In the example at this Google Drive link three images get merged into a result that contains almost no data from the least exposed image. But some pixels in the scene do actually need these data due to being overexposed in two other images. See the HDRMerge screenshot, with the spots circled with red:
And here's a screenshot from RawTherapee that reads the DNG saved from HDRMerge. Notice the black overexposure highlights.
Here's an attempt to reveal the actual data by lowering the Exposure Compensation value. The colors are too yellow, i.e. not orange enough — that indicates that red channel is indeed overexposed in the raw data.
For comparison, here's what RawTherapee shows for actual lowest-exposure .CR2 file.
Steps to reproduce the problem:
Run HDRMerge
Uncheck all the checkboxes (alignment, cropping and custom white level)
Press Add, choose the three photos from the link above
Look at the image, don't try to alter it, notice that only images 1 and 2 are active.
Try saving the DNG and loading it in RawTherapee (and switching to Neutral profile) to see the overexposure.
In the example at this Google Drive link three images get merged into a result that contains almost no data from the least exposed image. But some pixels in the scene do actually need these data due to being overexposed in two other images. See the HDRMerge screenshot, with the spots circled with red:
And here's a screenshot from RawTherapee that reads the DNG saved from HDRMerge. Notice the black overexposure highlights.
Here's an attempt to reveal the actual data by lowering the Exposure Compensation value. The colors are too yellow, i.e. not orange enough — that indicates that red channel is indeed overexposed in the raw data.
For comparison, here's what RawTherapee shows for actual lowest-exposure
.CR2file.Steps to reproduce the problem: