dont worry about it.what is matter, that how it looks in a game engine.as long as it looks good in engine, it doesnt really matter how it looks in max.
Looks like it was the edge padding. Good to know 8 and above is the norm. I had to go up to 13. On another note, I know you're suppose to split your smoothing groups when using a normal map, but I didn't do that for this mesh that has 4 smoothing groups and the normal map looks the same.
Here's what my uvs look like with the smoothing groups stitched
Here's what they would look like if I were to follow that normal map baking rule
Which should I be doing?
Also, I probably know the answer to this but I'm gonna ask it anyways. Is this due to my low poly not matching my high in this area or due to my cage? I spent awhile getting all of the ray misses out but this isn't going away so I'm assuming it's the low not matching the high...
I didn't know about that minimum 8 too, but when i think, its totally understandable. I never had problems with this btw.
Anyway Chase, your uv map is really strange.It causes brutal distortion on the texture.Just assign a checker map to your model, and you will see what i am talking about.
Obscura - I applied a checker texture to it before and tried relaxing the faces and edges so without adding more seams I don't know what else do to. I'm going to go back and retopologize this to get a more evenly distributed model. I just used topogun cause I find that easier to create the low poly. Is there any other more efficient ways? I know I could start from scratch but I'm not sure how to match the models the way topogun accomplishes it. This is my first bake.
Atticus - Nope in Max. What should I do about the smoothing groups and the normal map? I feel like I should be splitting the UV shells to follow the rules of a normal map bake, but if I'm not getting any errors why add more work/seams ya know.
You must have a UV split where you have smoothing splits. UVs are much more precise than texture maps so the edges will bleed and you will get artifacts, so you have to split up the UVs and add padding. It doesn't work the other way around though, if you have a UV split you don't necessarily need a smoothing split. There's a handplane video on the polycount news page that explains this really well.
To be honest I'm not sure that's what's causing your problem, it does look like you have hard edges inside your uv islands but the seams look like raycast misses from a non averaged cage but Max uses averaged cages by default so Im not sure. Hopefully someone more knowledgeable than me can point you in the right direction.
I know about the uv splits and smoothing splits but I wanted to make sure. I think that's what's causing the edge bleeding. Splitting up the uv seams should answer my distortion problem too. What's the difference between an average projection cage and the other type? I'm still confused on that
This was something I recently came across and me and my friend did some testing and came up with some weird results.
In simple, if your format to be is a .tga pad it a lot. Gaussian blur it, or re-render with extra padding, as others have mentioned this is the reason for those weird black edges.
However, my friend suggested I tried a different format so I tried .bmp bitmap and with or without any padding I didn't get any dark edges showing (in unity & marmoset).
I was and am still confused about this, I'm fairly new to texturing/mip-mapping etc. so if anyone has any suggestions as to why file format matters so much please share with me. The only reason I can think of is that mip-mapping isn't actually working with the bmp's at all, or is handling it differently compared to targas. Or perhaps the issue lies with the alpha channel in a targa, I'll do some more testing but as a newbie I still feel pretty in the dark here.
More Google searching to do.
Just PM me your email so I can send it to you. Also, I'm thinking of detaching the top cement block, retopoing the two pieces for cleaner geo since this was a quick and dirty test, and then bake using the match multi material IDs option. Or I could just retopo and combine them as one like I have. Any pros or cons to either?
Replies
Increase it to a very minimum of 6 (even if it means adjusting UV islands), and boost that texture filtering as high as you dare.
Here's what my uvs look like with the smoothing groups stitched
Here's what they would look like if I were to follow that normal map baking rule
Which should I be doing?
Also, I probably know the answer to this but I'm gonna ask it anyways. Is this due to my low poly not matching my high in this area or due to my cage? I spent awhile getting all of the ray misses out but this isn't going away so I'm assuming it's the low not matching the high...
Anyway Chase, your uv map is really strange.It causes brutal distortion on the texture.Just assign a checker map to your model, and you will see what i am talking about.
Are you baking in xnormal?
Atticus - Nope in Max. What should I do about the smoothing groups and the normal map? I feel like I should be splitting the UV shells to follow the rules of a normal map bake, but if I'm not getting any errors why add more work/seams ya know.
To be honest I'm not sure that's what's causing your problem, it does look like you have hard edges inside your uv islands but the seams look like raycast misses from a non averaged cage but Max uses averaged cages by default so Im not sure. Hopefully someone more knowledgeable than me can point you in the right direction.
In simple, if your format to be is a .tga pad it a lot. Gaussian blur it, or re-render with extra padding, as others have mentioned this is the reason for those weird black edges.
However, my friend suggested I tried a different format so I tried .bmp bitmap and with or without any padding I didn't get any dark edges showing (in unity & marmoset).
I was and am still confused about this, I'm fairly new to texturing/mip-mapping etc. so if anyone has any suggestions as to why file format matters so much please share with me. The only reason I can think of is that mip-mapping isn't actually working with the bmp's at all, or is handling it differently compared to targas. Or perhaps the issue lies with the alpha channel in a targa, I'll do some more testing but as a newbie I still feel pretty in the dark here.
More Google searching to do.