Feat/claude images mcp #3029
+102
−0
There are no files selected for viewing
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,90 @@ | ||
| --- | ||
| title: "Create Images with Claude and Hugging Face" | ||
evalstate marked this conversation as resolved.
Show resolved
Hide resolved
|
||
| thumbnail: /blog/assets/claude-images-mcp/thumbnail.png | ||
| authors: | ||
| - user: evalstate | ||
evalstate marked this conversation as resolved.
Show resolved
Hide resolved
|
||
| --- | ||
|
|
||
| # Create Images with Claude and Hugging Face | ||
|
|
||
| > [!TIP] | ||
| > **TL;DR:** Its easier than ever to generate detailed pictures using the latest AI models - by connecting Claude to Hugging Face Spaces. This article describes how, why, and introduces recently launched models which excel at producing natural images and including text. | ||
evalstate marked this conversation as resolved.
Show resolved
Hide resolved
There was a problem hiding this comment. maybe link some example text to image spaces here? and perhaps we can emphasise that these models are SoTA. essentially clause is SoTA for text/ code and these models are SoTA for Image* - combined we make a beast. |
||
| ## Introduction | ||
|
|
||
| Recent advances in Image Generation models have improved their ability to produce realistic outputs and incorporate high quality text. It's easier than ever to use these models by connecting them directly to Claude. | ||
evalstate marked this conversation as resolved.
Show resolved
Hide resolved
|
||
|
|
||
| The advantages of generating pictures using this way are: | ||
| - The AI can assist in building detailed prompts that may improve the output quality of generated images. | ||
| - The AI can "see" the generated images, then help iterate on designs and techniques to get perfect results. | ||
|
There was a problem hiding this comment. this is not always true tho no? :p - specially not in the case of claude |
||
|
|
||
| To get started, create a free [Hugging Face account](https://huggingface.co/join), then connect Claude from its "Search and tools" menu in the chat input box. The video below shows the exact steps needed: | ||
|
|
||
| <figure class="image flex flex-col items-center text-center m-0 w-full"> | ||
| <video | ||
| alt="claude-auth-flow.mp4" | ||
| autoplay loop autobuffer muted playsinline | ||
| > | ||
| <source src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/claude-images-mcp/claude-auth-flow.mp4" type="video/mp4"> | ||
| </video> | ||
| <figcaption>Conneting Claude to Hugging Face</figcaption> | ||
| </figure> | ||
|
|
||
| Confirm that that the Hugging Face tools are available from the "Search and tools" menu... and now we are ready to try some of the latest models. | ||
|
There was a problem hiding this comment. maybe here we can add a small blurb about the HF MCP Server There was a problem hiding this comment. Somewhere here, I think it might be good to talk about ZeroGPU too i.e. they are able to use these chonky models thanks to our serverless GPU platform. There was a problem hiding this comment. have added some text; i didn't want the article to become too technical/explanatory though - wanted to keep it accessible to Claude users that might already be uncomfortable with some of the terminology. not sure i've managed it though :D |
||
|
|
||
| ## Natural Images with Flux.1 Krea Dev | ||
|
|
||
| [FLUX.1 Krea [dev]](https://huggingface.co/black-forest-labs/FLUX.1-Krea-dev) aims to eliminate the telltale "AI look" that often plague generated images - for example plastic skin, oversaturated colors, or overly smooth textures. If you want images that look like they were shot by a professional photographer rather than generated by a computer, Krea delivers the realistic textures, natural lighting, and authentic aesthetics that other AI models struggle with. You can read more about how they did this [on their blog](https://www.krea.ai/blog/flux-krea-open-source-release). | ||
|
|
||
| <figure class="image text-center"> | ||
| <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/claude-images-mcp/bfl_krea_image_wide.avif" alt="Examples from Krea"> | ||
| <figcaption>Examples from the Krea Image Generator</figcaption> | ||
| </figure> | ||
|
|
||
|
|
||
|
|
||
|
|
||
| To try **Krea**, go to [`huggingface.co/mcp/settings`](https://huggingface.co/settings/mcp) and add `evalstate/flux1-krea` to your "Spaces Tools". This will add a tool named something similar to "`flux1_krea_dev_infer`" to Claude. This enables Claude to generate beautiful, realistic looking images. Note the precise name may vary slightly depending on your configuration. | ||
evalstate marked this conversation as resolved.
Show resolved
Hide resolved
|
||
|
|
||
| <figure class="image flex flex-col items-center text-center m-0 w-full"> | ||
| <video | ||
| alt="adding-mcp-space.mp4" | ||
| autoplay loop autobuffer muted playsinline | ||
| > | ||
| <source src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/claude-images-mcp/adding-mcp-space.mp4" type="video/mp4"> | ||
| </video> | ||
| <figcaption>Adding a Space in <a href='https://huggingface.co/settings/mcp'>MCP Settings</a></figcaption> | ||
| </figure> | ||
|
|
||
|
|
||
| You can then try a prompt in Claude like: | ||
|
|
||
| > "Use Krea to create an image of a Vibrant garden with victorian house". | ||
| To begin generating pictures. | ||
|
|
||
| ## Qwen Image | ||
|
|
||
| [Qwen-Image](https://huggingface.co/Qwen/Qwen-Image) is a powerful AI image generator that excels at creating images with accurate text rendering, making it ideal for designing posters, signs, infographics, and marketing materials where text quality matters. Read more about the Qwen-Image model from their [blog post](https://qwenlm.github.io/blog/qwen-image/). | ||
|
|
||
| <figure class="image text-center"> | ||
| <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/claude-images-mcp/qwen_sample.jpg" alt="Examples from Qwen"> | ||
| <figcaption>Examples from the Qwen-Image Generator</figcaption> | ||
| </figure> | ||
|
|
||
|
|
||
| To use **Qwen-Image** add `evalstate/qwen-image` from your [MCP Servers setting](https://huggingface.co/settings/mcp) page, then check it is enabled in Claude to start using it. | ||
|
|
||
| You can even keep both enabled and ask Claude to use both and compare results for similar concepts - for example: "Use Krea and Qwen to generate a street scene with 'Hugging Face' graffiti sprayed on the wall". | ||
|
|
||
|
|
||
| <figure class="image text-center"> | ||
| <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/claude-images-mcp/krea_hf_example.webp" alt="Hugging Face Graffiti Street Scene"> | ||
| <figcaption>Hugging Face Graffiti Street Scene</figcaption> | ||
| </figure> | ||
|
|
||
|
|
||
|
There was a problem hiding this comment. just to spice things up a bit should we add the ltx-fast example as a nice ending too? :p I think it could work quite well. |
||
|
|
||
| ## Conclusion | ||
|
|
||
| Hugging Face Spaces paired with the Claude Connectors gives you simple access to the very latest AI models - often on the day of launch. With a Hugging Face account, it's simple to add new tools to Claude and use them straight away. | ||
|
There was a problem hiding this comment. would be nice to end more enthusiastically There was a problem hiding this comment. @evalstate - we can create a community discussion here and ask people to share their feedback and generations there? wdyt? and give them a to-do to try some image to video models with the generated images too. think it could be nice way to end. There was a problem hiding this comment. Request re-review, have made it more optimistic. vb - i was planning to follow up with an article on a) a LoRA image space and b) using Kontext/explaining the |
||
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Uh oh!
There was an error while loading. Please reload this page.