Hi guys
i’m gradually becoming more intersted in photogrammetry
i’m going through the epic workflow and i didn’t actually get a part of the GDG2015:creating the open world kite explanation
“we worked entirely in linear color space,we want to make sure we avoided any kind of SRGB tonemapping”…“the file format is tiff 16 that gives a kind of black image,that’s a linear image,not very artist friendly and hard to paint so we use a custom version of tiff 16 ,that epic wrote, that applies a baked srgb look up that can be removed later with no loss of information”
so he’s talking about gamma correction??..albedo??ok for tiff but why 16 bits?removed later do to what then ?
The goal is to work in linear, and at a high bit depth, for as much of the pipeline as possible, simply to avoid artifacts as you manipulate the data you’ve captured. You want to leave as much data in the image for as long as possible throughout the pipeline
Source photos were RAW, then converted to 16 bit tiff which was readable by the photomodeling software. A texture and mesh were built from the photomodeling tool, which gives an unwrapped 16 bit per channel tiff for maximum value range. Then an SRGB curve is applied to get the asset into a visible range where it can be compared to the calibrated source photos (grey ball, etc), and then delit using Epic’s custom delighting process. The final result of all that is exported as an 8 bit per channel basecolor (albedo) texture into Unreal, and used.
I was not involved in the initial capturing pipeline so small aspects of my summary here might be off by little bits, but the broad idea is to maintain bit depth throughout the pipeline for greatest accuracy.
In the presentation, it was mentioned that some sort of kit might be released to help with the pipeline. Is there any ETA on that or an official announcement? I’d love be able to de-light my scans!