Jump to content
  • Advertisement
  • entries
    422
  • comments
    1540
  • views
    490100

Pick a picture... any picture. No! not that one...

Sign in to follow this  
jollyjeffers

152 views



Top-left image is (obviously) a photo, consider it as a reference image for varnished wooden flooring.

Of the other three, which one is the most correct?

They all come from the same source data (geometry, textures..) and have the same camera/light setup. The difference is which algorithm is applied to each one.

Bonus cookies go to the person who can identify which generic reflectance class each of the bottom two belong to [grin]
Sign in to follow this  


10 Comments


Recommended Comments

While I can't say the exact model, bottom left looks like some kind of anisotropic model, maybe ward? Bottom right looks like standard isotropic phong, but I can't say on a plane model :p

I prefer the bottom left look myself, but it feels too flat. Is it bumped?

Share this comment


Link to comment
Personally I'd have to suggest you shoot for a mix between to the two bottom images.

I mean the bottom left is the "cooler" looking of the three, but in my personal opinion it has a little too much going on. I'm not a fan of the recent games that have the mind set that more normal mapping makes everything better. The bottom left image looks to me like it could be part of one of those games. It still looks good, just not really my style :)

Nice work though.

Share this comment


Link to comment
I would guess... bottom right, since it seems to use the most complex shader.

Is the lighting model Ward?

Share this comment


Link to comment
Bottom right is probably the most accurate, though it appears too bump-mapped for my liking... Tone that back just a little bit, and it looks good.

Share this comment


Link to comment
(haven't looked at comments yet)

I'm going to say Phong for BR, and Ash-Shirley for BL.

Btw, I'd say the right is more realistic, but you need a bit of area lighting (one of the flaws of Phong, sadly)


Late edit: Area lighting and a higher exponent. I hate how small area lights have to be faked with small exponents. Screws everything up.

Share this comment


Link to comment
Thanks for the comments - not bad everyone [grin]

Top-right is just a "dummy" image with ambient lighting only. Not really a serious contender!

Bottom-left is indeed anisotropic Ashikhmin-Shirley (Well done Cypher!) and bottom-right is a regular Blinn-Phong.

Quote:
I prefer the bottom left look myself, but it feels too flat. Is it bumped?
Yup, exactly the same normal map as the bottom-right. I can't remember off the top of my head, but Ashikhmin-Shirley scales down the diffuse component a lot (much of the result comes from the specular term) so the bumps aren't quite so noticeable it seems...

Quote:
Bottom right is probably the most accurate, though it appears too bump-mapped for my liking... Tone that back just a little bit, and it looks good.
Yeah, thats an odd one - the Bottom-left uses exactly the same normal-map, yet generates much flatter bumps..

Quote:
Area lighting and a higher exponent. I hate how small area lights have to be faked with small exponents. Screws everything up.
Interesting - I originally had a much higher exponent for the bottom-right, but it looked much worse. Ended up with a very specific ellipse reflection that obviously didn't follow/match the underlying surface. I dropped the exponent until it looked a bit more convincing [lol]

Share this comment


Link to comment
Quote:
Interesting - I originally had a much higher exponent for the bottom-right, but it looked much worse. Ended up with a very specific ellipse reflection that obviously didn't follow/match the underlying surface. I dropped the exponent until it looked a bit more convincing


Yeah, but keep in mind that I said area lighting as well. The comparison you're doing is between an infintisimally (sp) small light and a very large light. The reason why you had to drop down to a lower exponent was becuase it covers a larger area and hence LOOKS like an area light. However, if you look at the edges of the source image, you'll notice that there's a relatively sharp falloff, which is indicative of a much higher exponent in play.

Share this comment


Link to comment
Good point.. I'll see what I can do about that, but I'm thinking that might just be a limitation [headshake]

Jack

Share this comment


Link to comment
It is, sadly. The best you could do is either do an approximation of a double integration of (R dot V)^n (not fun) or some kind of 2D/Cube texture lookup for the scene (special case! eewwwwww).

One option is to have a few spherical area lights, where the specular contribution is calculated through a texture lookup with the R dot V result as the input (This is actually similar to something I'm going to look into when I'm back in Waterloo)

Share this comment


Link to comment
Interesting ideas... Will have to remember those in the future [smile]

For this project I wanted to implement the exact lighting models rather than create my own (or hybrids) that appeared to work.

Cheers,
Jack

Share this comment


Link to comment

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
  • Advertisement
×

Important Information

By using GameDev.net, you agree to our community Guidelines, Terms of Use, and Privacy Policy.

We are the game development community.

Whether you are an indie, hobbyist, AAA developer, or just trying to learn, GameDev.net is the place for you to learn, share, and connect with the games industry. Learn more About Us or sign up!

Sign me up!