diff --git a/CHANGELOG.md b/CHANGELOG.md index 4f1720a..ac2ec2a 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -1,5 +1,13 @@ # Changelog +## 0.1.1 - Compatibility fixes + +- Added JPEG EXIF UserComment / Civitai metadata fallback +- Added UltimateSDUpscale / USDU sampler-like generator support +- Added nested SmartMetadataReader output resolution +- Improved FAILED / PARTIAL diagnostics +- Added regression tests for nested reader, Civitai metadata, and USDU workflows + ## 0.1.0 - Initial MVP - Python-only Smart Metadata Reader node diff --git a/README.md b/README.md index 80b927a..0542272 100644 --- a/README.md +++ b/README.md @@ -126,12 +126,18 @@ Phase 1 实机和回归测试已覆盖: - `DeepTranslatorTextNode` 文本来源。 - A1111 / Forge `parameters` fallback。 - PNG / JPEG / WEBP metadata 读取。 +- UltimateSDUpscale / USDU sampler-like metadata parsing。 +- nested `SmartMetadataReader` output resolution。 +- `CLIPTextEncode.text -> ShowText -> SmartMetadataReader positive/negative output`。 +- JPEG EXIF `UserComment` / Civitai metadata fallback。 ## 当前限制 - Phase 1 是 Python-only。节点内部三个大文本框和更像 SD Prompt Reader 的图片预览 UI 留到 Phase 2 JS 前端增强。 - 部分未知 custom conditioning 节点会返回 `PARTIAL` / unresolved,而不是猜测 prompt。 - 某些 custom node 的运行结果如果没有写入 metadata 缓存,可能无法还原。 +- 如果放大图引用了原图,例如 `003.png`,但原图不存在于可解析路径中,会返回 `PARTIAL` / unresolved;插件无法凭空恢复运行时输出。 +- 插件不会全局扫描 workflow 中所有 `ShowText` / `widgets_values` / `text` 字段,只解析最终 selected sampler/generator 的 `positive` / `negative` 链路。 - 被压缩或重新保存后丢失 metadata 的图片无法读取原始生成参数。 ## 后续计划 diff --git a/README_EN.md b/README_EN.md index 29a6d6f..10e2135 100644 --- a/README_EN.md +++ b/README_EN.md @@ -126,12 +126,18 @@ Phase 1 real-world validation and regression tests cover: - `DeepTranslatorTextNode` text source. - A1111 / Forge `parameters` fallback. - PNG / JPEG / WEBP metadata reading. +- UltimateSDUpscale / USDU sampler-like metadata parsing. +- Nested `SmartMetadataReader` output resolution. +- `CLIPTextEncode.text -> ShowText -> SmartMetadataReader positive/negative output`. +- JPEG EXIF `UserComment` / Civitai metadata fallback. ## Current Limitations - Phase 1 is Python-only. In-node large text boxes and a richer SD Prompt Reader style preview UI are planned for the Phase 2 JavaScript enhancement. - Some unknown custom conditioning nodes may return `PARTIAL` / unresolved instead of guessing. - If a custom node did not save its generated result into metadata cache, the parser may not be able to recover it. +- If an upscaled image references its source image, such as `003.png`, and that source image is not present in a resolvable path, the parser returns `PARTIAL` / unresolved; it cannot recover runtime outputs from nowhere. +- The plugin does not globally scan all `ShowText` / `widgets_values` / `text` fields in the workflow. It only resolves the final selected sampler/generator `positive` / `negative` chain. - Images with stripped metadata cannot provide original generation settings. ## Roadmap