CHAPTER 05 / 11

读懂选鸟结果 Reading the Results

0–3 星如何判定、Pick 精选是什么、连拍分组逻辑，以及输出目录结构和 EXIF 写入字段完全解析。 How star ratings are assigned, what Pick means, how burst groups work, and a complete breakdown of the output folder structure and EXIF fields written.

5.1

0–3 星的含义 Star Ratings

SuperPicky 慧眼选鸟 用 0–3 星加「无鸟」标记描述每张照片的质量。 4.5.0 起星级由同一批照片互相比较得出——先用硬门槛筛掉废片， 再把锐度与美学换算成批内百分位加权排序，最后按配额定星（配额怎么调见 3.2）： Photos are rated 0–3 stars plus a "no bird" marker. Since 4.5.0 the star level comes from photos competing within the same batch: hard gates drop the rejects, the rest are ranked by weighted batch percentiles of sharpness and aesthetics, and stars are assigned by quota (see 3.2):

等级Level 图标Icon 含义Meaning 核心判断依据How it's decided
三星优选 3-star ⭐⭐⭐ 精品 Excellent 批内（开识鸟时为本鸟种内）综合排名进入配额前段，默认前 30%； 另需满足锐度绝对兜底、眼睛可见度与连拍组上限 Ranks in the top quota slice of the batch — or of its species with Bird ID on — 30% by default, and clears the sharpness floor, eye-visibility and per-burst caps
二星良好 2-star ⭐⭐ 合格 Good 排名紧随其后的一段，默认再取 30%； 眼睛可见度低、或所在连拍组 3 星名额已满的照片也降到这里 The next slice, 30% by default; also where eye-obscured frames and burst-cap demotions land
一星普通 1-star ⭐ 留档 Average 过了硬门槛但排名落在配额之外的其余照片；或鸟眼 / 鸟喙完全不可见（角度不佳） Cleared the gates but ranked below both quota slices; or the bird's eye and beak are not visible at all (unfavourable angle)
零星放弃 0-star 0⭐ 极低质量 Reject 置信度不足、锐度极低，或严重过 / 欠曝 Confidence too low, extremely low sharpness, or severely over- / under-exposed
无鸟 No bird ❌ 未检测到鸟 No bird detected AI 未在画面中定位到鸟；照片仍保留在 0星_放弃 目录，不会丢失 AI found no bird in the frame; the photo is kept in the 0star_reject folder, not deleted

两个影响评分的修正机制： Two score modifiers you should know about:

  • 飞版加成 开启飞鸟检测（见第 03 章 3.4）后，检测到飞行状态的鸟自动获得锐度 ×1.2、美学 ×1.1 的加成，抵消飞版动态模糊对评分的影响，让飞版和栖鸟在同等真实质量下有公平的星级结果。 When BIF detection is on (see Chapter 03, Section 3.4), photos with a bird in flight receive a sharpness ×1.2 and aesthetics ×1.1 bonus — compensating for motion blur so flight shots compete fairly against perched birds of the same true quality.
  • 眼睛可见度 鸟眼或鸟喙不可见（正背面、被羽毛遮挡等）时，评分权重随可见度下降，避免「构图差」的照片与「清晰正面」的照片平等竞争。所有关键点均不可见时直接给 1 星。 When the bird's eye or beak is not visible (full back view, heavily obscured, etc.), the rating weight drops with visibility — preventing poorly composed shots from competing with sharp frontal ones. If all keypoints are invisible, the photo is capped at 1 star.
无鸟 ≠ 废片 No bird ≠ bad photo

未检测到鸟有可能是鸟在极端角度、距离太远、AI 置信度略低于阈值。这类照片安全保留在 0星_放弃 目录，不会被删除，可以手动翻看。 No bird detected can mean the bird was at an extreme angle, too distant, or just below the confidence threshold. These photos are kept safely in the 0star_reject folder — never deleted — and can be reviewed manually.

5.2

Pick 精选标记 Pick Flag

Pick 精选（🏆）在三星照片中进一步筛选出「值得优先后期」的那批，所有照片处理完毕后统一计算，算法为美学 × 锐度双排名取交集： The Pick flag (🏆) identifies the most post-processing-worthy photos among the 3-star pool. It is calculated after all photos have been processed, using a dual-ranking intersection of aesthetics and sharpness:

步骤 1 Step 1 全部三星照片按美学评分排序 → 取前 N% Sort all 3★ photos by aesthetics → keep top N%
→
步骤 2 Step 2 全部三星照片按锐度评分排序 → 取前 N% Sort all 3★ photos by sharpness → keep top N%
→
🏆 Pick 🏆 Pick 两个集合取交集——同时进入两个榜单前 N% 的照片获得 Pick 标记 Intersection of both lists — photos in the top N% for both metrics receive the Pick flag

N 的默认值是 25%（三星照片里美学和锐度都进前四分之一的才算 Pick）。可调范围 10%–50%，但没有图形界面入口——只能用命令行 --picked-top 或改配置文件，详见第 09 章。 N defaults to 25% — only 3-star photos that rank in the top quarter for both aesthetics and sharpness receive the flag. Adjustable from 10%–50%, but there is no GUI entry — use the --picked-top CLI flag or edit the config file; see Chapter 09.

Pick 标记在哪里体现： Where the Pick flag appears:

  • 日志 完成报告中显示为 └─ 🏆 N (xx% of 3★) Shown in the completion report as └─ 🏆 N (xx% of 3★)
  • EXIF 写入 XMP:Pick=1，Lightroom 中对应白色小旗 🚩（Pick Flag） Written as XMP:Pick=1 — shows as a white flag 🚩 (Pick Flag) in Lightroom
  • 浏览器 结果浏览器中带 🏆 标记，详见第 07 章 Displayed with a 🏆 icon in the Result Browser — see Chapter 07
Pick 是加分项，不是评级 Pick is a bonus, not a rating

没有 Pick 标记的三星照片依然是三星。三星是基础质量关，Pick 是额外的美学+锐度双优筛选，帮你在大量三星中快速找到「最值得打开后期软件精修」的那一批。 A 3-star photo without a Pick flag is still a 3-star photo. The star rating is the quality gate; Pick is an additional double-filter to help you quickly identify the most post-processing-worthy shots from a large 3-star pool.

5.3

连拍分组 Burst Groups

开启连拍检测后（详见第 03 章 3.5），每组连拍的整理逻辑： When burst detection is on (see Chapter 03, Section 3.5), each burst group is organised as follows:

  • 保留 每组中最高星级的那张留在正常目录（若有 Pick 标记则优先保留 Pick 照片） The highest-rated photo in each group stays in the main directory (a Pick-flagged photo takes priority if present)
  • 归组 同组其余照片移入目标目录内的 burst_xxx/ 子目录，不会被删除 The rest of the group moves into a burst_xxx/ subdirectory inside the same target folder — never deleted
  • 跳过 整组连拍全是 0–1 星时，不做合并，各自按正常流程放置 If every photo in the burst is rated 0–1 star, no grouping occurs — each photo is placed normally

burst 子目录以序号命名，格式为 burst_001/、burst_002/……示例目录结构（species-first 布局）： Burst subdirectories are named sequentially — burst_001/, burst_002/, etc. Example structure (species-first layout):

白腹海雕/3星_优选/ ├── DSC01234.ARW ← 本组连拍最高分，保留 / top-rated shot, kept here ├── DSC01234.xmp ← ARW 侧车文件 / XMP sidecar └── burst_001/ ├── DSC01235.ARW ← 同组其余成员 / rest of the burst group └── DSC01236.ARW
5.4

目录结构 Folder Layout

v4.3.0 的默认布局是 species-first（先鸟种后星级）： v4.3.0 uses species-first as the default layout — species folder on the outside, star rating on the inside:

照片原目录/ ├── 白腹海雕/ │ ├── 3星_优选/ │ └── 2星_良好/ ├── 大白鹭/ │ ├── 3星_优选/ │ └── 2星_良好/ ├── 其他鸟类/ │ ├── 3星_优选/ ← 未识别鸟种的 3 星 / unidentified 3★ │ ├── 2星_良好/ ← 未识别鸟种的 2 星 / unidentified 2★ │ ├── 1星_普通/ │ └── 0星_放弃/ ← 0 星 + 未检测到鸟 / 0★ + no-bird └── .superpicky/ ← 隐藏目录，报告与缓存，不含照片 / hidden, reports & cache

两条固定规则： Two fixed rules:

  • 规则 1 1 星及以下始终进「其他鸟类」——即使识别出了鸟种也如此。低质量照片不值得按鸟种分散管理，统一归入此处方便整体处理。 1-star and below always go into "其他鸟类 / Other Birds" — even if a species was identified. Low-quality photos aren't worth splitting by species; keeping them together makes bulk handling easier.
  • 规则 2 0 星和无鸟（❌）共享同一目录：均进 0星_放弃/，需要时可手动翻看，照片不会被删除。 0-star and no-bird (❌) share the same folder — both go into 0星_放弃/. Photos are never deleted and can be reviewed manually.
为什么 v4.3 改为 species-first 默认？ Why is species-first the default?

按鸟种归类后，同一只鸟的所有照片集中在一个目录里，导入后期时一次就能取齐；星级仍以子目录区分，不影响挑片顺序。 With species-first, every shot of the same bird sits in one folder, so a post-processing import picks them all up at once; star levels still separate them as subfolders.

如果偏好旧版 rating-first 布局（3星_优选/白腹海雕/），可在设置 → 输出中切换，详见第 09 章。 If you prefer the old rating-first layout (3星_优选/白腹海雕/), it can be switched under Settings → Output — see Chapter 09.

5.5

EXIF 写入字段 EXIF Fields Written

处理完毕后，SuperPicky 慧眼选鸟 通过 ExifTool 将以下信息写入每张照片的元数据： After processing, SuperPicky 慧眼选鸟 uses ExifTool to write the following metadata fields into each photo:

字段Field 写入内容Content written 说明Notes
Rating 0–3 0–3 Lightroom / 系统星级，与界面星级直接对应 Lightroom / system star rating, directly matching the interface
XMP:Pick 0 或 1 0 or 1 Lightroom Pick Flag（1 = 白旗精选） Lightroom Pick Flag (1 = white flag)
XMP:Title 鸟种名称 Bird species name 开启识鸟后写入（见第 08 章）；支持中文 / 英文 Written when Bird ID is on (see Chapter 08); Chinese or English
XMP-dc:Subject 鸟种关键词 Species keywords 4.5.0 新增：鸟名同时写进关键词，可在 Lightroom 关键词面板里按鸟种筛选。 采用合并写入，你原有的关键词一个都不会被动到 New in 4.5.0: the species name also goes into keywords, so you can filter by species in Lightroom's keyword panel. Written merge-add — your existing keywords are never touched
XMP:Description 详细评分说明 Detailed scoring breakdown 包含置信度、锐度、美学分值和判断理由 Confidence, sharpness, aesthetics scores and decision reason
XMP:Label 颜色标签 Color label 蓝色 = 飞版；绿色 = 头部精焦；红色 = 脱焦/失焦；无标签 = 其他 Blue = BIF; Green = head-focused; Red = out-of-focus; no label = other
XMP:City（城市） (City) 锐度分值（如 450.23） Sharpness score (e.g. 450.23) 借用地理字段存储评分数据；在 Lightroom 元数据面板中显示为数字属正常现象，可用于智能相册筛选 A geographic field repurposed to store scoring data; the numeric value in Lightroom's metadata panel is expected behaviour and can be used in Smart Collection filters
XMP:State（区域） (State) 美学分值（如 5.24） Aesthetics score (e.g. 5.24) 同上；借用地理字段存储美学评分 Same as above; geographic field repurposed for aesthetics score
IntellectualGenre IUCN 濒危等级 IUCN Red List category 如 LC / VU / EN（开启识鸟后写入，见第 06 章） e.g. LC / VU / EN (written when Bird ID is on — see Chapter 06)
Event GBIF 罕见度（0–100） GBIF rarity score (0–100) 全球出现频率评分（开启识鸟后写入，见第 06 章） Global occurrence frequency score (written when Bird ID is on — see Chapter 06)

RAW 文件（ARW / NEF / CR2 等）：SuperPicky 慧眼选鸟 不修改 RAW 原文件，所有元数据写入同名 .xmp 侧车文件。Lightroom / Bridge 会自动读取同目录下的 XMP 侧车，无需手动操作。 RAW files (ARW / NEF / CR2 etc.): SuperPicky 慧眼选鸟 never modifies the RAW file itself — all metadata is written to a .xmp sidecar of the same name. Lightroom and Bridge read XMP sidecars automatically.

JPEG 文件：EXIF 信息直接写入 JPEG 文件内部，无侧车文件。 JPEG files: EXIF data is written directly inside the JPEG — no sidecar needed.

在 Lightroom 中导入后如何利用这些字段 How to use these fields after Lightroom import

导入时选择「读取元数据和嵌入侧车」，Lightroom 自动显示星级和 Pick 旗标；Title 字段在元数据面板中可见；Label 颜色标签可用于筛选飞版（蓝色）或精焦照片（绿色），红色则是脱焦待删的那批；City / State 中的评分数值可用于智能相册条件筛选。完整的 Lightroom 导入流程详见第 07 章。 When importing, choose "Read Metadata and Embedded Sidecar." Lightroom will automatically display star ratings and Pick flags; the Title field is visible in the metadata panel; the Label color can filter BIF shots (blue) or head-focused shots (green), while red marks the out-of-focus ones to cull; and the City / State score values can be used in Smart Collection criteria. Full Lightroom import walkthrough in Chapter 07.