WB Figure Studio build1.1.40：六泳道标准验证例

本例沿用公开中文 build1.1.24 定量验证包中的两张合成练习原图和全部原始裁剪、泳道、选区坐标。原录屏与旧下载文件仍保留录制时结果。

新版算法 native-roi-v2 保留双线性采样的小数。对这两张 8 位原图采用相同选区重新计算时，12 个积分密度和 6 个比值与旧版略有差别。独立诊断对新版采样像素先四舍五入，再执行同一背景计算，12 个积分均精确回到旧值。原图、选区和背景分位数公式没有改变。

RGB 亮度用代数等价式 R + 0.7152×(G−R) + 0.0722×(B−R) 计算，使三个通道相等时精确保留该灰度值。离线脚本使用相同定义，保留浮点精度。

WB-Quantification-build1.1.40-zh.xlsx 和 -en.xlsx 使用本批软件真实 Excel 生成函数分别导出同一次复算。它们不是手工填写的模拟产品表格，也不是英文 build1.1.36 长视频的独立重建。结果页保留完整数值，显示格式可能只显示部分小数。Provenance 页含原图 SHA-256、原始裁剪、位深、算法及计算时间。计算结果已经实际生成，Excel 不会因手动改动单元格而重新读取图片计算。

quantification-build1.1.40-comparison.json 包含原图相对文件名与 SHA-256、旧新版逐道数值、裁剪/选区请求，以及本次真实 Excel 导出请求。用匹配校验值的原图在同一算法版本、相同明暗模式和选区下可复算。该文件中的 sources[].request 即本次原生计算的输入。

离线复核：解压全部文件，在该目录运行 node verify-quantification.mjs（Node.js 20 或更新版本）。脚本仅使用 Node 内置模块，无需安装软件或 npm 依赖，也不会联网。它直接解析本包的两张 8 位 RGB PNG，验证 SHA-256，独立完成双线性取样、选区背景和积分计算，并核对 12 个新版积分、6 个新版比值及四舍五入后恢复的 12 个旧积分。成功输出 passed: true。本脚本限定验证本包的图片格式，不是通用图片定量软件。

练习原图是公开合成素材，不是实验观察结果。页面和视频说明仍须结合实验自身的线性检测范围、曝光、背景和内参验证理解。不要将公开 Mac build1.1.37 视为已经包含 build1.1.40 的行为。


WB Figure Studio build1.1.40: six-lane reference example

This example reuses both synthetic public practice originals and the original crop, lane and ROI coordinates from the Chinese build1.1.24 verification package. The recorded video and original downloads retain their historical results.

The native-roi-v2 algorithm retains fractional values from bilinear sampling. Recalculating these 8-bit images with the same regions changes all 12 integrated densities and six ratios slightly. In an isolated diagnostic, rounding the new sampled pixels before the same background calculation reproduced all 12 old integrated densities exactly. The originals, regions and background-percentile formula are unchanged.

RGB intensity uses the algebraically equivalent expression R + 0.7152*(G-R) + 0.0722*(B-R), which exactly preserves the grayscale value when all three channels are equal. The offline script uses the same definition and retains floating-point precision.

WB-Quantification-build1.1.40-zh.xlsx and -en.xlsx are Chinese and English exports of this same recalculation through the application's real Excel generator. They are not hand-authored substitutes for software output, and they do not independently reconstruct the English build1.1.36 video. Result cells store the full numeric values even when display formatting shows fewer decimals. Provenance records source SHA-256 checksums, original crop geometry, bit depth, algorithm and calculation time. These are saved calculation results; editing cells does not cause Excel to reread the source images and quantify them again.

quantification-build1.1.40-comparison.json contains relative source filenames and SHA-256 checksums, old/new lane results, native crop/ROI requests and the actual Excel export request. Recalculate matching originals using the recorded algorithm, polarity and regions. The native inputs are in sources[].request.

Offline verification: extract all files and run node verify-quantification.mjs in that directory (Node.js 20 or newer). The script uses only built-in Node modules, requires no app or npm dependencies and makes no network requests. It decodes the two supplied 8-bit RGB PNGs, verifies their SHA-256 checksums, independently calculates bilinear samples, ROI backgrounds and integrals, and checks the 12 current integrals, six current ratios and 12 rounded-sampling historical integrals. Success prints passed: true. It is a reference verifier for these image files, not a general-purpose quantification program.

The source images are synthetic public practice material, not experimental observations. They do not establish the linear detection range, exposure, background or loading-control validity of an experiment. Public Mac build1.1.37 does not already include build1.1.40 behavior.

