[GH-ISSUE #8607] Feature: Offline-Ready Image Generation (Prompt-to-Image, Image-to-Image, In-Painting, Out-Painting) with Extended Model Support #85931

New Issue

2026-05-15T10:28:15-05:00

GiteaMirror commented

2026-05-15 10:28:15 -05:00

Originally created by @arudaev on GitHub (Jan 16, 2025).
Original GitHub issue: https://github.com/open-webui/open-webui/issues/8607

Is your feature request related to a problem?

There used to be a tool called OpenAI Labs, which offered cool features like Variations and Out-Painting. However, it is being sunsetted, forcing me and many other AI enthusiasts to search for alternative tools. The available paid tools are often subpar, striving for overly realistic results or adopting a chat interface similar to DALL·E 3 in ChatGPT. Additionally, HuggingFace models are frequently unstable, appearing and disappearing on a weekly basis. Currently, there is no environment that seamlessly integrates Prompt-to-Image, Image-to-Image, In-Painting, and Out-Painting operations. This gap necessitates the use of separate tools or services, complicating workflows and increasing setup overhead.

Describe the solution you'd like

I propose adding an offline-capable image generation module to Open WebUI. This feature will seamlessly blend text-based and visual AI, enabling users to perform a comprehensive range of image operations within a single interface. Key aspects include:

Unified Visual + Text AI
- Consolidate text and image operations under one platform, preserving Open WebUI’s simplicity.
Comprehensive Image Generation Features
- Prompt-to-Image: Create images directly from text prompts.
- Image-to-Image: Transform existing images based on new instructions.
- In-Painting: Fill missing or modified areas in an image.
- Out-Painting: Extend image boundaries seamlessly.
Extended Model Support
- HuggingFace: Integrate open-source models for offline usage.
- OpenAI DALL·E 2/3: Provide connectivity to external APIs for users who prefer or require online capabilities.
- Kaggle: Offer a path to pull or use models from Kaggle notebooks/datasets.
Resource Allocation & Management
- Allow configurable GPU/CPU usage, ensuring performance at both large and small scales.
- Support multiple concurrent requests without degrading the user experience.
Modular Architecture
- Maintain a separate, well-structured image-generation module to minimize impact on the core codebase.
- Facilitate future expansions (e.g., multi-GPU setups, advanced style transfers) by isolating image-processing logic.
Documentation & Examples
- Provide clear instructions for setting up and using the image-generation features.
- Offer sample scripts and best practices to help users get started.

Describe alternatives you’ve considered

External Tools:
- OpenOutpaint
- The better version: Stable Diffusion Infinity
- DALL·E 2 in Labs (that is being sunset this year)
- The best Image Enhancer
- And many other tools....

Additional Context

Why This Matters

Streamlined Workflows: Combining text and image capabilities in one interface saves time and fosters creativity.
Community Growth: Robust, multi-modal functionality can attract diverse users and contributors.
Future-Proofing: Aligns with Open WebUI’s commitment to extensibility and offline readiness.

Proposed Implementation Approach

New image-generation Plugin
- Clearly defined APIs for model integration (HuggingFace, OpenAI, Kaggle, etc.).
- Seamless adoption of Open WebUI’s RBAC and permission systems.
Configuration & Resource Management
- Extend existing infrastructure to handle GPU/CPU preference settings and concurrency controls.
- Use a lightweight scheduler or queue for task distribution.
UI Integration
- Add dedicated panels/tabs within the WebUI for Prompt-to-Image, Image-to-Image, In-Painting, and Out-Painting.
Error Handling & Logging
- Implement descriptive logs and robust error messages for troubleshooting.
- Ensure resource-limit scenarios are communicated effectively.
Testing & Validation
- Include unit tests, integration tests, and performance validations to guarantee reliability.
- Encourage community-driven testing for niche setups.

Feature Checklist

Prompt-to-Image Support
Image-to-Image Transformations
In-Painting Functionality
Out-Painting Functionality
HuggingFace Model Integration
OpenAI DALL·E 2/3 Integration
Kaggle Model Integration
Configurable GPU/CPU Resource Allocation
Example Scripts and Best Practices

Thank you for reviewing this request! I believe Offline-Ready Image Generation in Open WebUI will significantly enhance the platform, streamline AI workflows, and foster greater collaboration. I look forward to your feedback and any guidance on moving this proposal forward.

Originally created by @arudaev on GitHub (Jan 16, 2025). Original GitHub issue: https://github.com/open-webui/open-webui/issues/8607 ## **Is your feature request related to a problem?** There used to be a tool called OpenAI Labs, which offered cool features like Variations and Out-Painting. However, it is being sunsetted, forcing me and many other AI enthusiasts to search for alternative tools. The available paid tools are often subpar, striving for overly realistic results or adopting a chat interface similar to DALL·E 3 in ChatGPT. Additionally, HuggingFace models are frequently unstable, appearing and disappearing on a weekly basis. Currently, there is no environment that seamlessly integrates **Prompt-to-Image**, **Image-to-Image**, **In-Painting**, and **Out-Painting** operations. This gap necessitates the use of separate tools or services, complicating workflows and increasing setup overhead. --- ## **Describe the solution you'd like** I propose adding an **offline-capable image generation module** to Open WebUI. This feature will seamlessly blend text-based and visual AI, enabling users to perform a comprehensive range of image operations within a single interface. Key aspects include: 1. **Unified Visual + Text AI** - Consolidate text and image operations under one platform, preserving Open WebUI’s simplicity. 2. **Comprehensive Image Generation Features** - **Prompt-to-Image**: Create images directly from text prompts. - **Image-to-Image**: Transform existing images based on new instructions. - **In-Painting**: Fill missing or modified areas in an image. - **Out-Painting**: Extend image boundaries seamlessly. 3. **Extended Model Support** - **HuggingFace**: Integrate open-source models for offline usage. - **OpenAI DALL·E 2/3**: Provide connectivity to external APIs for users who prefer or require online capabilities. - **Kaggle**: Offer a path to pull or use models from Kaggle notebooks/datasets. 4. **Resource Allocation & Management** - Allow configurable GPU/CPU usage, ensuring performance at both large and small scales. - Support multiple concurrent requests without degrading the user experience. 5. **Modular Architecture** - Maintain a **separate, well-structured image-generation module** to minimize impact on the core codebase. - Facilitate future expansions (e.g., multi-GPU setups, advanced style transfers) by isolating image-processing logic. 6. **Documentation & Examples** - Provide clear instructions for setting up and using the image-generation features. - Offer sample scripts and best practices to help users get started. --- ## **Describe alternatives you’ve considered** - **External Tools**: - [OpenOutpaint](https://github.com/zero01101/openOutpaint) - [The better version: Stable Diffusion Infinity](https://github.com/lkwq007/stablediffusion-infinity) - [DALL·E 2 in Labs (that is being _sunset_ this year)](https://labs.openai.com/) - [The best Image Enhancer](https://huggingface.co/spaces/finegrain/finegrain-image-enhancer) - And many other tools.... --- ## **Additional Context** ### **Why This Matters** - **Streamlined Workflows**: Combining text and image capabilities in one interface saves time and fosters creativity. - **Community Growth**: Robust, multi-modal functionality can attract diverse users and contributors. - **Future-Proofing**: Aligns with Open WebUI’s commitment to extensibility and offline readiness. ### **Proposed Implementation Approach** 1. **New `image-generation` Plugin** - Clearly defined APIs for model integration (HuggingFace, OpenAI, Kaggle, etc.). - Seamless adoption of Open WebUI’s RBAC and permission systems. 2. **Configuration & Resource Management** - Extend existing infrastructure to handle GPU/CPU preference settings and concurrency controls. - Use a lightweight scheduler or queue for task distribution. 3. **UI Integration** - Add dedicated panels/tabs within the WebUI for Prompt-to-Image, Image-to-Image, In-Painting, and Out-Painting. 4. **Error Handling & Logging** - Implement descriptive logs and robust error messages for troubleshooting. - Ensure resource-limit scenarios are communicated effectively. 5. **Testing & Validation** - Include unit tests, integration tests, and performance validations to guarantee reliability. - Encourage community-driven testing for niche setups. ### **Feature Checklist** - [ ] **Prompt-to-Image** Support - [ ] **Image-to-Image** Transformations - [ ] **In-Painting** Functionality - [ ] **Out-Painting** Functionality - [ ] **HuggingFace Model Integration** - [ ] **OpenAI DALL·E 2/3 Integration** - [ ] **Kaggle Model Integration** - [ ] **Configurable GPU/CPU Resource Allocation** - [ ] **Example Scripts and Best Practices** --- Thank you for reviewing this request! I believe **Offline-Ready Image Generation** in Open WebUI will significantly enhance the platform, streamline AI workflows, and foster greater collaboration. I look forward to your feedback and any guidance on moving this proposal forward.

GiteaMirror closed this issue

2026-05-15 10:28:15 -05:00

Sign in to join this conversation.

Branches Tags

1 Participants

Notifications

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: github-starred/open-webui#85931