xNormal instead of Max's render-to-texture?
polycounter lvl 8
Hi,
Ok, I know this must be the most stupid question - but what are the actual benefits of using xNormal instead of Max's render to texture?
I have tried xNormal quite a few times, watched tutorials, calculated ray distances and even tried an external cage, but I just can't get a significantly better map with it, compared to what I can get from render to texture.
So, am I missing something here? Is it really a lot better than Max's built-in baking? What's the secret of it's popularity?
Ok, I know this must be the most stupid question - but what are the actual benefits of using xNormal instead of Max's render to texture?
I have tried xNormal quite a few times, watched tutorials, calculated ray distances and even tried an external cage, but I just can't get a significantly better map with it, compared to what I can get from render to texture.
So, am I missing something here? Is it really a lot better than Max's built-in baking? What's the secret of it's popularity?
Replies
and what you do, for smoothed cubes xnormaly is way better for example
Let's make up an example of a simple game prop. Let's say it can have a 1024 normal map and it's intended use is in Unity3D. Does it make a difference then?
Max Pros:
- All in the same app, no exporting and importing.
- Cage tweaking without a separate mesh.
- An interface I'm used to.
- I have a few different options for baking, Mental Ray, Vray, Scanline.
- I have full control over the lighting and the scene. I can do a skylight, or a custom light dome, or any other custom light setup.
- You can bake multiple meshes at the same time.
- You can use the standard include/exclude and renderable/non-renderable options in RTT along with cast/don't cast shadows. Good for floaters and low/high poly mesh intersection issues.
- Supports network rendering. Even with a network of only your computer, you can render in the background while still working in max. This lets you increase the samples on your AO which takes longer but doesn't tie up your machine. Considering AfterBurner is included in max and its easy to setup I don't know why more people don't use it.
Max Cons:- Slow for higher quality and bigger maps.
- If you're not used to the RTT interface, or the render settings it can be convoluted to deal with.
- The render preview window is almost totally useless and confuses more people then it helps. There isn't a quick way to check the actual maps it outputs without tracking them down and opening them. Most other paths in max have a right click option to open or reveal in explorer, so it just makes autodesk look incompetent and inconsistent.
- If you use a cage it normally needs to be reset and pushed before it is usable. Even resetting xform before applying Projection doesn't seem to fix it some times.
Xnormal Pros:- Its fairly fast from what I've used it for.
- Gives great results.
- Has a lot of options and features, it also has some crazybump like features.
- It's free
- If you know how to bake with it then it doesn't matter what app your studio is using you'll know how to get good results.
- I think it has some batching options?
- If you're a Maya user Transfer maps is a pain in the ass (at least for me) so xNormal is a excellent option. Same goes for Mudbox, Zbrush and Softimage.
xNormal Cons:That was the kind of list I was looking for. I can confirm all the pros of Max's rtt, except for the last one, that I haven't tried. (I find AfterBurner rather heavy and I don't currently use it, then again I enjoy mental ray's dbr very much with viz projects, but that's off topic)
Indeed, I find Max's "all-in-one" interface more comfortable to use. This is of course a matter of taste and what one is used to, but I really like the way you can tweak the exported cage and import it back again. It doesn't get much more convenient than that!
I think I will mainly stick with RTT, but I will continue my test baking with xNormal every now and then.
"No cage tweaking in the app, its a separate mesh you can import, which makes tweaking a bitch, so people normally stick to offset and any problems it causes, or spend extra time building a 3rd cage mesh."
This is not true xNormal does support both tweaking a existing cage or making a completly new one, altough it is realy impractilcal to work with compared to max.
I missed that. Indeed, it has and I have even tried it. But really, it was one of the most difficult ways of tweaking a mesh I have ever seen. Perhaps it will get better in some future version.
Oh... it is pretty unfriendly to use... oh well win some lose some, it might be ok to tweak a tiny area if needed.
Since I create mostly characters, some of the problem areas tend to be the armpits, and inner legs when using the offset. In these cases one of the more awesome features Xnormal has is you can use other geometry as a ray blocker. I simple poly plane in between the legs and underneath the armpits prevents a large offset from grabbing the normals from areas you don't want.
Though I will agree Max's cage is awesome its just so hard to get away from the speed of Xnormal.
http://www.polycount.com/forum/showpost.php?p=1169198&postcount=1984
Max has a way of tweaking edges so the blend better than Xnormal. It doesn't seem to matter the realtime dx render engine.
I think it might be an issue with max not treating the mesh like a game would, which I think xnormal does. If you break and bake in xNormal it will probably come out fine. However if you break and bake in max it will be correct also but when you preview/render it, you'll need a viewport shader like the 3point shader to display it properly.
OR is it the export format fubar'ing the geometry possibly?
I must admit that I just became a fan of xNormal
It is absolutely fantastic for baking maps for ZBrush models - I was previously trying to do this with ZBrush itself or Max, but the trouble of matching UVs, or -alternatively- having to decimate my "artwork" for baking in Max kind of turned me off.
Here xNormal really wins, you get all the details and no UV fuzz.
As an general rule you should try both on some example asset and at that time your understanding of workflow and benefits for you should be more visible...
NOTE: if you have monster GPU you can take advantage of GPU calculate AO in xNormal along with experimental grid rendering (like you have more machines as slaves like me)
As it was stated by Mark in coparision there is limitation in handling batch of objects with xNormal.. I overcome this problem by writing batch export and render script that takes care of this. I'm thinking of releasing it to public when it will be stable and mature enough.