README
Extensions for Kubin
Description
Repository for storing "official" extensions for Kubin.
Initially, it contained only extensions that were somehow related to enhancing the capabilities of main Kubin functions (generating images with Kandinsky models), but it currently has a lot of stuff not directly related to Kandinsky (but necessarily connected with image generation). Some are outdated and not quite finished at all, and very few are actively maintained. Extensions are:
Name | Description |
|---|---|
| kd-animation | GUI wrapper for Deforum-Kandinsky |
| kd-bg-remover | GUI wrapper for RMBG and BiRefNet background removal models |
| kd-flux | Basic π€ diffusers-based implementation of Flux.1-Dev T2I pipeline |
| kd-image-browser | Tools for navigating output image folders |
| kd-image-editor | GUI wrapper for Filerobot Image Editor |
| kd-image-tools | Currently only allows image similarity search |
| kd-image-to-video | Based on VideoCrafter model |
| kd-interrogator | Contains GUI for single image/folder-targeted usage of image interrogation models: CLIP Interrogator and VLM-based captioners: CogVLM2, InternLM-XComposer2-4KHD, JoyCaption Pre-Alpha, JoyCaption Alpha One, JoyCaption Alpha Two, MiniCPM-V 2.6, Molmo-7B-O, PaliGemma 2, Pixtral-12B-Captioner-Relaxed, Qwen2-VL-7B-Instruct, Qwen2-VL-7B-Captioner-Relaxed |
| kd-kwai-kolors | Basic implementation of Kolors T2I pipeline |
| kd-llm-enhancer | Basic LLM-based prompt enhancer, based on π€ transformers and Ollama API |
| kd-mesh-gen | Basic implementation of Shap-E I23D pipeline |
| kd-multi-view | Basic implementation of Zero123++ "Image to Multi-view" pipeline |
| kd-networks | Allows the use of Kandinsky 2.2 LoRA for inference |
| kd-pipeline-enhancer | Mostly useless π€ |
| kd-pixart | Basic π€ diffusers-based implementation of PixArt-Sigma T2I pipeline |
| kd-prompt-styles | Enables auto-enhancing prompts with community-collected styles |
| kd-sana | Basic implementation of SANA T2I pipeline |
| kd-segmentation | GUI wrapper for Segment Anything. Was intended for auto-extraction of inpainting masks and custom ADetailer implementation for Kandinsky, but was not finished π |
| kd-stable-cascade | Basic π€ diffusers-based implementation of Stable Cascade T2I pipeline |
| kd-switti | Basic implementation of the T2I pipeline for Switti |
| kd-training | GUI wrapper for some Kandinsky training scripts (K2.1 fine-tuning/K2.2 LoRA) |
| kd-upscaler | Tools for upscaling, currently only Real-ESRGAN and KandiSuperRes are supported |
| kd-video | GUI for a consumer-friendly (24Gb VRAM) implementation of Kandinsky Video T2V/I2V pipelines. The low-VRAM pipeline for KV1.1 is still flawed and outputs noise π |
| kd-video-tools | Various tools for working with video files |
Installation
Clone this repo into the "extensions" folder in Kubin root. More info here.
Contributors
seruva19
122 commits
Similar repos
Languages
Python
99.2%
README
Extensions for Kubin
Description
Repository for storing "official" extensions for Kubin.
Initially, it contained only extensions that were somehow related to enhancing the capabilities of main Kubin functions (generating images with Kandinsky models), but it currently has a lot of stuff not directly related to Kandinsky (but necessarily connected with image generation). Some are outdated and not quite finished at all, and very few are actively maintained. Extensions are:
Name | Description |
|---|---|
| kd-animation | GUI wrapper for Deforum-Kandinsky |
| kd-bg-remover | GUI wrapper for RMBG and BiRefNet background removal models |
| kd-flux | Basic π€ diffusers-based implementation of Flux.1-Dev T2I pipeline |
| kd-image-browser | Tools for navigating output image folders |
| kd-image-editor | GUI wrapper for Filerobot Image Editor |
| kd-image-tools | Currently only allows image similarity search |
| kd-image-to-video | Based on VideoCrafter model |
| kd-interrogator | Contains GUI for single image/folder-targeted usage of image interrogation models: CLIP Interrogator and VLM-based captioners: CogVLM2, InternLM-XComposer2-4KHD, JoyCaption Pre-Alpha, JoyCaption Alpha One, JoyCaption Alpha Two, MiniCPM-V 2.6, Molmo-7B-O, PaliGemma 2, Pixtral-12B-Captioner-Relaxed, Qwen2-VL-7B-Instruct, Qwen2-VL-7B-Captioner-Relaxed |
| kd-kwai-kolors | Basic implementation of Kolors T2I pipeline |
| kd-llm-enhancer | Basic LLM-based prompt enhancer, based on π€ transformers and Ollama API |
| kd-mesh-gen | Basic implementation of Shap-E I23D pipeline |
| kd-multi-view | Basic implementation of Zero123++ "Image to Multi-view" pipeline |
| kd-networks | Allows the use of Kandinsky 2.2 LoRA for inference |
| kd-pipeline-enhancer | Mostly useless π€ |
| kd-pixart | Basic π€ diffusers-based implementation of PixArt-Sigma T2I pipeline |
| kd-prompt-styles | Enables auto-enhancing prompts with community-collected styles |
| kd-sana | Basic implementation of SANA T2I pipeline |
| kd-segmentation | GUI wrapper for Segment Anything. Was intended for auto-extraction of inpainting masks and custom ADetailer implementation for Kandinsky, but was not finished π |
| kd-stable-cascade | Basic π€ diffusers-based implementation of Stable Cascade T2I pipeline |
| kd-switti | Basic implementation of the T2I pipeline for Switti |
| kd-training | GUI wrapper for some Kandinsky training scripts (K2.1 fine-tuning/K2.2 LoRA) |
| kd-upscaler | Tools for upscaling, currently only Real-ESRGAN and KandiSuperRes are supported |
| kd-video | GUI for a consumer-friendly (24Gb VRAM) implementation of Kandinsky Video T2V/I2V pipelines. The low-VRAM pipeline for KV1.1 is still flawed and outputs noise π |
| kd-video-tools | Various tools for working with video files |
Installation
Clone this repo into the "extensions" folder in Kubin root. More info here.
Similar repos
Contributors
seruva19
122 commits
Languages
Python
99.2%