how does ETC2 fares when compressing normal maps and bump/height maps?

Since we are talking about using an algorithm created for visual fidelity, applied to vector and numerical data ⃰ I'm quite sure that the answer is "horribly" and "quite bad" respectively, but since I didn't found much about the topic online, I would like to hear about your experience on the matter.

Thank you!

⃰ compressing a matrix using jpeg is the best comparison I might think of at the moment

  • \$\begingroup\$ The easiest way to find out is to try it on some data that's representative of your use case, and compare the output against your alternatives. It's rare to be able to give a completely objective answer "compression A is definitely better for this than compression B, no matter what your project is" — this is often going to come down to a judgement call for you as you balance size on disc/in RAM, performance, and fidelity according to the needs of your project. \$\endgroup\$ – DMGregory Nov 18 '15 at 13:29
  • \$\begingroup\$ @DMGregory thank you for the answer, and I understand your point (I'll probably end up trying myself if I don't get a good answer, as you suggested), though, to my defence my question is quite specific. Normal maps and height maps are well defined kinds of data, associated with specific algorithms, where the question "does compression A works?" makes sense, and can be answered within reasonable constraints (I'm looking specifically for the possibility of visual artefacts) without reinventing the wheel. \$\endgroup\$ – R1ck77 Nov 18 '15 at 13:36

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.