Results 1 to 9 of 9

Thread: Witness my shame: Nodes and UVs

  1. #1
    Super Member Nangleator's Avatar
    Join Date
    Sep 2007
    Location
    Massachusetts
    Posts
    2,122

    Witness my shame: Nodes and UVs

    I don't know how to do the simplest of node tasks, apparently.

    I have an object. It's a flat piece of metal that I wish to have a radial brushed aluminum look. I set up a UV texture using cylindrical projection with the axis running straight through the face of the object. Correct?

    Now, I'm trying to get a procedural bump to follow that UV. I can list things that don't work, but I know this is easy, once you've done it, right?

    I know about the anisotropic reflections and speculars. They let me choose my UV map, but the render output has nothing at all to do with reality.

  2. #2
    Army of 1 CC Rider's Avatar
    Join Date
    Jun 2006
    Location
    Charlotte, NC
    Posts
    586
    Can you post a screen shot?

  3. #3
    Quote Originally Posted by Nangleator View Post
    I don't know how to do the simplest of node tasks, apparently.

    I have an object. It's a flat piece of metal that I wish to have a radial brushed aluminum look. I set up a UV texture using cylindrical projection with the axis running straight through the face of the object. Correct?

    Now, I'm trying to get a procedural bump to follow that UV. I can list things that don't work, but I know this is easy, once you've done it, right?

    I know about the anisotropic reflections and speculars. They let me choose my UV map, but the render output has nothing at all to do with reality.
    I think that kind of mapping will put your texture on the outer edge of your piece of metal, and merely streak in straight towards the center. I don't know a way of mapping in a circular direction like that, I just use a map that's been radial blurred in photshop. You can start with pshop's noise and blur it radially a large distance, works well for things like disc brakes.

    But using it in the bump channel won't give you the results you're looking for; bump is on a per pixel basis, and getting anisotropy from bump would require dozens of samples per pixel. I've tried it and failed, and I've never seen it done by anyone else. I suppose it would work if you rendered at ten times bigger than your final image - hmm maybe I'll play with that

    btw the way I use the map is in the specular channel, it keeps the metal looking smooth like it should, bump never looks smooth.
    Confirmed -
    No Weapons of Mass Destruction
    or links to Al Queda or 9/11. (Sep. 2003)

  4. #4
    Valiant NewTeKnight Matt's Avatar
    Join Date
    Feb 2003
    Location
    San Antonio, Texas, USA
    Posts
    13,054
    Have to say, I've always struggled with this too.
    UI / UX Designer @ NewTek
    __________________________________________________
    www.pixsim.co.uk : LightWave Video Tutorials & Tools


  5. #5
    Pixel and Poly Pusher JeffrySG's Avatar
    Join Date
    Dec 2006
    Location
    Philadelphia, PA
    Posts
    5,738
    There was another thread about this same thing... I posed a pic of a radial blur that I created... I'll do a quick search for it.

  6. #6
    Pixel and Poly Pusher JeffrySG's Avatar
    Join Date
    Dec 2006
    Location
    Philadelphia, PA
    Posts
    5,738
    Here's the thread with my radial blur pic. You could use it as a planar map on your object.

    http://newtek.com/forums/showpost.ph...9&postcount=11
    http://newtek.com/forums/showthread....ht=radial+blur

    I would think there would be a way to get this same type of blur using a node but I have no idea how to do it.

  7. #7
    Super Member Nangleator's Avatar
    Join Date
    Sep 2007
    Location
    Massachusetts
    Posts
    2,122
    Thanks for the tips! I never knew about the failure of bump mapping for anisotropy. It goes along with what I've experienced, but I always figured I was doing something wrong.

  8. #8
    Registered User
    Join Date
    Jun 2003
    Location
    Chesterfield, Derbyshire, UK
    Posts
    78
    Quote Originally Posted by Nangleator View Post
    Thanks for the tips! I never knew about the failure of bump mapping for anisotropy. It goes along with what I've experienced, but I always figured I was doing something wrong.

    hmm, I'd be tempted to try to cyllidrically map a cone and then morph it to flat shape. Might work although I expect you'd get a seem.

  9. #9
    Quote Originally Posted by zogthedoomed View Post
    hmm, I'd be tempted to try to cyllidrically map a cone and then morph it to flat shape. Might work although I expect you'd get a seem.
    Holy crap! That's it! A cone!

    I was totally wrong! Anisotropy with bump mapping actually does work. You just need a huge bump texture with enough tiny lines to mimic a real brushed surface, and tons of aa to fight the ever-breeding moire patterns.

    The first thing I tried was a 10k texture made with the good ol' photoshop noise with a radial blur applied trick ( radial blur took about an hour on this size map ). Too noisy. Anywhere there's noise in the bump map you get a bump, and therefore a spec hit, so these would show up all over and ruin any anisotropy look. See first image.

    I needed grooves that were smooth, tiny, and followed the same path. But how to do it? ( After reading the above post, and slapping my forehead,) I planar mapped a cone on X, with a tiny line texture, and rendered it from above orthagonally (no morph), and voila, tiny rings that form circles out to 10,000 pixels.

    After messing with mapping and size for a while, I found out you don't need to render a huge size final image and scale down later, and 5k is pretty good size for the bump map, and easier to manage than 10k. 2500 pixels was too small for the lines to exist, and just baked a moire pattern into the texture.

    After that it was just a matter of placing point lights for a nice spec hit, which involved waiting several minutes for each test render.

    This was a blast! but totally impractical, my final result here is 500x312 and took 5 1/2 minutes to render, with no raytracing at all, no motion blur, no special shaders, just 770 polygons, diffuse, specularity, one map and 5 lights with no shadows. Perspective camera with adaptive set to .001, and there's still noise. The lines that you can see are a moire pattern, not the actual texture. Radiosity renders faster than this!

    The last image is a test with a reflection map HDR ( no raytracing, 8 1/2 minutes ), which doesn't give the right shapes in the aniso for some reason. A geometry environment rather than a spherical map might help.
    Attached Thumbnails Attached Thumbnails Click image for larger version. 

Name:	bump_aniso_01.jpg 
Views:	83 
Size:	123.0 KB 
ID:	60479   Click image for larger version. 

Name:	aniso_final.png 
Views:	112 
Size:	234.0 KB 
ID:	60486   Click image for larger version. 

Name:	aniso_bump_refl.png 
Views:	92 
Size:	840.9 KB 
ID:	60492  
    Confirmed -
    No Weapons of Mass Destruction
    or links to Al Queda or 9/11. (Sep. 2003)

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •