Dear Polycount,
I've just posted a new CrazyBump beta test. I would appreciate your feedback!
www.crazybump.com/release/CrazyBump_Beta_Feb4.exe
changes:
-New "extra fine detail" slider added to the normal map controls.
-New "slope influence" slider added to the specmap controls.
-New option to toggle horizontal and vertical wrapping.
-The normal mixer will tile your low-res detail textures, instead of scaling them up.
-Fixed a problem with non-responsive context menus on some PCs.
Replies
Downloading now
I plan to add 8-core support pretty soon, and I seek guinea pigs to help test it!
(I only have two cores over here. And a ball-mouse. And a pet dinosaur.)
One cool thing that i would love to see, i'm not sure how feasible this is however but with hard surface models i get a lot of minor artifacts when i run the detail pass thing, what happens is the gradients in the normals to compensate for lowpoly smoothing ends up in that detail map, and it makes the end result look like i've got smoothing errors where i really dont. So i spend a good deal of time editing these things out. It would be cool if there was some sort of smart filter there to detect that sort of thing, but i'm not sure how you would do it.
You can see what i mean here:
This may not seem like a big deal, but when you've got a real complex mesh, you start to see these sort of artifacts all over the place, and it gets to be a real chore to get clean results.
Here is the TGA used in those shots:
http://dl.dropbox.com/u/499159/mateba_normals.rar
Thanks for the idea!
The simplest solution I can think of would be to give CrazyBump the ability to open object-space normal maps. That way, those artifacts wouldn't be present in the first place.
I'm not imagining having CB create object-space maps from scratch. Rather, I'm imagining having CB load existing object-space maps, so that you can modify them, generate other maps from them, or add details using the mixer.
Would that be a feature that many people would use?
... Now to save up the dosh to get a copy. >.<
Totally awesome idea! I was using OS maps before I was laid off, probably will again. When I do, I'll be wanting a mixer. I remember EQ saying it used to be a bit of a pain for him, with a custom tool. I haven't spent enough time with OS maps to know though.
But the preview doesn't support object-space, so it'll be useless. Likewise the spec / occlusion / displacement tabs will be useless.
But if you only want to mix details onto an existing obj-space map, you can try loading it into the mixer.
@EarthQuake:
Would you be able to give me an object-space version of your mateba_normals, to play with?
Yeah i think that would be very useful, then its just a matter of rendering an extra map, which is very easy.
How about another asset that uses OS?
http://dl.dropbox.com/u/499159/cb_os_test.rar
TGA+OBJ from a different gun.
Here's a test build with 8-thread support:
www.crazybump.com/release/CrazyBump_Beta_Feb7.exe
I hope to learn how it performs on a quad-core i7 processor... to learn whether CrazyBump benefits from hyperthreading.
If you load a large image ( 2048 x 2048 ) on your i7 machine, does CrazyBump run faster when the toggle is set to 8 threads?
Thanks very much for your help!
I get a consistent crash:
1. Load a 2048x2048 normalmap TGA.
2. Go to the Occlusion tab.
3. Click left and right on the Intensity slider, once every 1/2 second. I can click about 10 times before it dumps me to the desktop.
There's a brief message window before CB is gone completely, too fast to read though.
Edit... system specs: WinXP 32 sp3, Intel Core2 Quad @ 2.40GHz, 2 GB RAM, GeForce 8800 GTX.
@EricChadwick: Thanks very much for the crash report! That definitely needs fixing before 1.2 can be officially released
It should also make better use of multiple CPUs. Would you mind testing it, Whargoul? I'm hoping it does better than 45% this time!
http://www.crazybump.com/release/CrazyBump_Beta_Feb15.exe
This build automatically uses whatever number of cores it detects. Alas, it seems like CB doesn't really benefit from more than 4 cores at present. Further investigation is required!