BUG in pfscalibration

[jeremy.r...@gmail.com]

Hi,

I've recently tried to use pfstools to calibrate the response curve of grayscale cameras and bumped into some weird behaviour.
To make my grayscale images into 3-channel images, I simply duplicated the same data to all 3 channels of a png image and ran the following:
pfsin my_file.hdrgen | pfshdrcalibrate -v -s response.m | pfsout test.exr

Problem is, I get a response curve similar to the one below, where red is clearly wrong and green and blue seem to agree on the linearity of my sensor. Now the reason why I'm puzzled by this plot is because R,G and B channels all contain the exact same data ...

Cheers,

Jeremy

[Rafal]

Jeremy,

On Monday, 8 May 2017 16:35:36 UTC+1, jeremywrote:

Problem is, I get a response curve similar to the one below, where red is clearly wrong and green and blue seem to agree on the linearity of my sensor. Now the reason why I'm puzzled by this plot is because R,G and B channels all contain the exact same data ...

If you could share your images and the hdrgen file that produce this behaviour, I can check what could be wrong.

You can try adding "-r linear" to initialize optimizer with a linear response.

Rafal