I realize that the R5 is fairly new (and in short supply), but it seems when I use exiftool with the Cr3 files it produces, it doesn't seem to have any of the white balance information, black levels, white levels, etc, available yet. I'm assuming there's a plan to add this at some point? Having that information available via exiftool is really handy for my raw processing software so it can auto detect that metadata and I don't have to manually enter it when processing the Cr3 files.
If it's a matter having sample files, I have an R5. I'll happily generate whatever sample files anybody needs, just tell me what you need.
Thanks
Adrian
Hi Adrian,
Right. Thanks for the reminder. The R5/R6 contain a new ColorData record that needs to be decoded. I'll see about doing this for the next release (ExifTool 12.09).
- Phil
Edit: ExifTool 12.09 is now available
You da man...
Let me know if you need any cr3 files to look at.
Hi Adrian,
Thanks, but I had already found some sample R5/R6 CR3/HIF/JPG files that I used for testing. The ColorData structure was essentially the same as for the M50, but shifted by 14 bytes.
Let me know if you have any problems with the new version.
- Phil
Many thanks. So far it looks good, though NormalWhiteLevel and LinearityUpperMargin being the same value (at ISO 100, I've not looked at other ISOs just yet) is interesting.
In my other (older) EOS bodies, LinearityUpperMargin is usually around 10000. Not saying it's wrong here, just a note for others, if they get magenta in their highlights, it's because the green channel went non-linear and started to roll contrast off before hitting the normal white level, which means LinearityUpperMargin should be adjusted to a lower value than the normal white level. There's almost a full stop difference between the green channel levels and the Red and Blue channel levels at the R5s native white balance of ~5300K (the point where the red and blue multipliers are almost the same if not the same), so there's a distinct chance of that happening and being quite visible if it does. I've not had a chance to test if that actually happens yet, but just looking at the meta output, that caught my eye. If it doesn't happen, then Canon really went to town on that sensor and there's *a lot* more actually usable linear dynamic range available than older sensors. Going all non-linear at 10000 and above pretty much kills off highlight robustness.
All the other usual meta suspects that my processing code looks at looks to be in order otherwise.