feat(ai): image generation builder with local-file & multi-image support #125

Closed

despiegk wants to merge 0 commits from development_ai_image_generation into development

despiegk commented

2026-04-16 03:09:11 +00:00

Owner

Summary

Add ImageGenerationRequest builder for text-to-image AND image-conditioned generation (URLs, raw bytes, local paths).
New image_io module: MIME detection, format conversion (PNG/JPEG/WebP/GIF), byte-budget resizing.
Multi-image response parsing (Vec<GeneratedImage>).
Feature-gated behind default-on image-io; legacy generate_image* API unchanged.

Closes #124

Changes

crates/ai/Cargo.toml — optional image + base64 deps behind image-io feature.
crates/ai/src/error.rs — AiError::ImageIo variant.
crates/ai/src/image_generation/image_io.rs (new) — loader, format conversion, byte-budget shrink.
crates/ai/src/image_generation/request.rs (new) — ImageGenerationRequest builder, ImageInput, GeneratedImage, ImageGenerationResult, multi-image parser.
crates/ai/src/client/mod.rs — image_request() entry point; pub(crate) accessors for the builder.
crates/ai/src/lib.rs — re-exports.
crates/ai/examples/image_generation_builder.rs (new) — runnable example.
crates/ai/examples/README.md — entry for the new example.

Test Results

cargo check -p herolib_ai: OK
cargo build -p herolib_ai --examples: OK
cargo test -p herolib_ai: 61 unit + 4 integration + 7 doc-tests, 0 failures.

## Summary - Add `ImageGenerationRequest` builder for text-to-image AND image-conditioned generation (URLs, raw bytes, local paths). - New `image_io` module: MIME detection, format conversion (PNG/JPEG/WebP/GIF), byte-budget resizing. - Multi-image response parsing (`Vec<GeneratedImage>`). - Feature-gated behind default-on `image-io`; legacy `generate_image*` API unchanged. ## Related Issue Closes https://forge.ourworld.tf/lhumina_code/hero_lib/issues/124 ## Changes - `crates/ai/Cargo.toml` — optional `image` + `base64` deps behind `image-io` feature. - `crates/ai/src/error.rs` — `AiError::ImageIo` variant. - `crates/ai/src/image_generation/image_io.rs` (new) — loader, format conversion, byte-budget shrink. - `crates/ai/src/image_generation/request.rs` (new) — `ImageGenerationRequest` builder, `ImageInput`, `GeneratedImage`, `ImageGenerationResult`, multi-image parser. - `crates/ai/src/client/mod.rs` — `image_request()` entry point; `pub(crate)` accessors for the builder. - `crates/ai/src/lib.rs` — re-exports. - `crates/ai/examples/image_generation_builder.rs` (new) — runnable example. - `crates/ai/examples/README.md` — entry for the new example. ## Test Results - `cargo check -p herolib_ai`: OK - `cargo build -p herolib_ai --examples`: OK - `cargo test -p herolib_ai`: 61 unit + 4 integration + 7 doc-tests, 0 failures.

despiegk added 1 commit

2026-04-16 03:09:11 +00:00

feat(ai): image generation builder with local-file & multi-image support

Lint / lint-linux (pull_request) Failing after 4s

Details

Tests with Services / test-linux (pull_request) Failing after 39s

Details

19ffd708fa

Add ImageGenerationRequest builder on top of the existing Gemini 3.1
Flash Image Preview integration. Supports text-to-image and
image-conditioned generation from URLs, raw bytes, or local files. The
image-io module loads images from disk, detects MIME, converts between
PNG/JPEG/WebP/GIF, and resizes to fit a max-bytes budget. Response
parsing now returns every image (not just the first). Feature-gated
(image-io) so consumers can opt out.

#124