thank guys 
I was thinking Iād try and see if I can help you out with the failing normal mapping on the knee pads.
Did you bake the normal maps for these in Blender? If so which version?
Also which version of Blender did you use for the rendering?
Would you be able to setup a testscene with just one of the pads and the hi-res such that the scene is setup for baking?
Cheers,
Morten
hi mmikkelsen, i have used the last version āBlender 2.59 (stable)ā http://www.blender.org/download/get-blender/
the problem of the normal map is only in blenderā¦ in Modo 501 in marmoset in thearender i dont have this problem of normal map!
so u made the normal map with blender as well? Can you show me the unwrap for the knee-pad? and also could u show me the normal map?
I never quite understood blender internal nor_map supporting.
Is BI maya or LW compatible?
BTW, MESCH973, in the marmoset render version, thereās still a problem on these knee-pads.
ok friends the uv, the normal and modo rendering, in modo i dont have the same problem, is good
Attachments
It looks like the mesh was baked with incorrect vertex normals down the centre of the mirrored parts, or it was baked correctly and you forgot to weld the centre verts.
This isnt Blenders baking tools being broken, its a problem with the baking mesh or the final LP mesh 
The issue with the seam is a very dependent on how exactly you baked the map.
Did you use the mirror modifier with merge checked?
Did you mirror the geometry manually and remove doubles?
Did you mirror the geometry manually and not remove doubles?
Did you bake with half the mesh and then mirror?
Did you bake before of after you mirrored the mesh?
All these factors can cause issues when baking mirrored geometry and if you donāt know how to fix them, it can be very frustrating 
When you bake with only one half of the low poly mesh, or with the mirror modifier without merge on, the vertex normals that run down the centre are no longer going to be correct when you connect the two part together again and remove the doubled verts.
The correct work flow is to do the following:-
- Delete half of the LP mesh
- Unwrap that remaining half
- Add the mirror modifier so that the mesh is mirrored
- Make sure Merge is checked
- Bake with the mirror modifier on and visible to the renderer
- Apply the mirror modifier (with merge checked)
- Done!
You can also mirror manually for the same effect.
If you can let me know exactly how you baked the mesh, then we should be able to get a fix sorted out for you
Thank you. And I think Andy could very well be right. Could I persuade you to make a blend available with your lo-res kneebad and the hi-res so we can take a look?
Heās correct that you canāt just bake using one half because then after the mirror operation youād get a different set of normals down the mirror seam due to normals being shared/average after the mirror operation (as they should be).
So you need to apply the mirror operation with merging before baking.
Thereās an example of someone else falling into that trap here and his scene was fixed.
http://projects.blender.org/tracker/?group_id=9&atid=498&func=detail&aid=26996
Are you by any chance using instancing for the other half using just a negative scale on the object? Though I am not 100% sure I think this might not work on normal mapping in Blender at this point. In which case this support has to be put in.
Many game engines donāt support non uniform scaling at all though but obviously it should be supported by Blender.
As explained here http://wiki.blender.org/index.php/Dev:Shading/Tangent_Space_Normal_Maps they are almost all incompatible. Though many packages agree with each other on channel layout this does not mean that their tangent space is computed the same way. Almost every package out there computes tangent space a different and often proprietary way which creates problems when you try to move tangent space normal maps across packages.
xNormal and Blender uses the same spaces and same channel layout. However, there are still ways in which things can go wrong in regards to normal mapping in any tool and hopefully weāll get cleared up whatās going on in this case.
Wow, this is amazing. It definitely belongs to the gallery. Love to see better and better works made using Blender in the workflow.
This isnāt the case!
You donāt understand that these maps are rendered correctly in so many other apps except blender!
So many years Iām complaining for this, Iām not alone. Do some search around the net and see how bad reputation blender has on this. Blender can not support normal maps correctly! 2.49b or 2.5x. This last 2.5x canāt support even the 2.49b projects having nor maps.
In general, the only wrong on the above posted maps is the bad placement of the seams. Seams are more or less visible on normal maps. We have to live with it. So, next time select darker areas to place them.
Tested using blender all BI versions, zbrush, 3dcoat, Thea.
Donāt insist that all other apps are wrong on this, except blender.
@michalis,
Iām sorry my friend, but you are incorrect 
Pre 2.57, mirrored normal maps were not supported in the new Blender code as tangents were not stored with a sign. Post 2.57, mirrored normal maps work perfectly as the sign is now stored correctly. I have no problem making maps with mirrored normal maps in Blender 2.57+
The issue is that people are baking the maps incorrectly and not that Blender 2.57+ is broken and doesnāt support mirroring.
I would be happy to post some images later to prove that they work correctly 
btw, I would trust mmikkelsenā¦he knows what he is talking about 
@metalliandy
Iām not incorrect.
Can you, please, inform me if BI is compatible with maya or LW format (normal-maps) ?
Blender uses the same swizzle as Maya (X+Y+Z+) but the tangent basis is different, as it is between almost all applications, so there is never a guarantee of a perfect match. Unless the tangent basis is 100% matched with the application that is baking the map and with the offline renderer, game engine or viewport shader that is displaying you will never get a perfect normal map when transferring them between applications.
The colours and gradients in the normal map are there to represent the differences between the low and high poly models and if the tangent basis doesnāt match then these colours and gradients cant be translated accurately into correct lighting, which is why you can get the smoothing errors (which are simply the visual difference between the tangent basis of the baked map and the renderer)
This is not an issue with Blender itself, it a problem with lack of a tangent basis standard between applications.
Blender uses the same tangent basis as xNormal and is a pretty decent match with Marmoset (98%) and most other applicationsā¦certainly much better than the old tangent basis that Blender used.
This problem is once of the reasons why people use smoothing groups on game assets.
Adding smoothing groups reduces the gradients and in doing so, also reduces the errors because there is less chance of an incorrect translation happening as the areas with smoothing groups tend to be flatter and resulting values tend to be much closer to 128,128,255 (in addition to other factors)
As I said in my previous post, many people do not know the correct way to bake normal maps correctly (mirrored or otherwise) and in my experience, errors are very rarely down to errors within the applications. Itās the same reason why people get the black line issue when baking with smoothing groups. Itās just a lack of understanding, the spreading of misinformation, or simply that people donāt know any betterā¦
I donāt know much about Lightwave, so I cant really comment on their normal maps, but itās almost guaranteed that they are using their own implementation normal maps, with a custom tangent basis that was developed by Newtek
Mirrored normal maps work perfectly in Blender and I have zero issues with using them.
Just wanted to say what metalliandy is saying is absolutely correct. All the packages use different implementations to generate tangent space unfortunately. Itās only the swizzle that matches between different tools. So unless two packages use the same tangent space implementation you will not be able to move normal maps back and forth without some issues.
That being said you are correct that historically many people have pointed out that Blenderās normal mapping had its issues but Blenderās tangent spaces were subjected to a major overhaul which has been in since 2.57+ and mirroring has been confirmed to work many many times so I suggest you keep an open mind and try it out.
xNormal and Blender use the exact same standard now ā>
http://wiki.blender.org/index.php/Dev:Shading/Tangent_Space_Normal_Maps
And it is explained there why this is a very good standard that others should be encouraged to follow.
The overall point is that this implementation promises to generate the same tangent spaces even when the mesh data has undergone certain modifications which are common when transferring assets between tools. Other implementations do not offer this.
I think you should give the new spaces a chance. They are pretty rock solid.
Theyāve solved all seams related problems for us here at work and weāve shipped two games with them now. Before we put them in there were frequent complaints about seams. Afterwards, not a peep. The artists are extremely happy with them.
The only issue in Blender I can think of at this point is object instancing using negative scales. Everything else should work incl. the mirror modifier which we have also confirmed.
Here are a couple of other artists who were having issues that might be typical among artists.
http://projects.blender.org/tracker/?func=detail&atid=498&aid=28260&group_id=9
http://projects.blender.org/tracker/?group_id=9&atid=498&func=detail&aid=26996
And in both cases it came down to a lack of understanding (I mean no disrespect, it is indeed complicated).
In the future if you have an issue you are welcome to submit a bug report and we would be happy to look into it.
Be sure to attach a blend file if you do.
Cheers,
Morten.
Awsome, wish I had as much skill as you
Looks very formidable What is this for? Recreation?
Really like the amount of detail in this render, and the glare and cigarette were nice touches.
The boss of bosses! Fantastic 10 +