Combine -> Images Different EV's -> hdr format. >1.0f
Posted: 2012-01-02T05:05:18-07:00
hi
I'm piping raw 48bit RGB video frames from avisynth to linear EXR via IM successfully thanks to help on this forum.
But I now have a new twist to the process which is videos shot with interleaved exposures of set EV within the video frames ie: dark frame -> light frame -> dark frame etc. Dark to Light is user set before capture 1EV to 5EV. The camera basically jumps ISO between frames, capturing a progressive stream of alternating exposures.
The final output I'd like to produce is a hdr image format, floating point, unscaled which encompasses the extent of the exposure range of EV's without normalising 0.0 - 1.0 ie: perhaps take dark image as 0.0 - 1.0 and bright image data 1.0 - 2.0. Now I think that's what I need to do, more than happy to be told this is wrong approach, wrong theory.
The hdr images would then be manually tonemapped in an external application. Image formats supported would be amongst exr, tiff, ppm in order of preference.
As a starting point for discussion I have this on the CLI:
I'm piping raw 48bit RGB video frames from avisynth to linear EXR via IM successfully thanks to help on this forum.
But I now have a new twist to the process which is videos shot with interleaved exposures of set EV within the video frames ie: dark frame -> light frame -> dark frame etc. Dark to Light is user set before capture 1EV to 5EV. The camera basically jumps ISO between frames, capturing a progressive stream of alternating exposures.The final output I'd like to produce is a hdr image format, floating point, unscaled which encompasses the extent of the exposure range of EV's without normalising 0.0 - 1.0 ie: perhaps take dark image as 0.0 - 1.0 and bright image data 1.0 - 2.0. Now I think that's what I need to do, more than happy to be told this is wrong approach, wrong theory.
The hdr images would then be manually tonemapped in an external application. Image formats supported would be amongst exr, tiff, ppm in order of preference.
As a starting point for discussion I have this on the CLI:
Last query, I've managed to add color profile info to the output as it goes through the chain, but is it possible to add a simple exif tag for the EV value in similarly for use with enfuse, that reads the exif EV tag for a alternative approach of two separate images for enfusing rather than one exr holding both exposures.wine ./avs2yuv.exe -raw "avisynthscript.avs" -o - | convert -limit memory 500MiB -limit map 1GiB -size 1920x1088 -depth 16 -define quantum:format=floating-point -define quantum:scale=65536.0 -profile Linear_sRGB_D65.icc -set colorspace RGB -colorspace RGB rgb:- "IM_%06d.exr"