Denizens of Polycount:
CrazyBump Beta 8.5 is ready for testing. This one makes better-looking normal maps.
-New code makes a nicer blend of shape recognition with image detail.
-Mixer combines normal map layers more accurately than before.
-Overall awesomeness improved by 27%.
Download from
http://crazybump.com/beta/upgrade.html
Replies
cheers,
Rom
snowblindstudios.com
CS3 uses Direct3D, while older photoshops don't, so I suspect the performance trouble might have something to do with D3D.
Does photoshop have an option to disable use of D3D? If so, does it seem to help?
Are you seeing these performance issues in phtoshop, or in crazybump, or both?
thanks very much,
-Ryan
The thing is that i think with older versions of PS that wouldn't really affect PS at all, just the 3d apps, oh well likely a photoshop problem. =(
Download and try it out, you won't regret it.
This needs further investigation. CrazyBump should play nice with other 3D software... if it's not doing that, then something needs fixing!
Hi Sherrell,
CrazyBump is a multipurpose bumpmap creation/manipulation thingy which really needs a more informative website.
-It uses image recognition to make normal & displacement maps from photo-source textures.
-It combines normal maps in multiple layers, and blends them properly in 3D. This looks much better than trying to blend them with a photoshop action.
-It removes directional light from photo-source textures, to make them suitable for use as diffuse maps.
-It has a tool to help create specmaps
-It has a tool to make fake (but decent-looking) ambient occlusion maps.
I'm sure it has features that I'm forgetting to mention. If you're running windows, then please download the free beta test, and give it a spin!
thanks,
-Ryan
I gave the new version a spin, the first image I tried creating a normal map for I didn't get very good results but the second I got amazing results. I think it might be because the first image the light source was adjacent to the the object, I dunno. What is the ideal image to use in crazy bump?
Would you be willing to show me your first image... the one which got poor results? If you could show it to me, then I might be able to improve my algorithm to handle it better, or at least offer specific advice about it...
I realize shape recognition is very difficult but it seems a lot of images I try it on sort of have a diagonal gradient across the heightmap. So like instead of parts that are the same height being the same-ish value the parts on one corner are lighter and the ones on the opposite are darker.
Anyways, love the program it's great to see you are always improving on it.
here is a link to that first image if you are still interested. http://img86.imageshack.us/img86/5909/ornaments0098slk7.jpg