Hey, sorry for the delayed reply. I took a video game break (basically me gaming for most of the day when I'm in a mess). Okay, here's the thing. When I think I have the workflow down, you guys introduce all sorts of other stuff to it which is quite overwhelming for me. Things like having a medium poly mesh or a 'triangulation before export' confused me. So what I ask is, could you list every thing I need to do in the process? I'm really confused on what meshes I need to have, what types of geometry they need to consist of and so and so... I don't want any surprises, it messes with me when I think I have the whole process down and then it turns out I'm doing a lot of things wrong; it just confuses me even more. I'll post a pic of all my meshes conforming together soon and I'll sort out the smoothing groups.
Blockout based on reference or blueprints, focus on silhouette
Build your Low Poly off of your block out and reference
After the low poly looks good, proportions are correct, looks acceptable, follows every aspect of the reference, add Supporting Edge Loops on Low Poly to serve as the pre-smooth mesh for the high poly (This is the Mid Poly of sorts)
Smooth your mesh and add any floaters or additional geo for bake purposes, this is the High Poly
Triangulate Low Poly and/or High Poly for export, duplicate before process for quad versions
Go through a final Final smoothing pass of your low poly, ensure smoothing looks good on both high poly and low poly. Ensure you
check for Ngons on your low poly, non manifold geo, errors in general.
Make sure Low poly conforms to the high poly, make sure it is UV'd
properly. Make sure the reference is represented accurately in 3D.
Project/create UVs for Low poly
Create Envelope from low poly
Bake the Normal, AO, etc
Texture
Export to engine
Apply/create material
Set up lighting
---
This is only if you are making the mesh in maya/max. If you are making something in zBrush you can make the base mesh in max first then export to zBrush then make the high poly and then create the low poly thereafter in the program of choice, topo gun, maya, max etc.
Every artist has slight variations, you will find yours.
Okay so to confirm, I should have 3 meshes? the game resolution mesh, the high resolution mesh for detail projection and then the mid poly mesh where I can alter edge loops any time I want? Correct? I am able to alter the mid poly until I get the result I want right? If that's not what the mid poly is used for could you explain what is?
I obviously will find my variations in time but yeah I need a proven base standard to work from, a foundation.
The mid poly, is essentially the low poly, with supporting edges to maintain a smooth flow for edges.
Four meshes really, counting the envelope, but yes.
You need the low poly for the game, for display. The mid poly is to determine and maintain edge flow, and smooth edges for the high poly, to create the high poly. The high poly to bake detail, the detail onto the low poly. The envelope for baking capture.
Millenia talks about it in that Ithaca shotgun video. Not sure if he calls it a mid poly.
You make the low poly. Once decided it is finished you begin to add edges near the border edges to maintain the overall look and proportions of the mesh when you smooth the mesh to get the high poly. The mid poly is just the low poly with additional edges to support the border edges.
Ahhh okay okay. So once I've created that, all I need to do is actually bake the geometry from the turbosmooth modifier preview onto the actual mesh to make the high poly. I can always go back and make edits to the mid poly if I want to change the form of my high poly. Correct?
Hey man, sorry I've been away for quite some time. Been busy and distracted with a lot of things. But I'm back and refreshed and motivated to carry on if you're willing to help again. I'll post a pic of my meshes conforming together tomorrow so we can hopefully figure out why I'm getting a blank normal canvas. I really want to get baking down properly and correctly. I want to get the basics of it right and correct.
Okay, so here is my HP, LP conforming together. My cage is at the side. Hope this helps. Oh and I know I need to sort out my smoothing groups for the meshes, But I should atleast get some normal detail on my normal map, I don't think it's linked to that. But what do I know huh?
Okay I need some help with UV'ing. I've created a small piece of geometry just for a little UV study. See, I've split everything up and projected the UV's in the right axis so now there is no distortion what so ever. Everything is perfect. However the thing is, I don't know how to put it together... how I should put it together.
I remember you talking about something like, hard edges should always be split up and soft edges can stay connected. What is that?
In conclusion, what is the best and the most optimal way for me to assemble my UV's in the 0-1 space so that they can be ready for export?
Here are some pics: Notice how my UV's are split up? I need them assembled but I don't know the best and the most optimal way to do them.
Go ahead and try your best at fitting the UVs within the 0to1 space. Maintain proportions when doing so however and try to scale down evenly.
If you have a hard edge, that edge must be separated in the UV
map. That hard edge needs to be separated to bake properly. When it is a
soft edge it needs to be sewn in the UVs. Keep in mind if the edge is
hard separate in the uvs, but if the uv is separated it does not NEED to
be hard.
Okay. So soft edges can be split up too? I also need to think about what somebody will see most in the model right? For example in this one, they would see the outside 4 polygons and the inside 4 polygons. The rims of the model will be seen the least.
Also, previously you listed this as in the steps I need to take when making a mesh asset:
Build and hunt for Reference
Blockout based on reference or blueprints, focus on silhouette
Build your Low Poly off of your block out and reference
After
the low poly looks good, proportions are correct, looks acceptable,
follows every aspect of the reference, add Supporting Edge Loops on Low
Poly to serve as the pre-smooth mesh for the high poly (This is the Mid
Poly of sorts)
Smooth your mesh and add any floaters or additional geo for bake purposes, this is the High Poly
Triangulate Low Poly and/or High Poly for export, duplicate before process for quad versions
Go
through a final Final smoothing pass of your low poly, ensure smoothing
looks good on both high poly and low poly. Ensure you
check for Ngons on your low poly, non manifold geo, errors in general.
Make sure Low poly conforms to the high poly, make sure it is UV'd
properly. Make sure the reference is represented accurately in 3D.
Project/create UVs for Low poly
Create Envelope from low poly
Bake the Normal, AO, etc
Texture
Export to engine
Apply/create material
Set up lighting
I assume this is still valid? Makes a little bit more sense to me now.
Hey is there anything I need to know about overlapping UV's? Here's what I got so far: I took a shot of the actual model with the checker texture to show you that I thought about proportion. The outside polygons have more resolution than the anything else since that's what the eye will see the most. The inside is second priority, and then the rest is just the rims and side parts.. you know.
I kind of know what I'm doing and I kind of don't. I'm definitely not confident when I'm doing it though. I guess that will come in time the more I do this.
However, you need to do better. Overlapping UVs are fine, but I would not bother doing so with this asset as it is for practice. Additionally overlapping UVs will complicate things for you when you bake and take this into the texturing phase, at least it can. The rims/smaller surfaces should stay sewn together. The edges are soft (or should be) and therefore especially in this case should be sewn. The other reason being is that they are not 90 degrees or a similar angle. If the angle is close to that of a flat surface, generally speaking it should be soft and sewn. The bottom most sides should be hard + separated. The only instances I can see that should be indeed sewn and intact is the Top rim and that lower rim.
Hopefully that makes sense, give it another go. Keep the UVs tighter, rotate them to fit and use more space!
Answering your questions, yes soft edges can be split, but I would say rarely, only when absolutely needed. You would indeed need to consider what part of the mesh will be seen the most that should/can dictate what UV hull is what size. That bulleted list is indeed still valid. I would have you avoid overlapping UVs for now, again it can lead to issues.
Ahh yes! I remember this! It's all coming back man Okay got it. Hold on, could I have my UV's diagonal? But then my textures would come out diagonal... but then I could just rotate them right? Lol. I will also need to take in fact that if I texture purely using SP then SP will take care of it for me using the trilinear option. See because I've not done any proper texturing yet, I don't have a feel for what I have to prepare for.
I will give it another go. Will post results. Oh and I'll make sure to avoid overlapping UV's for now.
Okay! Progress. I have thought about proportion, although there may be errors in that department. I have thought about what needs to sewn together and what doesn't. No overlapping UV's. All clean. No doubt my UV's could be better but I'm getting there. On the top picture, the square UV's that are highlighted in the top left are the only UV's sewn together. I get that part 100% now. Although this is a simple model, I'm sure I'll get the hang of it on more complex models too.
The UVs you sewed are now incorrectly representing the 3D version in the 2D UVs. You have them square but they are in 3D Rectangular, retain that shape. You still have a great amount of empty/negative space, utilize that! Your UVs are extremely tight in some instances, give them some space, remember the textures will bleed on to each other or share pixels. Stay away from diagonal rotations, its just easier if they are rotated by 90, unless you have plenty of UVs where you need to optimize the space.
UVs must retain a representation of the 3D shape, yes. There are times where you can get away with skewing the shape, a bit of scaling, but try to avoid it when possible.
Okay. I know my UV's aren't 100% packed but that's the best I can do. Proportion is thought about. 2D UV's retain the same form as the 3D mesh. All hard edges are split. No diagonal UV's. No overlapping UV's.
Would be helpful if you told me how so. Lol. Or should I move onto baking?
Okay check this out. I'm really aching to learn this stuff. So I started another little UV study. Applying those same rules onto this little piece of geometry. It's supposed to represent an anvil. I'm suprised I even modelled that without reference. I thought about all the stuff you taught me. Proportion, Hard and soft edges.. everything. Is there anything else I should know? But yeah I feel so much more confident in UV mapping now. Don't feel scared at all. Would you say my UV'ing ability is pratical? Would they make my textures look decent? Given appropriate resolution of course.
HEEYYYYY. Look! This is the crate without the normal map: There are tiny artifacts in the cornor. I acknowledged this when doing a test render in 3ds max before actually going into XNormal. It's just so that I know my cage covers my model good and no rays are missed. A tiny amount of rays are missed. No biggie. Here are both my maps. The padding may be a lot, it's set to 32 in XN. Also, is my AO supposed to look like that? Getting there man.
Hey. Also, I think I've isolated my baking problem. I get confused and completely lose it when I'm dealing with objects that are made up of multiple pieces. I'm going to try and keep trying to make sure I do everything in an organised manner. Also going to use the layer system and get better at is. Putting objects on new layers makes it much easier to work with. As soon as I tackle that, I think I can confidently say I've got the hang of basic baking. Unless there's anything else I need to know in terms of basics. You're the boss.
So, this is what I would do for the UVs, I have scaled the uvs slightly in one direction or another to maximize space:
--- Also, always model with reference!
Your current UV-ing is practical yes, but still needs to be improved to fully utilize the 0 to 1 space.
There is more to learn with UVs, like tiling UVs, using the second and third UV sets for other aspects, having UVs share the same space (overlapping/mirrored UVs), and advanced models with various smoothing groups.
Those bake errors need to be resolved. Whats your cage look like?
Hey. What do you mean you have scaled the uv's in one way or another to maximize uv space? Don't get you there sorry. Yeah I get that my UV's are not as good as they can be. Also, I know to model using reference. I just try to make small pieces of geometry, very basic, in the use of practicing UV'ing. They aren't worth much to me, I can scrap them when I'm done UV'ing them.
I'm getting more used to fixing missed ray errors. I go in, fix them, things actually work for me now. :pleased: Let me see if I can't fix the cage errors my self. I'll show you my cage if it doesn't work out.
As for the anvil. Here are my baking results: RED=CAGE BLUE=HIGHPOLY GREY=LOWPOLY Anvil didn't turn out too great. But again, it's really a matter of me going in and fixing those missed rays which I'll get to work on. Heres my AO and NM rendered in XN: What can I do to make my AO better? I get those funny triangular shading things faintly appear in the maps.
Hey, there's also something else I need to talk about. I was reading a thread and if you go down to look at "Quack!"'s first post, he states:
"In a modern work flow we are depending less and less on high poly unique
baked assets and more on high tri count low polys with material blending
and custom normals. In this case I wouldn't bake a unique normal map
and instead use a tiling texture and some edge bevels with custom
normals. Most assets like this, even in the past, would not be baked
down from a high poly, major time waster."
See, I've been skeptical about this too. When I video game, I never really see anything like normal maps be put into place for basic general assets. The assets look really, really nice, but they have sharp edges. Which doesn't really matter. Here, I took a screen of an asset or two in Farcry 3 for example: Sorry if the pics are large. Don't know why. >_^ 99% of the time I never really notice normal maps in assets. In the last picture, the crate you can see next to the table defo has a normal map. This is because that crate is scattered around the world and it's supposed to be seen and used by the player. It's an important asset. I can see the round curvature of the mesh and the map. But generally I would say 80% of assets in the scene have no NM.
I've been doing some research lately and I've made a list of the things I want to be fully confident in:
Texel Density
Custom Normals
Material Blending
Tiling Textures
And having multiple assets on a single UV map.
The things I'm most unsure about is all of those except for Texel Density. I've been hearing 'Texel Density' since the beggining of time. Which is btw when I started 3D. :P Don't get me wrong though, I barely know anything about Texel Density.
So what is with that? And why haven't you told me that it's in-efficient to produce a NM for every asset? I always thought that was how it was done.
I slightly scaled them either in the U or V direction to fill up space.
Regarding your AO bake errors:
Make sure the Cage encompasses the entire HP and LP
Make sure you have "Closest hit if ray fails" unchecked within baking options and "Ignore backface hits" unchecked as well within AO settings plus uncheck "limit ray distance"
Make sure your High Poly geo is not needlessly Smoothed or tessellated
I see smoothing errors in you HP that may be causing the AO issues as well
Try the above see if any of that resolves your errors.
So, regarding the High Poly to Low Poly workflow and what Quack is talking about, that is for a professional large scale workflow. What you are doing is learning the basics and learning 3d in general. Simply making a Low Poly asset and adding custom normals will leave you lacking in the HP to LP workflow. Everything Quack said was correct, but he knows how to make a high poly and bake that detail down. His team knows how to do that, and they know how to increase their efficiency because they know the basics of baking down detail. I believe learning the HP to LP workflow is important, and it is certainly important to acquiring a job. Check a position's job description, you will find it stating you need knowledge in HP to LP baking, sculpting, etc. It is still required knowledge. Perhaps one day we can get away with simply creating a one level of detail object but that time is not now especially considering most studios are creating games for various systems with various levels of performance. There are assets that do not need a normal map because they are so unimportant however you should not concern yourself with that at the moment. I would say after you understand the process better and nail the steps for baking then you can worry about creating something like that. Tiling textures will get you into this, where you create one level of detail object without an HP and only use tillable textures. A perfect example of this are modular buildings, a HP is often not needed.
Alright. Understood. I'll try those solutions. So the highpoly needs to have appropriate smoothing groups too?
So what you're essentially saying is: Learn the methods and nail them down perfectly. Then I can choose what I want to use in what scenario. For example I could use custom normals in huge scenes where I need to batch produce loads of assets. On the other hand, I would be better off using normal maps for scenes that aren't as big. Hypothetically speaking.
Also, I don't know why, I just can't visualise tiling textures. When the time comes, you gotta show me an example or what they are. I'll probably also look it up my self. The point is.. I can't visualise it. So explanations are useless in those terms.
Also, after baking, is it on to texturing? I know for now, I've got baking to conquor.
Your highpoly has smoothing errors, that does not necessarily mean smoothing groups but instead geo could be pinching or super tight which causes smoothing errors on your mesh and in turn on your LP when you bake. Smoothing groups are mainly for the LP so it bakes properly.
Hey, letting ya know, turns out I won't be near my computer just for today. So progress is halted.
Anyway, really really nice interesting read on the thread you sent me. Understood half of it, didn't understand the other half. I like how EQ is the Dr.House of Polycount.
Yeah so which one should I use? Averaged Projection Mesh or Explicit Mesh Normals? Or should I use both depending on what situation I'm confronted with? Could you explain to me the method I'm not currently using? Currently, I think all my seams are hard edges. I think. Whats the other method? I believe I'm using Averaged Projection Mesh, right?
Another thing I need to clarify. Do I need to use the Ray Distance Calculator at all in XNormal if I'm using a cage in the first place? A user asked this and here's EQ's reply: "To get an "averged projection mesh" in Xnormal, you will be using a
cage, either exported from your 3d app(In max you can set up a
projection modifier and then export as SBM with your cage) or by
creating a cage in Xnormal's 3d viewer. So you don't need to use those
settings, as the cage overides it."
The cage overrides the RDC's values anyway. So confirmed, I should stop bothering with it?
I also 100% agree with this guys post. Most importantly, this part:
"The point i want to make is this information was horribly
communicated/documented by epic/unreal in the first place. you guys :MM,
polygoo, earthquake, are at the Top of of the chain in terms of game
art and understanding, and look at this muddy discussion. where does
that leave the new guys, and less technical 3d artists?"
Either method is really up to you. I prefer the cage method, and I would venture to say it is the most used as well. To better understand, read EQ's thread on just that:
Alright. I've created a little bench using reference. Posted pics of the lowpoly and the highpoly. Tris can be seen in the top left (selected). The handrests and the seat it self are different objects. Meaning, the hand rests are one, seat is another. UV mapped it. I'm still going to do some finishing touches. I know that the hand rests currently have more resolution than the actual seat itself which is incorrect.
Will post updates shortly. Crit me. Reference:
LP:
HP:
UV + Checker Proportions:
UV's:
It's red because I selected them. Easier for viewing. Thanks.
Update: Optimised the uv's by a good 30%. Proportions aren't looking too bad now. Kept in mind that the back faces of the bench and the faces under the seat are obviously going to be low res since they're barely going to be seen. Back faces will be seen more so I've not scaled it down too small. In the top left I've got a few faces that aren't really going to be seen at all. A mistake I made which was not deleting them. Will keep it in mind for my next mesh. Not worth it going back and deleting them.
Moving to baking. See where I'm at with making optimal cages.
Okay so this is what I wanted to talk to you about. I can't proceed without you instructing me on this. Exploding for bakes. I want to know how to do it fresh. I barely remember anything. The only thing I do remember is, it was horrible for me. So how do I explode this mesh? Consisting of 2 pieces. Would I select them both and add a projection modifer together? Or would I need to select the seat and the hand rests individually and make a cage for them like that? (So I would have 2 seperate cages).
Do they need to be moved? So maybe translate the seat in the Z axis? Like this:
Duplicate your LP and HP and keep the initial LP/HP hidden to have unexploded
meshes. Have your other Low poly and High poly meshes fully
exploded, so select the armrests and pull them away from the backrest. After that is done I like to triangulate the HP and LP. Then create a cage from your LP. You should now have three meshes. Export for baking.
What you have looks fine for baking, just make sure you have a proper smoothing pass for your LP.
You have no idea how good I feel after reading 'UVs look much better'. Oh the satisfaction. Feels good to have improved in that section. I remember when I was completely clueless in that subject. Yeah alright, working on the smoothing. I'll post updates. Check back often now, won't be stalling this no longer. :P
Question: Does my highpoly HAVE to be triangulated? Or can I just have it where I only need to triangulate my cage and lp. It's just that triangulation can give minor artifacts when using turbosmooth. Confused. All my life I've been told you should not have triangles when using turbosmooth yet here you are telling me it needs to be triangulated.. Or maybe you didn't know since you're using Maya and smoothing works differently?
Another question. See, I still have my hand rest and seat as different objects. Would I need to attach them to each other and create a uniform projection? Or would I keep them seperate?
Edit: Okay so I've attached both the handrest and the seat to each other for both the cage and the lowpoly. The highpoly, lowpoly and the cage are all triangulated. The lowpoly and the cage have exactly the same smoothing groups. In XNormal I have added in the HP the LP and I made sure to click 'Browse external cage file' to add in the cage. Now to see if the cage to lp relationship would work I went on the ray distance calculator and ran it but I get this message: No idea what to do. Tried everything I know.
If you get artifacts from triangulation on your HP, then no you do not need to do it. I just have a habit of triangulating my HP and LP, neither is truly needed it just reduces error messages in xNormal.
You should combine all your meshes into one mesh. So you have the HP and LP then create the cage from your LP and you will have 3 meshes for export.
Regarding the error, make sure the cage is exactly the same as the LP. If you make an edit to the LP, you need a new cage. Some modifiers in max can impact the topology of the cage. Also ensure you have the HP, LP, and Cage in the same group or not even in a group. xNormal does throw odd errors sometimes, often the answer to the error is not easy to find. Keep at it, export your HP, LP, and Cage and try again.
Alright. So I did things very slowly. And a bit of progress. I swear I'm not dumb, but when it comes to layers and groups and working with multiple meshes and different prefixes... my mind literally does this: Anyway, rendered out the normals and ao. Got a few errors here and there. Nothing adjusting the cage won't fix. Getting better and better day by day but I just feel progress is incredibly slow. Not as fast as I want it to be.
Here are the ao and normals btw: Where I've outlined the black. No need to fix those. They won't be seen and are faces that should have been deleted by me. Hence the reason they're messed up anyway. AO isn't too bad. Not getting any weird triangle shadows.
I've applied the normal map but the thing is... well, I've noticed this more than once with multiple of my models: The normal map isn't significant. I mean look at this: I can see that the normal is there and it has some effect but it's so small. I think I'm missing something. How do I make my normal maps more significant? I know at the end of the day they are texture maps but I'm sure I can reap more gold from them. If this is the result then it makes doing this almost pointless and in-efficient.
Replies
I don't want any surprises, it messes with me when I think I have the whole process down and then it turns out I'm doing a lot of things wrong; it just confuses me even more. I'll post a pic of all my meshes conforming together soon and I'll sort out the smoothing groups.
---
This is only if you are making the mesh in maya/max. If you are making something in zBrush you can make the base mesh in max first then export to zBrush then make the high poly and then create the low poly thereafter in the program of choice, topo gun, maya, max etc.
Every artist has slight variations, you will find yours.
---
Let me know if that makes sense.
I obviously will find my variations in time but yeah I need a proven base standard to work from, a foundation.
Four meshes really, counting the envelope, but yes.
You need the low poly for the game, for display.
The mid poly is to determine and maintain edge flow, and smooth edges for the high poly, to create the high poly.
The high poly to bake detail, the detail onto the low poly.
The envelope for baking capture.
You make the low poly. Once decided it is finished you begin to add edges near the border edges to maintain the overall look and proportions of the mesh when you smooth the mesh to get the high poly. The mid poly is just the low poly with additional edges to support the border edges.
I'll post a pic of my meshes conforming together tomorrow so we can hopefully figure out why I'm getting a blank normal canvas. I really want to get baking down properly and correctly. I want to get the basics of it right and correct.
Perhaps you do not have correct inputs for xnormal? Are all UVs in the 0 to 1 space?
You can send them to me... not sure honestly.
I remember you talking about something like, hard edges should always be split up and soft edges can stay connected. What is that?
In conclusion, what is the best and the most optimal way for me to assemble my UV's in the 0-1 space so that they can be ready for export?
Here are some pics:
Notice how my UV's are split up? I need them assembled but I don't know the best and the most optimal way to do them.
If you have a hard edge, that edge must be separated in the UV map. That hard edge needs to be separated to bake properly. When it is a soft edge it needs to be sewn in the UVs. Keep in mind if the edge is hard separate in the uvs, but if the uv is separated it does not NEED to be hard.
I also need to think about what somebody will see most in the model right? For example in this one, they would see the outside 4 polygons and the inside 4 polygons. The rims of the model will be seen the least.
Also, previously you listed this as in the steps I need to take when making a mesh asset:
- Build and hunt for Reference
- Blockout based on reference or blueprints, focus on silhouette
- Build your Low Poly off of your block out and reference
- After
the low poly looks good, proportions are correct, looks acceptable,
follows every aspect of the reference, add Supporting Edge Loops on Low
Poly to serve as the pre-smooth mesh for the high poly (This is the Mid
Poly of sorts)
- Smooth your mesh and add any floaters or additional geo for bake purposes, this is the High Poly
- Triangulate Low Poly and/or High Poly for export, duplicate before process for quad versions
- Go
through a final Final smoothing pass of your low poly, ensure smoothing
looks good on both high poly and low poly. Ensure you
check for Ngons on your low poly, non manifold geo, errors in general.
Make sure Low poly conforms to the high poly, make sure it is UV'd
properly. Make sure the reference is represented accurately in 3D.
- Project/create UVs for Low poly
- Create Envelope from low poly
- Bake the Normal, AO, etc
- Texture
- Export to engine
- Apply/create material
- Set up lighting
I assume this is still valid? Makes a little bit more sense to me now.Here's what I got so far:
I took a shot of the actual model with the checker texture to show you that I thought about proportion. The outside polygons have more resolution than the anything else since that's what the eye will see the most. The inside is second priority, and then the rest is just the rims and side parts.. you know.
I kind of know what I'm doing and I kind of don't. I'm definitely not confident when I'm doing it though. I guess that will come in time the more I do this.
However, you need to do better. Overlapping UVs are fine, but I would not bother doing so with this asset as it is for practice. Additionally overlapping UVs will complicate things for you when you bake and take this into the texturing phase, at least it can. The rims/smaller surfaces should stay sewn together. The edges are soft (or should be) and therefore especially in this case should be sewn. The other reason being is that they are not 90 degrees or a similar angle. If the angle is close to that of a flat surface, generally speaking it should be soft and sewn. The bottom most sides should be hard + separated. The only instances I can see that should be indeed sewn and intact is the Top rim and that lower rim.
Hopefully that makes sense, give it another go. Keep the UVs tighter, rotate them to fit and use more space!
Answering your questions, yes soft edges can be split, but I would say rarely, only when absolutely needed. You would indeed need to consider what part of the mesh will be seen the most that should/can dictate what UV hull is what size. That bulleted list is indeed still valid. I would have you avoid overlapping UVs for now, again it can lead to issues.
Okay got it. Hold on, could I have my UV's diagonal? But then my textures would come out diagonal... but then I could just rotate them right? Lol.
I will also need to take in fact that if I texture purely using SP then SP will take care of it for me using the trilinear option. See because I've not done any proper texturing yet, I don't have a feel for what I have to prepare for.
I will give it another go. Will post results. Oh and I'll make sure to avoid overlapping UV's for now.
I have thought about what needs to sewn together and what doesn't.
No overlapping UV's. All clean.
No doubt my UV's could be better but I'm getting there.
On the top picture, the square UV's that are highlighted in the top left are the only UV's sewn together. I get that part 100% now. Although this is a simple model, I'm sure I'll get the hang of it on more complex models too.
The UVs you sewed are now incorrectly representing the 3D version in the 2D UVs. You have them square but they are in 3D Rectangular, retain that shape. You still have a great amount of empty/negative space, utilize that! Your UVs are extremely tight in some instances, give them some space, remember the textures will bleed on to each other or share pixels. Stay away from diagonal rotations, its just easier if they are rotated by 90, unless you have plenty of UVs where you need to optimize the space.
Try once more.
Extremely tight UV's are a no go.
No diagonal rotations for now.
Will post updates.
I know my UV's aren't 100% packed but that's the best I can do.
Proportion is thought about.
2D UV's retain the same form as the 3D mesh.
All hard edges are split.
No diagonal UV's.
No overlapping UV's.
How is it?
Or should I move onto baking?
Okay check this out. I'm really aching to learn this stuff. So I started another little UV study. Applying those same rules onto this little piece of geometry. It's supposed to represent an anvil. I'm suprised I even modelled that without reference.
I thought about all the stuff you taught me. Proportion, Hard and soft edges.. everything.
Is there anything else I should know?
But yeah I feel so much more confident in UV mapping now. Don't feel scared at all.
Would you say my UV'ing ability is pratical? Would they make my textures look decent? Given appropriate resolution of course.
This is the crate without the normal map:
There are tiny artifacts in the cornor. I acknowledged this when doing a test render in 3ds max before actually going into XNormal. It's just so that I know my cage covers my model good and no rays are missed. A tiny amount of rays are missed. No biggie. Here are both my maps. The padding may be a lot, it's set to 32 in XN.
Also, is my AO supposed to look like that?
Getting there man.
---
Also, always model with reference!
Your current UV-ing is practical yes, but still needs to be improved to fully utilize the 0 to 1 space.
There is more to learn with UVs, like tiling UVs, using the second and third UV sets for other aspects, having UVs share the same space (overlapping/mirrored UVs), and advanced models with various smoothing groups.
Those bake errors need to be resolved. Whats your cage look like?
Yeah I get that my UV's are not as good as they can be.
Also, I know to model using reference. I just try to make small pieces of geometry, very basic, in the use of practicing UV'ing. They aren't worth much to me, I can scrap them when I'm done UV'ing them.
I'm getting more used to fixing missed ray errors. I go in, fix them, things actually work for me now. :pleased:
Let me see if I can't fix the cage errors my self. I'll show you my cage if it doesn't work out.
As for the anvil. Here are my baking results:
BLUE=HIGHPOLY
GREY=LOWPOLY
Anvil didn't turn out too great. But again, it's really a matter of me going in and fixing those missed rays which I'll get to work on.
Heres my AO and NM rendered in XN:
What can I do to make my AO better? I get those funny triangular shading things faintly appear in the maps.
"In a modern work flow we are depending less and less on high poly unique baked assets and more on high tri count low polys with material blending and custom normals. In this case I wouldn't bake a unique normal map and instead use a tiling texture and some edge bevels with custom normals. Most assets like this, even in the past, would not be baked down from a high poly, major time waster."
Thread: http://polycount.com/discussion/172863/environment-art-workflow-questions-texel-density-importance-when-to-bake-highpoly#latest
See, I've been skeptical about this too. When I video game, I never really see anything like normal maps be put into place for basic general assets. The assets look really, really nice, but they have sharp edges. Which doesn't really matter. Here, I took a screen of an asset or two in Farcry 3 for example:
Sorry if the pics are large. Don't know why. >_^
99% of the time I never really notice normal maps in assets. In the last picture, the crate you can see next to the table defo has a normal map. This is because that crate is scattered around the world and it's supposed to be seen and used by the player. It's an important asset. I can see the round curvature of the mesh and the map. But generally I would say 80% of assets in the scene have no NM.
I've been doing some research lately and I've made a list of the things I want to be fully confident in:
The things I'm most unsure about is all of those except for Texel Density. I've been hearing 'Texel Density' since the beggining of time. Which is btw when I started 3D. :P
Don't get me wrong though, I barely know anything about Texel Density.
So what is with that? And why haven't you told me that it's in-efficient to produce a NM for every asset? I always thought that was how it was done.
Regarding your AO bake errors:
Try the above see if any of that resolves your errors.
So, regarding the High Poly to Low Poly workflow and what Quack is talking about, that is for a professional large scale workflow. What you are doing is learning the basics and learning 3d in general. Simply making a Low Poly asset and adding custom normals will leave you lacking in the HP to LP workflow. Everything Quack said was correct, but he knows how to make a high poly and bake that detail down. His team knows how to do that, and they know how to increase their efficiency because they know the basics of baking down detail. I believe learning the HP to LP workflow is important, and it is certainly important to acquiring a job. Check a position's job description, you will find it stating you need knowledge in HP to LP baking, sculpting, etc. It is still required knowledge. Perhaps one day we can get away with simply creating a one level of detail object but that time is not now especially considering most studios are creating games for various systems with various levels of performance. There are assets that do not need a normal map because they are so unimportant however you should not concern yourself with that at the moment. I would say after you understand the process better and nail the steps for baking then you can worry about creating something like that. Tiling textures will get you into this, where you create one level of detail object without an HP and only use tillable textures. A perfect example of this are modular buildings, a HP is often not needed.
So what you're essentially saying is: Learn the methods and nail them down perfectly. Then I can choose what I want to use in what scenario. For example I could use custom normals in huge scenes where I need to batch produce loads of assets. On the other hand, I would be better off using normal maps for scenes that aren't as big. Hypothetically speaking.
Also, I don't know why, I just can't visualise tiling textures. When the time comes, you gotta show me an example or what they are. I'll probably also look it up my self. The point is.. I can't visualise it. So explanations are useless in those terms.
Also, after baking, is it on to texturing? I know for now, I've got baking to conquor.
Read this whole thread, alot of people asking questions some related to what you are doing:
http://polycount.com/discussion/107196/youre-making-me-hard-making-sense-of-hard-edges-uvs-normal-maps-and-vertex-counts/p1
Getting late but I'll get on fixing those errors tomorrow.
Anyway, really really nice interesting read on the thread you sent me. Understood half of it, didn't understand the other half. I like how EQ is the Dr.House of Polycount.
Yeah so which one should I use? Averaged Projection Mesh or Explicit Mesh Normals? Or should I use both depending on what situation I'm confronted with?
Could you explain to me the method I'm not currently using? Currently, I think all my seams are hard edges. I think. Whats the other method? I believe I'm using Averaged Projection Mesh, right?
Another thing I need to clarify. Do I need to use the Ray Distance Calculator at all in XNormal if I'm using a cage in the first place?
A user asked this and here's EQ's reply:
"To get an "averged projection mesh" in Xnormal, you will be using a cage, either exported from your 3d app(In max you can set up a projection modifier and then export as SBM with your cage) or by creating a cage in Xnormal's 3d viewer. So you don't need to use those settings, as the cage overides it."
The cage overrides the RDC's values anyway. So confirmed, I should stop bothering with it?
I also 100% agree with this guys post. Most importantly, this part:
"The point i want to make is this information was horribly communicated/documented by epic/unreal in the first place. you guys :MM, polygoo, earthquake, are at the Top of of the chain in terms of game art and understanding, and look at this muddy discussion. where does that leave the new guys, and less technical 3d artists?"
Will post soon. Thanks.
http://polycount.com/discussion/81154/understanding-averaged-normals-and-ray-projection-who-put-waviness-in-my-normal-map/p1
I would just say continue with the cage method, in the future you can try the other method.
Don't respond.
The handrests and the seat it self are different objects. Meaning, the hand rests are one, seat is another.
UV mapped it. I'm still going to do some finishing touches. I know that the hand rests currently have more resolution than the actual seat itself which is incorrect.
Will post updates shortly. Crit me.
Reference:
LP:
HP:
UV + Checker Proportions:
UV's:
It's red because I selected them. Easier for viewing. Thanks.
Optimised the uv's by a good 30%. Proportions aren't looking too bad now. Kept in mind that the back faces of the bench and the faces under the seat are obviously going to be low res since they're barely going to be seen. Back faces will be seen more so I've not scaled it down too small. In the top left I've got a few faces that aren't really going to be seen at all. A mistake I made which was not deleting them. Will keep it in mind for my next mesh. Not worth it going back and deleting them.
Moving to baking. See where I'm at with making optimal cages.
Okay so this is what I wanted to talk to you about. I can't proceed without you instructing me on this. Exploding for bakes. I want to know how to do it fresh. I barely remember anything. The only thing I do remember is, it was horrible for me. So how do I explode this mesh? Consisting of 2 pieces.
Would I select them both and add a projection modifer together? Or would I need to select the seat and the hand rests individually and make a cage for them like that? (So I would have 2 seperate cages).
Do they need to be moved? So maybe translate the seat in the Z axis? Like this:
Duplicate your LP and HP and keep the initial LP/HP hidden to have unexploded meshes. Have your other Low poly and High poly meshes fully exploded, so select the armrests and pull them away from the backrest. After that is done I like to triangulate the HP and LP. Then create a cage from your LP. You should now have three meshes. Export for baking.
What you have looks fine for baking, just make sure you have a proper smoothing pass for your LP.
Yeah alright, working on the smoothing. I'll post updates. Check back often now, won't be stalling this no longer. :P
I'll try both methods.
Okay so I've attached both the handrest and the seat to each other for both the cage and the lowpoly. The highpoly, lowpoly and the cage are all triangulated. The lowpoly and the cage have exactly the same smoothing groups. In XNormal I have added in the HP the LP and I made sure to click 'Browse external cage file' to add in the cage. Now to see if the cage to lp relationship would work I went on the ray distance calculator and ran it but I get this message:
No idea what to do. Tried everything I know.
You should combine all your meshes into one mesh. So you have the HP and LP then create the cage from your LP and you will have 3 meshes for export.
Regarding the error, make sure the cage is exactly the same as the LP. If you make an edit to the LP, you need a new cage. Some modifiers in max can impact the topology of the cage. Also ensure you have the HP, LP, and Cage in the same group or not even in a group. xNormal does throw odd errors sometimes, often the answer to the error is not easy to find. Keep at it, export your HP, LP, and Cage and try again.
Anyway, rendered out the normals and ao. Got a few errors here and there. Nothing adjusting the cage won't fix.
Getting better and better day by day but I just feel progress is incredibly slow. Not as fast as I want it to be.
Here are the ao and normals btw:
Where I've outlined the black. No need to fix those. They won't be seen and are faces that should have been deleted by me. Hence the reason they're messed up anyway.
AO isn't too bad. Not getting any weird triangle shadows.
I've applied the normal map but the thing is... well, I've noticed this more than once with multiple of my models: The normal map isn't significant. I mean look at this:
I can see that the normal is there and it has some effect but it's so small. I think I'm missing something. How do I make my normal maps more significant?
I know at the end of the day they are texture maps but I'm sure I can reap more gold from them. If this is the result then it makes doing this almost pointless and in-efficient.