did an experiment converting the tif normal map to png (in 4.6)
Mistara
Posts: 38,675
cranked the strength up to 400% percents to compare the level of detail.
Michael 6.
Post edited by Mistara on

Comments
I think the main reason to use .tif files for normal maps is to allow greater bit depth, which means more fine detail. As far as I know, definitely .jpg and perhaps .png are limited to 8 bits per RGB colour channel (plus another one for the alpha channel in a .png image). The .tif format is a bit of an oddity, it allows extra features to be bolted on. One of these features is 16-bit colour channels — it allows really smooth bump, displacement and normal capabilities, even in extreme closeups, but it does have the drawback that the .tif files can be quite a bit bigger.
the top one is the tif, bottom the png.
tdlmaker triggered on both.
testing next in ca, doesn't seem to be tdl in ca, unless it's in the background.
:) after i finish squeezing oranges for my martini
It also helps with color banding issues, sometimes if you have a slight curve across a good distance you get that stairsteppy effect. It does help to bake normal maps to 16bpp, and save out to a more sane deliverable format- There's a dither process when converting from 16 to 8 which alleviates the banding while moving down. Photoshop does this fairly well, I can't speak for other apps.
PNG also supports 16-bit images. I use 16-bit PNGs for the normal maps I create in ShaderMap.
Both TIF and PNG are lossless file formats. The same image stored in both formats will decode to the exact same in-memory representation (assuming they were saved correctly so you don't have one as 16-bit and the other as 8-bit). If there are differences in the render, then tdlmake is buggered.