docs: add image generation documentation and image-to-image editing tutorial

[nabinchha]

📋 Summary

Add comprehensive documentation for the image generation feature and a new tutorial demonstrating image-to-image editing with seed datasets.

🔄 Changes

✨ Added

• Tutorial 6: Image-to-Image Editing — New tutorial that loads animal portraits from HuggingFace (AFHQ), streams them as a seed dataset, and uses ImageColumnConfig with multi_modal_context to generate accessorized versions (sunglasses, top hats, bow ties, etc.)
• Image Columns section in docs/concepts/columns.md — Documents the 🖼️ Image Columns type (diffusion vs autoregressive, Jinja2 prompts, preview vs create output modes, multi_modal_context)
• Image Inference Parameters section in docs/concepts/models/inference-parameters.md — Documents ImageInferenceParams fields, examples for diffusion and DALL·E-style models, API auto-selection

🔧 Changed

• mkdocs.yml — Added Tutorial 5 (Generating Images) and Tutorial 6 (Image-to-Image Editing) to the site navigation
• docs/code_reference/models.md — Updated intro to mention ImageInferenceParams, added link to Tutorial 6
• docs/notebook_source/5-generating-images.py — Added link to Tutorial 6 in next steps section
• docs/concepts/columns.md — Updated column count from "ten" to "eleven"
• docs/concepts/models/inference-parameters.md — Updated intro to mention all three inference parameter types

#317 should merge first

[greptile-apps]

Greptile Overview

Greptile Summary

This PR adds comprehensive documentation for the image generation feature, including a new Tutorial 6 that demonstrates image-to-image editing using seed datasets and multi_modal_context. The changes include:

• New Tutorial 6: Complete walkthrough of loading images from HuggingFace AFHQ dataset, passing them as context to autoregressive models via multi_modal_context, and generating edited versions with accessories and style variations
• Documentation enhancements: Added Image Columns section to columns.md and Image Inference Parameters section to inference-parameters.md, explaining diffusion vs autoregressive models, output modes, and configuration
• Tutorial improvements: Fixed image display code in Tutorial 5 and added cross-references between tutorials
• Navigation updates: Added new tutorials to mkdocs.yml site structure

The documentation is clear, well-structured, and consistent with existing tutorials. All Colab notebooks were properly regenerated from source files. The previously reported model classification issue has been addressed in a prior commit.

Confidence Score: 5/5

• This documentation-only PR is safe to merge with no risk to production code
• All changes are documentation updates (tutorials, concept docs, and generated notebooks). No production code is modified. The content is accurate, well-written, and follows existing patterns. Tutorial code examples are functional and educational in nature.
• No files require special attention

Important Files Changed

Filename	Overview
docs/concepts/columns.md	Added Image Columns section documenting diffusion/autoregressive models, Jinja2 prompts, output modes, and multi_modal_context; updated count from ten to eleven column types
docs/concepts/models/inference-parameters.md	Added comprehensive Image Inference Parameters section with fields table, examples for autoregressive and diffusion models; updated intro to mention all three parameter types
docs/notebook_source/6-editing-images-with-image-context.py	New tutorial demonstrating image-to-image editing with AFHQ dataset, multi_modal_context for autoregressive models, and before/after comparisons
docs/notebook_source/5-generating-images.py	Fixed image display code using IPImage filename parameter, added link to Tutorial 6, minor typo fix
mkdocs.yml	Added Tutorial 5 and Tutorial 6 to navigation structure
docs/colab_notebooks/6-editing-images-with-image-context.ipynb	Generated Colab notebook from Tutorial 6 source with installation and API key setup cells

[greptile-apps]

_{13 files reviewed, 2 comments}

_{Edit Code Review Agent Settings | Greptile}

[greptile-apps]

Additional Comments (1)

docs/colab_notebooks/6-editing-images-with-image-context.ipynb
Missing API-key setup

This tutorial states “Set OPENROUTER_API_KEY in your environment”, but the notebook’s setup cell only populates NVIDIA_API_KEY. As written, a fresh Colab run will fail/authenticate incorrectly for the chosen provider unless the user manually sets OPENROUTER_API_KEY outside the notebook.

Prompt To Fix With AI

This is a comment left during a code review.
Path: docs/colab_notebooks/6-editing-images-with-image-context.ipynb
Line: 1464:1472

Comment:
**Missing API-key setup**

This tutorial states “Set `OPENROUTER_API_KEY` in your environment”, but the notebook’s setup cell only populates `NVIDIA_API_KEY`. As written, a fresh Colab run will fail/authenticate incorrectly for the chosen provider unless the user manually sets `OPENROUTER_API_KEY` outside the notebook.

How can I resolve this? If you propose a fix, please make it concise.

[greptile-apps]

_{17 files reviewed, 1 comment}

_{Edit Code Review Agent Settings | Greptile}

[greptile-apps]

Additional Comments (1)

packages/data-designer-config/src/data_designer/config/utils/image_helpers.py
Broken .jpeg detection

SUPPORTED_IMAGE_EXTENSIONS lowercases the enum values (fmt.value.lower()), but ImageFormat.JPEG.value is already "jpg" (see detect_image_format() mapping) while is_image_path()/is_image_url() tests expect .jpeg to be supported. With the current list, .jpeg files/URLs will no longer be recognized.

Fix by including both .jpg and .jpeg (or avoid lowercasing/altering enum values and keep get_supported_image_extensions() behavior). Also update the list type to a tuple or set if immutability/perf matters.

Also appears in: is_image_url() usage of SUPPORTED_IMAGE_EXTENSIONS in the same file.

Prompt To Fix With AI

This is a comment left during a code review.
Path: packages/data-designer-config/src/data_designer/config/utils/image_helpers.py
Line: 20:22

Comment:
**Broken `.jpeg` detection**

`SUPPORTED_IMAGE_EXTENSIONS` lowercases the enum values (`fmt.value.lower()`), but `ImageFormat.JPEG.value` is already `"jpg"` (see `detect_image_format()` mapping) while `is_image_path()`/`is_image_url()` tests expect `.jpeg` to be supported. With the current list, `.jpeg` files/URLs will no longer be recognized.

Fix by including both `.jpg` and `.jpeg` (or avoid lowercasing/altering enum values and keep `get_supported_image_extensions()` behavior). Also update the list type to a tuple or set if immutability/perf matters.

Also appears in: `is_image_url()` usage of `SUPPORTED_IMAGE_EXTENSIONS` in the same file.

How can I resolve this? If you propose a fix, please make it concise.