AntonPalmqvist / physically-based-api

A database of physically based values for CG artists.
https://api.physicallybased.info/
Creative Commons Zero v1.0 Universal
400 stars 15 forks source link

material suggestion: ink #52

Open Aeva opened 4 days ago

Aeva commented 4 days ago

The database already has office paper which is great, but I think it would be useful to also have reference points for how printed images should look for a variety of common printing methods (inkjet at minimum). I think measured values for the sRGB primaries (as in the color space of the image not the color space of the printer) as well as a black ink only swatch would be sufficient?

My thinking here is this would be useful for making printed materials in PBR renders look more realistic, but it would also be useful for digital artists who are new to working with PBR to help them conceptualize how PBR behaves relative to a familiar color space.

AntonPalmqvist commented 3 days ago

Thank you for this interesting idea!

It's indeed a good exercise to conceptualize the idea of printed colors vs. what we see in a render engine. Let's say we pick a color of 0,0,0 in our software and print it. It will never reflect 0% light, because it's not physically possible to have a material that absorbs all light that hits it.

This guy did quite a thorough test and measured black ink on white office paper, printed a consumer printer: https://youtu.be/vtUhUeOo5mM?si=GYFZTiR8AbBp71Fi&t=726

It came out at 0.04 in sRGB linear the first time he printed it. The second time he printed, he printed it twice to make it cover the paper better and it came out at 0.03 sRGB linear. That's quite close to charcoal at 0.02, which is one of the darkest naturally occurring materials. It's however possible to go even darker, for example the darkest paint he could find measured only 0.008. That's only 0.8% reflected light!

I'll consider adding the following materials to the database as I think they are the most interesting reference points:

For red, green, and blue measurements, I suspect there will be too much variation depending on the type and quality of the printer (laser, ink, etc.), and the quality of the paper, to give a relevant reference point for those values. Not that it wouldn't be an interesting exercise!