateInpaintFromConditioning, OutFluxModelSamplingPred, PAGAttention, PrepareAttnBank, QwenEditResolution, RFDoubleBlocksOverride, RFSingleBlocksOverride, RegionalStyleModelApply, SEGAttention, TBG_masked_attention