After reading all normal map threads a few times, I think I understand a lot of it now. But a few things are still unclear to me. Does a normal map overwrite the normals of a low poly or does it only alter it? And which normals will be affected? The face normals or the vertex normals? Or both?
From my understanding, assigning a normal map to a low poly will replace the low polys face normals. But I am not sure if this is correct. I would be grateful if someone could help me figuring this stuff out.
And... sorry for my poor english. It is not my native language.
Tangent space normal maps only alter the vertex normals, a world or object space normal map overwrites all vertex normal information, but is more limited and not commonly used much.
Also there's different ways to generate normal maps because of how the vertex normals are taken to account, and can cause your normal map and game engine (or render) to be out of sync, and can cause minor shading artifacts.
Ah! That makes sense. Because the tangent space normal map relies on the vertex normals and therefore I should not delete any geometry. But I can do it for world or object space normal maps, because both of it are not relative but use an absolute space and doesn't care for the vertex normals. Right?
Thank you, ZacD!
Unfortunately I have another question regarding gradients in normal maps. I have trouble to define these. This time I'm just guessing, because I have no clue: Gradients comes with soft edges. The steeper the angle between two connected faces, the greater the gradient, in order to let the connecting edge appear soft. And what I see in the viewport of maya when shaded as a soft edge, is then baked as a gradient into the normal map to let again the shader know how to calculate the... I don't know... lighting information or vertex angle, shading???
I'm 100% sure this is a horrible wrong wording and the spelling is terrible.
Can someone come up with a proper explanation what those gradients exactly are and why they appear as they do, please? I'm trying to understand all of this stuff.
[Edit: I think I figured it out now. Gradiation in a normal map is created when the vertex normals pointing in a different direction than the face normals. This means: The greater the deviations in the angle between the vertex normals and the face normals are the more extreme the shading occurs.
If we would look at two neighboring faces (90°) whose edge was set to Soft Edge, the vertex normals of the edge would be averaged (to 45°). In the viewport the impression of a soft shading occurs.
And in the end this shading is exactly what is transmitted later in the normal map as gradiation represented in colors. Correct?]
It's rather simple actually, and you don't need to think of it in terms of hard or soft to understand the concept. The gradients in the normal map compensate for the mesh normals on the low poly. The more extreme the mesh normals, the stronger the gradient.
By more extreme, I mean specifically: the more the angle of the vertex normal diverges from the angle of the face normal. So yes, the last half of your post exactly. Amusingly, I responded to your question before reading the part below the edit line and wrote an almost identical summary to yours.
It's rather simple actually, and you don't need to think of it in terms of hard or soft to understand the concept. The gradients in the normal map compensate for the mesh normals on the low poly. The more extreme the mesh normals, the stronger the gradient.
By more extreme, I mean specifically: the more the angle of the vertex normal diverges from the angle of the face normal. So yes, the last half of your post exactly. Amusingly, I responded to your question before reading the part below the edit line and wrote an almost identical summary to yours.
Yay! I'm correct! Thank you for the confirmation, Earthquake!
Sorry for necroing this thread. So it's okay to add supporting loops to either side of those cylinders in order to get a non-skewed bake. Do you guys delete those loops after the successful bake or do you leave it for future potential baking?
Okay I read about how the extra support loops should contribute to the design of the object, not only just for capturing good bakes.
Sorry for necroing this thread. So it's okay to add supporting loops to either side of those cylinders in order to get a non-skewed bake. Do you guys delete those loops after the successful bake or do you leave it for future potential baking?
Okay I read about how the extra support loops should contribute to the design of the object, not only just for capturing good bakes.
Whatever geometry you use for the bake should also be used for the final mesh that you are using the bake on. If you remove the supporting geometry (loops in this case) it will alter the vertex normals and cause shading errors that you want to avoid.
One thing you can do if you cant afford extra geometry is add the extra loops> bake to an Object space normal map> remove the loops> convert the OS normal map back to tangent space using the no loops mesh. You can do this in xNormal (tools menu) and in Handplane.
Sorry for necroing this thread. So it's okay to add supporting loops to either side of those cylinders in order to get a non-skewed bake. Do you guys delete those loops after the successful bake or do you leave it for future potential baking?
Okay I read about how the extra support loops should contribute to the design of the object, not only just for capturing good bakes.
Whatever geometry you use for the bake should also be used for the final mesh that you are using the bake on. If you remove the supporting geometry (loops in this case) it will alter the vertex normals and cause shading errors that you want to avoid.
One thing you can do if you cant afford extra geometry is add the extra loops> bake to an Object space normal map> remove the loops> convert the OS normal map back to tangent space using the no loops mesh. You can do this in xNormal (tools menu) and in Handplane.
Hope that helps!
You can add geometry for an object map bake, and remove it, but for a tangent space normal map, you cannot add and remove geometry, unless that area is completely flat and it's not changing the shading at all.
Sorry for necroing this thread. So it's okay to add supporting loops to either side of those cylinders in order to get a non-skewed bake. Do you guys delete those loops after the successful bake or do you leave it for future potential baking?
Okay I read about how the extra support loops should contribute to the design of the object, not only just for capturing good bakes.
Whatever geometry you use for the bake should also be used for the final mesh that you are using the bake on. If you remove the supporting geometry (loops in this case) it will alter the vertex normals and cause shading errors that you want to avoid.
One thing you can do if you cant afford extra geometry is add the extra loops> bake to an Object space normal map> remove the loops> convert the OS normal map back to tangent space using the no loops mesh. You can do this in xNormal (tools menu) and in Handplane.
Hope that helps!
You can add geometry for an object map bake, and remove it, but for a tangent space normal map, you cannot add and remove geometry, unless that area is completely flat and it's not changing the shading at all.
Indeed! Should have clarified that in my post really...
You can add geometry for an object map bake, and remove it, but for a tangent space normal map, you cannot add and remove geometry, unless that area is completely flat and it's not changing the shading at all.
I just wanted to clarify my previous clarification a little
You should always avoid editing the geometry used to bake when the target is a tangent space normal map. Though an area may seem flat, in cases where CC subdivision is used you will only ever get flat geometry in very rare cases such as totally disconnected planes (I'm not even 100% sure if they are totally flat off the top of my head) and though it may be imperceivable at first glance there is almost always some slight curvature to the underlying surface, which will be baked into the TS normal map and thus the geometry will not match the bake if altered (this is the exact reason why 8bit normals can still fail on seemingly flat surfaces where the banding is present). These differences can be especially noticeable with high spec + high smoothness/low roughness metallic materials.
Though it may be possible to get away with doing this where a user is experienced enough to know the neuances of normal maps and what exactly the result will be when the changes are made (users such as yourself ZacD), it is not something that I think should be encouraged as breaking the rule can lead to the slippery slope of people editing meshes post bake because they heard is was ok and then told someone else etc. After a while we may well end up in the situation where we were a few years ago where misinformation was rife and was propagated by people who don't really understand the core issues that we have worked so hard to solidify in threads such as these.
Hope that all makes sense.
I will post some examples of this when I get back to my main PC later.
You can add geometry for an object map bake, and remove it, but for a tangent space normal map, you cannot add and remove geometry, unless that area is completely flat and it's not changing the shading at all.
I just wanted to clarify my previous clarification a little
You should always avoid editing the geometry used to bake when the target is a tangent space normal map. Though an area may seem flat, in cases where CC subdivision is used you will only ever get flat geometry in very rare cases such as totally disconnected planes (I'm not even 100% sure if they are totally flat off the top of my head) and though it may be imperceivable at first glance there is almost always some slight curvature to the underlying surface, which will be baked into the TS normal map and thus the geometry will not match the bake if altered (this is the exact reason why 8bit normals can still fail on seemingly flat surfaces where the banding is present). These differences can be especially noticeable with high spec + high smoothness/low roughness metallic materials.
Though it may be possible to get away with doing this where a user is experienced enough to know the neuances of normal maps and what exactly the result will be when the changes are made (users such as yourself ZacD), it is not something that I think should be encouraged as breaking the rule can lead to the slippery slope of people editing meshes post bake because they heard is was ok and then told someone else etc. After a while we may well end up in the situation where we were a few years ago where misinformation was rife and was propagated by people who don't really understand the core issues that we have worked so hard to solidify in threads such as these.
Hope that all makes sense.
I will post some examples of this when I get back to my main PC later.
Thanks very much for the clarification! So that means the studio that gave me the test doesn't know how to bake their normal maps. I inset-ed a single flat polygon in order to avoid skewed normal maps for my bolts, while still staying within their requested triangle count. And they were so anal about it and pointed it out as my failure....
I posted for a problem like these and I was sent here so here my problem: I'm trying to bake a normal map and after reading here I added more topology to my low poly mesh and added some support loops around my hard edge but I still get some wierd artifacts.
with normal map:
without:
I baked the normal in Substance Painter 1.7.2 without cage. It's the same error or it's not related ?
Just bake with a cage... Not doing so is just asking for trouble, and it's certainly the source of your problems here. The support loops you have added have done nothing other than exponentially ramp up how costly this model is to rasterize.
I do not believe that support loops on a low poly are ever a good solution for controlling shading, when the same amount of extra geometry could be used to actually round out those edges and minimize the difference between the high and low poly. I challenge anyone to find a single form that shades better by throwing a hack on top rather than just giving the whole model enough detail.
Yes! I have but i'll be brutally honest. I don't understand how that relates to my problem and if i don't understand it, I can't apply it to my situation.
Maybe you can do a dumb down version for me? because the post reads like they are talking about baking flat faces like on a crate and that's not the problem i'm having.
Yes! I have but i'll be brutally honest. I don't understand how that relates to my problem and if i don't understand it, I can't apply it to my situation.
Maybe you can do a dumb down version for me? because the post reads like they are talking about baking flat faces like on a crate and that's not the problem i'm having.
The post describes your situation almost exactly, even the example images have cylindrical geometry that is similar to yours. If you can give me more information as to where you are stuck I may be able to help but I can't see how I can make it any more applicable to what you're doing.
The first issue is explained in detail in the first post, you are not going to get a better explain, break down, and possible solutions. The last image just a projection issue where the left and right sides are projecting onto each other because they are so close. You'll need a cage and to make sure the cage fits into that area well, so that there isn't any cross projection across that gap.
Alright! Not a clue as to what cage means. My only point of reference is using smooth mesh and i can't understand how that works in this situation.
Are you talking about envelopes?
As for the front page, I'm still clueless on its implementation. It sounds like he's talking about not using 90* degree faces on an object because they won't appear in the bake.
I apologies if this is annoying, I've been stuck on this problem for a week.
Yeah! Thanks. I've read that along with everything else I've been able to find on the subject, Still. I've got my model up if you want to have a look at it.
I'm going to come out and say what no-one is outright saying:
Asking someone to point out step by step exactly what is wrong, how to fix it, and (from you posts) the information behind this is not a huge ask. But it is asking someone else to give their time end energy to help you directly. This is not a bad thing, it's something that polycount thrives on. If someone steps in to help you in this way, that's great.
However. Everything you need to know has been put in front of you. Any one part of it you don't understand is probably a couple of clicks away from that. What you might be missing is the experience of learning from your own mistakes and experiments. Like trying out the different shapes illustrated on the first page of this thread.
For my own advice: Try chaging your cage so that the envelope pushes out at closer to a 90* angle and the bevels are baked from the sides.
Alright! That's good to know in the future, I won't ask for people to check my model.
Anyhow, Been playing around with my low poly mesh and adding in slopes after a friend explained how Maya's bakes normal's.
It's slightly better, still has some waviness to it but only if you view it directly from a 90 degree angle not every direction.
Still gotta do all the other wavy edges and the crack seems like a lost cause, but its progress~!
What's an acceptable amount of waviness in an object? Because for me I'd think it would need to be absolute but i don't think that's do-able even with adjustments and suttee slopes there's still minor waves.
The other option is just painting the normal's but that's just a stop gap measure.
So part of the problem your having is that the high poly model has a very tight edge that is only visible up close. And your low poly doesnt have a lot of sides - so when viewed up close it looks very blocky.
Think about how the model will be viewed and create an edge that is easily readable from that distance.
Well, My model will be viewed very close up as its a first person weapon for an older game and everything that your talking about theses tight edges is art style of the game.
It's all very low poly but hard edged and the lowest subdivision is about 8 sides for my cylinders while the highest is 16.
If you see waviness and baked your asset properly, that is the correct interpretation. Can't stress enough that waviness is not an error or artifact, it is a natural representation of what you baked.
If you don't want waviness you have to author your assets with that in mind. An 8 sided high poly cylinder with rounded tops and bottoms will not cause waviness on an 8 sided low poly cylinder. A high poly cylinder with many sides causes waviness because the low poly mesh does not match its curvature:
As a side note I wanted to mention you saying this, "Alright! That's good to know in the future, I won't ask for people to check my model." This comes off, at least to me, as shutting out helpful advice because it isn't giving you all the answers upfront. Don't be put off from sharing your work when you need feedback because people tell you that you should use learning resources and spend time working towards understanding them. However, I do think they're right in suggesting that you need to spend more time understanding the process you're using. It seems like you're doing the same thing over and over without knowing why it's not getting any better.
Alright! I think your saying that waviness is dictated by the amount of
sides an object has? for example 16 > 8 will create some waviness
however if its 16 > 16 there will be none, in which case that's
raises the question of what's the point if your models geometry isn't
any lower?
However, If you went to something like 12 sides instead of 16 that's loosely about the same amount would that be better? then after that bake down from 12 to a lower subdivision.
As for what your saying below, it sounded like the
person was giving me a helpful hint that it wasn't the best thing to
ask. I took the hint, Was i wrong to?
Further more, Your completing right as I'm just work through this as quickly as possible. I'm starting to understand it now that someones filled me in on something however i keep coming back to the same thing, its really driving me nuts if I'll be honest.
1. Not exactly, the fewer the sides on the low poly, the wavier it will be, but you can avoid the waves if the lowpoly and high poly match up. It's more of a fix than a general solution.
2. Not sure exactly what you are getting at.
3. No comment
4. Honestly cubes and cylinders are worst case scenarios for baking, 90 degree angles don't bake well because you should avoid them or use them in areas that wont be seen close up. In modern games you can normally just add more geometry. If you are doing something super lowpoly like 200 tris weapons, then getting perfect 90 degree bakes are important, but for a modern 10k weapon it's not an issue.
Don't sweat the little things. And try to avoid 90 degree angles.
"2. Not sure exactly what you are getting at." I'm getting at is the
following. "1. Not exactly, the fewer the sides on the low poly, the
wavier it will
be, but you can avoid the waves if the lowpoly and high poly match up.
It's more of a fix than a general solution." If i increase the polycount
of my low res mesh to something a little higher like 12 it should bake
more correctly?
As for cylinders, I'm just glad that i had this
problem now not later when i'm working on other things. It's allowed me
to take notes.
What I think is the problem is that you're fixated on counting sides when you should be understanding the process.
"If i increase the polycount of my low res mesh to something a little higher like 12 it should bake more correctly?" This is what's flawed in your thinking: Waviness is not incorrect. If it's present, it is the direct and absolutely correct output of whatever you input.
When you bake a normal map, you're capturing details that you apply to a lower detailed model for cheaper than you could give it that actual detail. It's data that is telling your model to shade based off different surface normals than the geometry actually has. Waviness is just what you get when you have a drastic difference in high and low geometry at a 90 degree angle like that.
That's the joke of the thread title: you put waviness in your normal map, because that's how you made your models.
If I were you I'd take a break, if you can afford to, and focus on understanding how a normal map gets produced in the first place. Knowing high + low + cage = normal is not nearly enough. You need to understand what those normal map colors represent, how the rays are fired from mesh to mesh in order to produce it, how mesh normals and thus mesh geometry are inherently tied into all these things.
"If I were you I'd take a break, if you can afford to, and focus on
understanding how a normal map gets produced in the first place. Knowing
high + low + cage = normal is not nearly enough. You need to understand
what those normal map colors represent, how the rays are fired from
mesh to mesh in order to produce it, how mesh normals and thus mesh
geometry are inherently tied into all these things."
Well, I understand completely the colour values and how normal's function and how they are created but things like the workflow and techniques is what i want to know. There are a few things that have helped theory wise such as finding out how Maya approaches baking which did explain a lot of what the thread talks about and keywords that people still seem to use from older suites.
Strangely enough there isn't much information on workflow and techniques
Dude that doesn't make any sense, why would you make a normal map of your low mesh?
It's suppose to be like this, who ever said that your high poly object is suppose to be low is wrong.
You don't understand the point being made. They were asking how to remove waviness in the normal map at the caps on an extremely low poly cylinder. The example with the blue cylinders I made is exactly how to do that.
You should share your example file, It sounds very useful.
Sure. I've redone it real quick, I didn't bother to bake the assets before as I hoped it was clear enough as is: The low poly is the most basic 8 sided cylinder possible.
Notice that your example has waviness present as well due to the same reason: The low poly doesn't match the high. There's not really anything special about this though. We're simply seeing the natural differences between various meshes compared to a low poly. This is the same information in the OP just presented in a different manner. Now I'm not necessarily advocating for or against baking like this either, just answering the original question of how to remove the waviness they didn't want present. By making the high and low cylinders have equal segments, logically the only difference between mesh normals is the rounded corners that you may want to capture.
Wooh! You did it. I feel stupid now for asking but how do i use theses?
Could you create some steps on how to bake them? I did bake with
waviness_MedHigh and waviness_LowHigh onto waviness_Low and i want to
know if this is right?
Objects are waviness_Low with respected maps applied.
It looks right to me but i'm unsure if
those black corners and edges are the shader or the map? Also, Why is
the shell set to 21? What are the benefits of having a larger shell and
how does the bevel effect waviness and finally if i go to avenger it and
bake again the waves reappear but if its got hard tops then they don't.
Why is that?
You've got my full and undivided attention, Tell me your secrets.
I baked using Substance Designer. Not sure what the steps are in Maya, but _LowHigh _MedHigh and _SuperHigh should be baked down to _Low using the waviness_Cage file for the projection, which I guess is what you meant by "shell". It has to be pushed out like that because otherwise there will be projection errors. The cage must always encompass the high poly model entirely and have averaged normals.
Your bakes look like they have gaps from not using the cage, but it's hard to say if you're using different software. If you bake with the cage file your normal maps might look slightly different due to Maya's baker but the shaded result should be identical.
The waviness_Low file has split normals/smoothing groups on the top and bottom only to reduce the gradients in the normal map, but if it were set to all averaged normals like how the cage file is, there'd be no effective difference other than stronger gradients. It would look practically the same when applied to the model.
As for the beveled edge affecting waviness: a larger/smaller bevel might make it more or less apparent but it won't combat waviness which is only due to the number of sides on the cylinders.
Been busy reading up on techniques recently but i have all my control edges in and its worked on everything but theses tapered ends are still giving me a bit of trouble and and it turns out there's another way to do those rounded edges that "somedoggy" mention before.
Still at the end of the day, it comes back to this taper ends, a lot of my models use tapers for the ends and i'd like to know how to get them to bake correctly.
I can't stress enough the fact that waviness is not incorrect. It is the natural result of correctly baked normals between two cylindrical objects with a non-equal # of sides. If you don't want it, just match the number of sides:
I've got a question about normals. I noticed in blender that the smooth shading looked different depending on whether or not I had triangulated my mesh yet, so I assumed that I should triangulate it and then bake from high poly down onto normals. The result I'm getting in 3DO looks kind of weird though. Any thoughts? Should I have used an edge split or just flat shading instead, since it's such a boxy shape?
You should always triangulate before you bake, because the your modeling application, your baking application, and your target engine might triangulate the mesh differently.
The shading errors you are seeing are most likely from not using a sync'd normal map workflow.
You should always triangulate before you bake, because the your modeling application, your baking application, and your target engine might triangulate the mesh differently.
The shading errors you are seeing are most likely from not using a sync'd normal map workflow.
So does this mean adding supporting edges or chamfers or something? Is the problem in how I baked the normal map or in the model itself? It looks fine in blender. It's only when I take it out that it messes up. I noticed that the normal map I get if I bake it in xNormal (right) looks wacked compared to the one I get from Blender itself (left).
Replies
Thank you, ZacD!
Unfortunately I have another question regarding gradients in normal maps. I have trouble to define these. This time I'm just guessing, because I have no clue: Gradients comes with soft edges. The steeper the angle between two connected faces, the greater the gradient, in order to let the connecting edge appear soft. And what I see in the viewport of maya when shaded as a soft edge, is then baked as a gradient into the normal map to let again the shader know how to calculate the... I don't know... lighting information or vertex angle, shading???
I'm 100% sure this is a horrible wrong wording and the spelling is terrible.
Can someone come up with a proper explanation what those gradients exactly are and why they appear as they do, please? I'm trying to understand all of this stuff.
[Edit: I think I figured it out now.
Gradiation in a normal map is created when the vertex normals pointing in a different direction than the face normals.
This means: The greater the deviations in the angle between the vertex normals and the face normals are the more extreme the shading occurs.
If we would look at two neighboring faces (90°) whose edge was set to Soft Edge, the vertex normals of the edge would be averaged (to 45°). In the viewport the impression of a soft shading occurs.
And in the end this shading is exactly what is transmitted later in the normal map as gradiation represented in colors. Correct?]
By more extreme, I mean specifically: the more the angle of the vertex normal diverges from the angle of the face normal. So yes, the last half of your post exactly. Amusingly, I responded to your question before reading the part below the edit line and wrote an almost identical summary to yours.
Okay I read about how the extra support loops should contribute to the design of the object, not only just for capturing good bakes.
One thing you can do if you cant afford extra geometry is add the extra loops> bake to an Object space normal map> remove the loops> convert the OS normal map back to tangent space using the no loops mesh. You can do this in xNormal (tools menu) and in Handplane.
Hope that helps!
You should always avoid editing the geometry used to bake when the target is a tangent space normal map. Though an area may seem flat, in cases where CC subdivision is used you will only ever get flat geometry in very rare cases such as totally disconnected planes (I'm not even 100% sure if they are totally flat off the top of my head) and though it may be imperceivable at first glance there is almost always some slight curvature to the underlying surface, which will be baked into the TS normal map and thus the geometry will not match the bake if altered (this is the exact reason why 8bit normals can still fail on seemingly flat surfaces where the banding is present). These differences can be especially noticeable with high spec + high smoothness/low roughness metallic materials.
Though it may be possible to get away with doing this where a user is experienced enough to know the neuances of normal maps and what exactly the result will be when the changes are made (users such as yourself ZacD), it is not something that I think should be encouraged as breaking the rule can lead to the slippery slope of people editing meshes post bake because they heard is was ok and then told someone else etc. After a while we may well end up in the situation where we were a few years ago where misinformation was rife and was propagated by people who don't really understand the core issues that we have worked so hard to solidify in threads such as these.
Hope that all makes sense.
I will post some examples of this when I get back to my main PC later.
Thanks again!
with normal map:
without:
I baked the normal in Substance Painter 1.7.2 without cage.It's the same error or it's not related ?
I do not believe that support loops on a low poly are ever a good solution for controlling shading, when the same amount of extra geometry could be used to actually round out those edges and minimize the difference between the high and low poly. I challenge anyone to find a single form that shades better by throwing a hack on top rather than just giving the whole model enough detail.
Norm
Low
High
Maybe you can do a dumb down version for me? because the post reads like they are talking about baking flat faces like on a crate and that's not the problem i'm having.
Here's a copy of the mesh, Enjoy!
https://drive.google.com/open?id=0B8P_s34EMKFGRUZjQnZEek54ckE
Are you talking about envelopes?
As for the front page, I'm still clueless on its implementation. It sounds like he's talking about not using 90* degree faces on an object because they won't appear in the bake.
I apologies if this is annoying, I've been stuck on this problem for a week.
http://polycount.com/discussion/107196/youre-making-me-hard-making-sense-of-hard-edges-uvs-normal-maps-and-vertex-counts/p1
http://wiki.polycount.com/wiki/Texture_Baking
That would be easier than shots in the dark.
Asking someone to point out step by step exactly what is wrong, how to fix it, and (from you posts) the information behind this is not a huge ask. But it is asking someone else to give their time end energy to help you directly. This is not a bad thing, it's something that polycount thrives on. If someone steps in to help you in this way, that's great.
However. Everything you need to know has been put in front of you. Any one part of it you don't understand is probably a couple of clicks away from that. What you might be missing is the experience of learning from your own mistakes and experiments. Like trying out the different shapes illustrated on the first page of this thread.
For my own advice: Try chaging your cage so that the envelope pushes out at closer to a 90* angle and the bevels are baked from the sides.
Anyhow, Been playing around with my low poly mesh and adding in slopes after a friend explained how Maya's bakes normal's. It's slightly better, still has some waviness to it but only if you view it directly from a 90 degree angle not every direction.
Still gotta do all the other wavy edges and the crack seems like a lost cause, but its progress~!
The other option is just painting the normal's but that's just a stop gap measure.
Think about how the model will be viewed and create an edge that is easily readable from that distance.
It's all very low poly but hard edged and the lowest subdivision is about 8 sides for my cylinders while the highest is 16.
Here's my workflow.
If you don't want waviness you have to author your assets with that in mind. An 8 sided high poly cylinder with rounded tops and bottoms will not cause waviness on an 8 sided low poly cylinder.
A high poly cylinder with many sides causes waviness because the low poly mesh does not match its curvature:
As a side note I wanted to mention you saying this, "Alright! That's good to know in the future, I won't ask for people to check my model."
This comes off, at least to me, as shutting out helpful advice because it isn't giving you all the answers upfront. Don't be put off from sharing your work when you need feedback because people tell you that you should use learning resources and spend time working towards understanding them. However, I do think they're right in suggesting that you need to spend more time understanding the process you're using. It seems like you're doing the same thing over and over without knowing why it's not getting any better.
However, If you went to something like 12 sides instead of 16 that's loosely about the same amount would that be better? then after that bake down from 12 to a lower subdivision.
As for what your saying below, it sounded like the person was giving me a helpful hint that it wasn't the best thing to ask. I took the hint, Was i wrong to?
Further more, Your completing right as I'm just work through this as quickly as possible. I'm starting to understand it now that someones filled me in on something however i keep coming back to the same thing, its really driving me nuts if I'll be honest.
2. Not sure exactly what you are getting at.
3. No comment
4. Honestly cubes and cylinders are worst case scenarios for baking, 90 degree angles don't bake well because you should avoid them or use them in areas that wont be seen close up. In modern games you can normally just add more geometry. If you are doing something super lowpoly like 200 tris weapons, then getting perfect 90 degree bakes are important, but for a modern 10k weapon it's not an issue.
Don't sweat the little things. And try to avoid 90 degree angles.
As for cylinders, I'm just glad that i had this problem now not later when i'm working on other things. It's allowed me to take notes.
"If i increase the polycount of my low res mesh to something a little higher like 12 it should bake more correctly?"
This is what's flawed in your thinking: Waviness is not incorrect. If it's present, it is the direct and absolutely correct output of whatever you input.
When you bake a normal map, you're capturing details that you apply to a lower detailed model for cheaper than you could give it that actual detail. It's data that is telling your model to shade based off different surface normals than the geometry actually has. Waviness is just what you get when you have a drastic difference in high and low geometry at a 90 degree angle like that.
That's the joke of the thread title: you put waviness in your normal map, because that's how you made your models.
If I were you I'd take a break, if you can afford to, and focus on understanding how a normal map gets produced in the first place. Knowing high + low + cage = normal is not nearly enough. You need to understand what those normal map colors represent, how the rays are fired from mesh to mesh in order to produce it, how mesh normals and thus mesh geometry are inherently tied into all these things.
Well, I understand completely the colour values and how normal's function and how they are created but things like the workflow and techniques is what i want to know. There are a few things that have helped theory wise such as finding out how Maya approaches baking which did explain a lot of what the thread talks about and keywords that people still seem to use from older suites.
Strangely enough there isn't much information on workflow and techniques
Thanks for posting this, it took me a while to understand and test.
It's suppose to be like this, who ever said that your high poly object is suppose to be low is wrong.
The low poly is the most basic 8 sided cylinder possible.
Link to the files:
https://drive.google.com/open?id=0B4vo_o3utgw6Zlc5djBGYl8xUDA
Notice that your example has waviness present as well due to the same reason: The low poly doesn't match the high. There's not really anything special about this though. We're simply seeing the natural differences between various meshes compared to a low poly. This is the same information in the OP just presented in a different manner. Now I'm not necessarily advocating for or against baking like this either, just answering the original question of how to remove the waviness they didn't want present. By making the high and low cylinders have equal segments, logically the only difference between mesh normals is the rounded corners that you may want to capture.
Objects are waviness_Low with respected maps applied.
It looks right to me but i'm unsure if those black corners and edges are the shader or the map? Also, Why is the shell set to 21? What are the benefits of having a larger shell and how does the bevel effect waviness and finally if i go to avenger it and bake again the waves reappear but if its got hard tops then they don't. Why is that?
You've got my full and undivided attention, Tell me your secrets.
Your bakes look like they have gaps from not using the cage, but it's hard to say if you're using different software. If you bake with the cage file your normal maps might look slightly different due to Maya's baker but the shaded result should be identical.
The waviness_Low file has split normals/smoothing groups on the top and bottom only to reduce the gradients in the normal map, but if it were set to all averaged normals like how the cage file is, there'd be no effective difference other than stronger gradients. It would look practically the same when applied to the model.
As for the beveled edge affecting waviness: a larger/smaller bevel might make it more or less apparent but it won't combat waviness which is only due to the number of sides on the cylinders.
As for this cage file and using it, I can't find any documentation on autodesk knowledge. It just doesn't exist for Maya.
Still at the end of the day, it comes back to this taper ends, a lot of my models use tapers for the ends and i'd like to know how to get them to bake correctly.
Doing Cylinders, Cubes and So on..
The shading errors you are seeing are most likely from not using a sync'd normal map workflow.
So does this mean adding supporting edges or chamfers or something? Is the problem in how I baked the normal map or in the model itself? It looks fine in blender. It's only when I take it out that it messes up.
I noticed that the normal map I get if I bake it in xNormal (right) looks wacked compared to the one I get from Blender itself (left).