by a-test-user-in-the-wrold
A locally runnable interface for Qwen-Image-2.1 with censorship filters removed, supporting both image generation and editing through a Gradio frontend. The Space exposes an MCP server endpoint, meaning builders testing tool-use workflows or agent pipelines can plug it in as an external image tool without writing additional wrapper code. If you are evaluating uncensored multimodal models for content that standard Qwen deployments refuse, this gives you a quick baseline with two model variants available side by side.
by AST-1320
Combining Qwen's image editing capabilities with FLUX.2 Klein models, the Space lets you make targeted edits to images and then run them through a FLUX.2 Klein enhancement pass. The pipeline draws on several vision-language models alongside age classification and internvl components, giving builders a concrete environment to see how chained image editing and generative refinement interact. If you are exploring lightweight enhancement workflows that combine instruction-based editing with diffusion-based upscaling or refinement, this is a reasonable place to test that kind of composition.
by hugging-apps
BFS Best Face Swap uses the BFS LoRA on top of Qwen-Image-2.1 and its Viggle turbo variant to perform head swapping on images. Builders exploring identity transfer or character consistency workflows can use this to see how a fine-tuned LoRA handles the task compared to a base model, and the included MCP server support makes it straightforward to wire into an agent pipeline for testing.
by kulkas2pintu
Wan2.2 14B Preview takes an input image and a text prompt and produces a short video using Wan2.2 I2V models, including a lightning-distilled variant for faster inference. The pipeline also incorporates RIFE for frame interpolation, making it a reasonably complete image-to-video stack to poke at. Builders exploring Wan2.2-based workflows or comparing the standard and lightning model variants have a concrete starting point here, with MCP server support available if they want to wire it into a larger agentic setup.
by arudradey
A Gradio interface running on ZeroGPU that lets you prompt Qwen-Image-2.1 through an uncensored GGUF variant, giving you outputs without the content filtering applied in the base model. The MCP server tag suggests it exposes an API endpoint alongside the UI, which is worth checking if you need to integrate image generation into a broader toolchain. Builders curious about behavioral differences between the standard Qwen-Image-2.1 weights and the uncensored fork have a direct way to compare outputs without any local setup.
by hugging-apps
Qwen-Image-2.1 provides a Gradio interface for generating and editing images using Qwen's latest image model, with a content safety guard layered in via hfmlsoc/ncii-guard-v02. It also exposes an MCP server endpoint, which means builders experimenting with tool-use pipelines or agent workflows can wire image generation directly into an MCP-compatible client without writing a custom wrapper.
by Pepe104
Pepe104's Space combines MiniMax H3 Turbo with LoRA fine-tuning to generate video alongside a synchronized soundtrack, pulling from a small cluster of related MiniMax model variants. Builders curious about pairing audio and visual outputs in a single generation pipeline can use it as a reference point for how LoRA weights integrate with the base H3 Turbo architecture.
by ravenrose996688
Krea 2 Turbo bundles text-to-image generation and image editing into a single Gradio interface, drawing on multiple model variants including a dedicated identity-edit model and several LoRA options. Builders exploring the Krea 2 model family can use this space to compare outputs across configurations without setting up separate pipelines locally.
by kulkas2pintu
A collection of Qwen Image Edit LoRAs is demoed here through a Gradio interface, letting you run targeted image editing tasks across styles including anime conversion, manga tone, noir comic panels, Pixar-inspired 3D, polaroid photo, upscaling, and multi-angle generation. Each LoRA is independently selectable, so a builder can directly compare how different fine-tunes handle the same input image without switching between separate repos or inference setups. If you are evaluating which Qwen-based editing LoRA fits a specific production use case, this is a practical way to audit output quality across more than a dozen style variants in one place.
by pnemrow
A demo interface for testing a curated collection of Qwen Image Edit LoRAs, pulling together over two dozen specialized models covering relighting, face swapping, pose control, skin detailing, upscaling, and style transfer into a single Gradio-based tool. Builders experimenting with Qwen-based image editing pipelines can use it to compare outputs across multiple LoRAs in one place rather than spinning up separate environments for each. The MCP server tag also makes it worth a look for anyone exploring how to wire image editing workflows into agent or tool-use setups.
by Saravutw
Omni Video Custom-fast-motion is a Gradio-based tool that accepts text prompts or images as input and generates video output, with support for extending existing video clips. Builders working on multimodal pipelines may find it worth a quick session to compare how the text-to-video and image-to-video paths handle the same subject, or to test how the video extension feature holds up across a short sequence.
by multimodalart
MiniMax H3 generates video alongside a synchronized soundtrack, combining MiniMax's H3 model with an audio component to produce video and sound together in a single output. Builders working on multimedia generation pipelines can use it to explore how video and audio synthesis can be coordinated rather than handled as separate post-processing steps. It runs through a Gradio interface, making it straightforward to test inputs and inspect outputs without writing custom inference code.
by microsoft
Mage-VL runs Microsoft's 4B parameter multimodal model directly in a Gradio interface, handling both image and video understanding through codec-native processing. Builders working on pipelines that need a single model to reason across static and moving visuals can use this space to test how the model handles video inputs without requiring separate preprocessing steps.
by jasfn
A Gradio interface wrapping LTX 2.3 that converts a source image into a short video clip with native audio output baked in. Builders working on multimodal pipelines may find it worth testing to see how the native audio integration behaves alongside the video generation, since that combination is less common in typical image-to-video setups.
by kulkas2pintu
Wan2.2 14B Fast Preview takes an input image and a text prompt and generates a short video using Wan-AI's 14B image-to-video diffusion model, with RIFE handling frame interpolation for smoother output. It also exposes an MCP server interface, which makes it worth a quick look for builders who want to wire image-to-video generation into an agent pipeline or tool-calling workflow rather than just testing it through a standalone UI.
by Sneak-Moose
Pro Realism Edit Studio is an image editing tool built by Sneak-Moose that accepts one or two input images and applies edits using a stack of several Qwen-based vision-language models. The multi-model setup, which includes both general-purpose and fine-tuned variants like lovis93/next-scene-qwen-image-lora-2509 and Phr00t/Qwen-Image-Edit-Rapid-AIO, makes it a practical testbed for comparing how different Qwen checkpoints handle the same editing task. Builders experimenting with instruction-driven image editing pipelines could use it to quickly evaluate prompt behavior across model variants before committing to a single backbone.
by prithivMLmods
A collection of Qwen Image Edit LoRAs is assembled here into a single Gradio interface, letting you apply named styles like Noir Comic Book Panel, Pixar-Inspired 3D, Polaroid Photo, Midnight Noir Eyes Spotlight, and Manga Tone to images through a common editing pipeline. Builders working on image transformation workflows can use it to quickly compare how different LoRA checkpoints handle the same input, covering a range from anime conversion and style transfer to upscaling and portrait enhancement. If you are evaluating which specialized fine-tune to integrate into your own pipeline, having roughly twenty distinct LoRAs accessible in one place cuts down on the time spent spinning up separate environments for each.
by selfit-camera
Omni Image Editor bundles several common image manipulation tasks into a single Gradio interface: text-to-image generation, image editing, upscaling, and watermark removal. The NSFW detection model from Falconsai runs alongside these tools, which gives builders a concrete reference point for how content filtering can be integrated into a multi-feature image pipeline. Developers building their own generation or editing workflows may find it worth testing to see how the moderation layer behaves across the different output types.
by ideogram-ai
Ideogram 4 runs state-of-the-art open-weights image generation through a Gradio interface, offering both fp8 and nf4 quantized model variants alongside a vision-language component built on Qwen3-VL. Builders evaluating quantization tradeoffs for deployment can directly compare how the two precision levels affect output quality without setting up their own inference pipeline.
by webml-community
Bonsai Image WebGPU runs image generation entirely in the browser using WebGPU, drawing on 1-bit and 2-bit quantized 4B-parameter models to keep compute requirements low enough for client-side inference. The quantization approach is worth examining closely if you are exploring how far model compression can go before output quality degrades. Builders working on edge or offline deployment pipelines can use this as a reference point for what ternary and binary weight schemes look like in a real, running system.
by r3gm
Wan2.2 14B Fast Preview takes an input image and a text prompt and outputs a generated video, running on the Wan2.2 I2V 14B model with RIFE for frame interpolation. The space is also set up as an MCP server, so builders integrating image-to-video generation into an agent pipeline can test tool compatibility directly rather than wiring it up manually first.
by TencentARC
Pixal3D converts a single image into a 3D asset by combining depth estimation, background removal, and mesh generation through a pipeline that draws on models from Microsoft, Meta, and Black Forest Labs. The output maintains pixel-level alignment with the source image, which is worth testing if you are evaluating how well current image-to-3D methods preserve surface detail and texture fidelity. Builders working on 3D content pipelines or scene reconstruction tools will get a clear sense of where this approach holds up and where it breaks down.
