[Bf-committers] Normal Map issue.
metalliandy666 at googlemail.com
Thu Nov 6 00:00:16 CET 2014
Thanks for the reply. I want to stress that mirroring normal maps
without error should be perfectly possible and has been since 2.57 iirc.
Here are some screenshots of perfectly mirrored normal maps, which have
been rendered in Cycles and BI. This case has most amount of mirroring
that is possible so as you can see it works perfectly well :)
The last thing we want to do is propagate the idea that people need to
flip channels or do anything crazy like that in order to get it working
correctly, as it should work perfectly out of the box.
If you are finding that you do need to flip channels then it is either
simple user error or a massive bug that should be reported along with a
.blend file so that devs. can reproduce the error. As it stands I cannot
reproduce such an error.
Can you point me to one of the thread where people are talking about
On 05/11/2014 21:26, R VR wrote:
> Hello, so I came across a thread in the net and tried a few suggestions out, and it turns out that when you have a mirrored model which is exhibiting behavior of flipped normals on either side, all you need to do is separate the RGB, invert the Green, combine RGB and the normals sort themselves out properly.
> Seeing as I have seen this talked about quite a bit and it's a tricky fix, hopefully this information can make it's way out to the broader blender community.
> Cheers, and thanks for the help, more normal map support in cycles for the future would be great though.
>> Date: Mon, 3 Nov 2014 14:42:41 +0000
>> From: metalliandy666 at googlemail.com
>> To: bf-committers at blender.org
>> Subject: Re: [Bf-committers] Normal Map issue.
>> Hi Rob,
>> AFAIK, mirrored baking should work perfectly with a mirror modifier. I
>> just did a quick test and everything looked as I would expect, but maybe
>> I am misunderstanding what you mean.
>> Could you post a picture and a .blend so that I can reproduce the error
>> Also, what kind of normal map are you baking (Tangent Space/Object
>> Space) & which render are you using to bake the normal maps with?
>> On 03/11/2014 08:34, R VR wrote:
>>> Hello, I have a suggestion for what I feel would be a minor feature addition with large large upside. It has been raised by others, but has never been implemented for whatever reason.
>>> When working with a mirrored symmetrical object with a uv map, the normals on the mirrored side are flipped in the red color I believe, which has a very undesirable result. One side is as expected and the other side is completely the opposite. When attempting to make photo-realistic skin pores on an animated character, in which as much uv space as possible needs to be allocated to the face, having a mirrored normal only enhances the level of detail that can be achieved, while having to map the entire surface to have the normals right is a real obstacle for that goal.
>>> This issue is imposing a fairly reasonable obstacle in my workflow (as I'm sure is the case with others), I cannot work in progression with any symmetrical model in which normal maps are a large feature, which is often, seeing as how any texture I make while the object is mirrored and being tested will need to be completely and utterly redone down to the last pixel when it comes time to finalize the model. Not only this but I would also be losing detail in having to share space with the other half of the model.
>>> If the mirror modifier had an implementation that automatically switched the normal map in the right direction for the mirrored side, this would take care of the issue, and seeing as how there are uv options already I don't feel like this would be a big step to take.
>>> You guys are awesome!
>>> Cheers, Rob.
>>> Bf-committers mailing list
>>> Bf-committers at blender.org
>> Bf-committers mailing list
>> Bf-committers at blender.org
> Bf-committers mailing list
> Bf-committers at blender.org
More information about the Bf-committers