What I did is that I separated the base mesh to different parts so when I use it in Zbrush I can split it to subtools and as TeriyakiStyle suggested save myself a lot of headache (in sculpting and polypainting).
So yeah the vertices are not connected.
Here, each color indicates a separate mesh
You don't need to separate everything. TeriyakiStyle only sent 3 meshes to ZBrush I think, which were the blade and both straps. Then he separated the 3 meshes to separate subtools and then he split the blade into different polygroups using group by normals. I'd suggest you to try to export something like this:
You don't need to separate everything. TeriyakiStyle only sent 3 meshes to ZBrush I think, which were the blade and both straps. Then he separated the 3 meshes to separate subtools and then he split the blade into different polygroups using group by normals. I'd suggest you to try to export something like this:g[/IMG]
How did he do that exactly? I get these weird groups when I grouped by normal.
On another note, shouldn't this thread be stickied like the regular competitions?
Try to adjust the angle on MaxAngle. It can be a pain sometimes. If you can't get it to work perfectly, you can just try to group them manually, by selecting those artifacts, the part where they're supposed to be and pressing CTRL+W.
Sculpted the blade wrap as I need practice in that dept, total noob there. I think the result is OK, just mediocre. I'm happy with it as my first cloth wrap sculpt, but i wouldn't use it for a bake I don't think.
I could probably do a much better job just modelling and sub-ding it, a lot quicker too.
I guess that's one thing I need to learn is when to sculpt and when it doesn't need it.
That's a great start! I'd only advise you to be careful with some of the geometry. I see that you don't have anything under the straps near the blade, which is pretty obvious where the straps meet the metal, or rather don't meet it, because when you smoothed it the mesh got "smaller" and you can see holes there. Same thing on the handle, when smoothed you can tell that it's not connected to the blade, so I'd recommend you to try and merge these two and extend the mesh on the blade. Also, there's some weird smoothing going on at the circles, can I see the base mesh?
Here's a quick paintover with some of the problems:
Hey sorry I missed this. Can you show me an image of what results you are getting? For me I am starting out with my low poly then I set the angle to around 10 (degrees) and it splits the planes well for me.
Hi guys, got my LOD0, LOD1, bakes & diffuse done. Time to move onto the masks & get this baby in game! Will finish this tomorrow evening.
I'm surprised I got a decent bake onto the LOD1, there are some weird parts but I figure being LOD1, people would not notice such details whilst playing. Any advice on making LOD1 though would be great, I worked from LOD0 & removed edges & moved verts to keep the same silhouette.n The automatic reduction tool was just too messy.
@quockhanhlk - it looks like you're going in the right direction, but be careful not to add too much geometry when it's not needed. You currently have a whole lot of subdivides on the handle, for example. If geometry isn't defining silhouette, it probably means you don't need it! Of course, those can later be used to create the handle strap, depending on your own work flow.
@FlashL Your UV packing looks good! I typically make my LOD1s from the LOD0 by almost exclusively removing edges. Modo's reduction tool is too messy yeah, I'd only use it if I was desperate or lazy that day. Or maybe if only both.
@Snowstorm Thanks! I've been working really hard on improving my UV layouts, there's still room for improvement but compared to 12 months ago I am light years ahead.
As for the texture, I coloured my high poly with basic coloured shaders & baked it down to my LOD0, this provided my base colours to hand paint into using CS6. I used my AO bake on a multiply layer above the base colours to create depth, & then painted in some extra detail. I also used a metal texture overlay from CGtextures for the scratches on the metal.
Since I decided to finally get into some modelling I find this a great opportunity to start out with my very first ever finished HP model and my first ever attempt at sculpting.. I work with Blender for maybe 2 weeks or so and this is my first output:
edit: I'll probably start all over again because it was pretty much a tryout to see where my current modeling "skills" will get me to..
Nice start! I have a few suggestions though. The blade seems REALLY thick, I'd recommend you to maybe go for half of that or even less. The shape looks spot on though. Be careful with those cuts and stuff, right now they look too detailed, something you would probably get from rocks, not steel. Try to check out some of Motenai's videos on YouTube and check how he does those cuts. Apart from that, some extra polish on the straps could really help.
It's a great start! I'm glad that you're joining the fun
@rock as belkun said make it thiner ,it's way to thick .If you want to make something that will look realistic i sugest you using the orb cracks brush,it's making a very nice shape combined with pinch at the endis to make it sharp.
also you can try some custom alpha with the dragrect stroke.
after all is a good start
Updates on the noob competition
My lowpoly baking mesh with UVs
Are the places where I split UVs fine? I always wondered the extent of splitting the UVs.
The method I used was that I used the base mesh (before sculpting) as a low poly base, then I tweaked the topology on how I changed the sculpted topology.
@Ghost,
yeah it looks good. No reason to split the top of the blade imo. The distortion you'd get on the little tri up top won't be noticeable anyway.
You're only creating 2 more verts but it's 2 more than needed. Not really an issue but something to keep in mind.
(It's hard to give feedback here because everyone is doing pretty good anyway)
It does appear that you still have tris under the blade wrap, deleting those will give you more for rounding out the blade a bit)
@Ghost Yeah, I'd revisit your UVs. Minimal islands is going to cause less headaches down the road and a better texture/result. If you don't have to split them I wouldn't.
I didn't see much else he could do with the uv's though, I probably would have done similar.
The only thing he really split was handle from blade (different material anyway, good place to split), and the blade in half... Less distortion than bending it around a corner. (Square map/long rectangle blade). He floated the blad geo so that's already split.
I dunno, I guess the blade wrap's verts could be combined to those of the blade once those extra polys are gone.
That would save a few faces, a few verts and make 2 islands one.
----
If you do go that route ghost I would leave the circle uv's intact and put the distortion in the wrap. The circle is much more noticeable if it's distorted.
It doesn't make sense why the handle is taking as much space as the blade. You won't see the blade in game, so why is it so big?
At the very least, it would be worth it to have the blade be a single island in the same way the fuller/back part of blade is. It will be much easier to get precise texture then having to combat with un-natural seams.
@heboltz
I didnt scale the hilt it was just like that when I unwraped and I just jigsawed the uvs.
So here's my point. With minor uniform scaling, you can get the two pieces of the blade together in a diagonal that will keep you from getting a seam where your islands would meet. Heres a mock up.
The only thing I scaled was the hilt, I'm not exactly sure how the blue/green fit together, so I just over lapped them, while taking into account how they uvs "might" stack up. Hope this helps/explains what I mean.
If you want maximum uv space,and are willing to accept a little distortion (honestly I don't see it showing much in the blade anyway as it's mostly shiny steel) I'd go this route.
You need to get rid of those hidden tris anyway (That's 48 tris, can make the blade nice and round with that). If you can't see it it's wasted (it can be useful for baking, editing but not useful for in game).
I'm pretty anal retentive about my uv's
Though now I need to start worrying a little less about distortion since everything in Dota is baked from high (a lot easier to bake around distortion than hand paint around it)
So how should I scale the uv islands from each other.
The blue and green is one piece of the blade but I just split it at a corner area so I can maximize space.
I noticed that the yellow area is distorted a lot, what are the limits of the distortion? (I'm thinking not to have long stretched tris or quads)
The hidden area was just going to be the lod0 and I was going to delete the hidden areas after the lod1. I'm still kind of confused on how to create the lod1 from the lod0, I saw valve's highpoly/lowpoly models and they shorten the mesh.
Back to UV mapping, generally how should the proportions be? I see on heads the proportion is like 2 for the head and 1 for everything else.
Yeah, they tend to put more uv space on heads, it's an important part of the character and there are a lot of small details on a face. More animations and whatnot.
Generally it's best to have 100% no distortion and equal pixel density. But that's impossible without every poly on it's own island.
So I try to put most of it where it's less likely to be noticed. Like the handle, since in game it is under the hand.
That's why heboltz3 said to shrink that island, give yourself more space for the important stuff.
Of course the paintover I did was just pasting your bits together. You'd definitely want to relax the uv's some so the distortion is spread evenly across the entire blade. It'll be less noticeable then if just a few polys have all the distortion.
(Still, you won't get that much distortion over what you have
That blade is fairly even colors. If it had text on it the distortion would be harder to hide.
Do you use a checker map to check uv's? If you use the checkerboard pattern (max has it in uv editor, I suppose most programs do now)) you can see in the view how even those checkers are in shape and size. Best way to get proper scale.
======
Still, on high or low poly (not talking high res bake obj) you don't want those polys that are hidden. They do no good. Wasted. And in game they are bad because they are still drawn, but they are overdraw.
The engine has to draw them in front of terrain/characters, etc..., then it has to draw the cloth on top of that too. One model might not make a big difference but if every model in game was like that it would stack up quick.
Is this better? I merged the green and blue UV spaces. There is little to no manual distortion on this UV map and all I did was just scaled down the hilt. I still don't really understand distortion stuff.
definitely better. You could probably still squeeze a few more pixels out of it but it depends on how picky you want to be. If you lay it out like I showed you will get more better pixel density (more pixels to color it with, and not much distortion - stretched pixels).
But splitting the circle and wrap apart work too.
I've noticed Valve's Dota2 uv's really aren't very efficient. A lot of the items I have looked at have large areas of blank space.
Basically when uv mapping you are looking at 3 things:
1:space usage
2:distortion
3:seams.
Space usage needs to be optimized by arranging the uv's so they pack tightest. No wasted space. this image shows how much space you wasted compared to mine (approximately). You wasted a 3rd of your space, I wasted maybe a 5th. I get more pixels and detail. (for that image I just filled in wasted spaces with black, then compressed those 'islands' into one area as a rough example)
Distortions, it's always more obvious on things that are seen easily and things that have crisp details. Text would be easier to see distortion than on your gradient painted blade. The gradient is already kind of like blurred pixels, so distortion ddoesn't show well.
Some text would be hard to read and blurry if stretched a lot.
Seams: Seams are hard to paint out, so obvious in game. They also split the verts on your mesh, making more verts, making your model more complex to render. Best to split on hard edges, between materials (like blade steel and leather handle) and in places like armpits that the player won't see much.
You just have to find the best balance. Use as much space, hide seams and distortions where they won't be noticed and try to use as few islands as possible so you're not creating extra verts.
If you put this material on your model (sorry it was quick) you can see how even your uv's are. The squares will all be even sized and square if your uv's a really clean. If not you can adjust them until they look as good as possible.
@ghostdetector Don't be afraid to distort your UV's much like baddcog has shown. Because the dota2 workflow generally bakes from a painted highpoly, UV distortion (in most general cases) is automatically dealt with. The curving of the blade is almost essential for an effective UV layout for this asset. Here's mine:
I am still improving my UVing ability, I used to waste upwards of 45% of ym UV space, & now I'm averaging around approx. 25% ~ 20% depending on what I'm UVing of course.
On main topic, I've been really busy this week, but I plan on implementing my sword into game later tonight. What a fun challenge! Definitely an easier scope creating 1 great asset rather than creating a set. I'll leave sets alone for a while.
Here are more UV stuff. How did I do? Is there a way to check how much space I'm wasting in my uvs or do I have to paint spaces and jigsaw them together?
I don't know about other programs but in blender shows the distortion by a stretch feature. It color coded by gradients and it shows how warm the UV spaces are. Blue being the coldest (not distorted) and yellow being the warmest (high distortion).
You can see the worst distortion is on the 'blade spine'. It has a few very long squares in the middle. shorter ones by the wrap. If you adjust those verts by hand you can minimize that and even them out some.
The larger squares on your handle mean it is getting less pixel density, but that's alright as you've chosen that (since it will be seen less).
As far as wasted space, you just have to eye that. Use up space as good as possible and that's all you can do, but be aware of it.
I only did that as an example to illustrate it better. I'll probably never do it again
Yeah, that's the idea. You could still work that distortion more evenly through that island by straightening the far right edges. And lengthwise it could be evened out a bit so the squares are even sized the full length of it.
@rock - looking good! Some things to point out is how the handle connects to the ring part. It makes it seem as though the handle is separate from the actual blade. Another thing is how sharp top of the ring is (where the cloth and the blade meet.
Slow progress but I made the diffuse texture (color map) by just using the Hi-vertex as the base colors and adding the cavity/ao/normal map channels on top of it. I also added some scratches and grunge from cgtextures but I made it really transparent. Is it too sutle?
@TeriyakiStyle I have to thank you! I was still using Maya 2013 and your post inspired me to upgrade to 2014 and I'm so glad I did. The crease tool is so intuitive my only regret is that I didn't upgrade earlier.
@FlashL Very inspiring high poly model considering it looks like you've modeled it strictly within one modelling package.
@Baddcog Nice model! Practice makes perfect as they say. Your colours definitely catch the eye.
@rock Nice to see you taking constructive criticism and producing a great model afterwards.
Can somebody help me with rigging? With previous rigs/items, all I had to do was just move the mesh ontop of where the default item is. In this case, the sword is tilted.
In other zip files (the more recent ones i guess) the mesh is usually straight up and down to the left or the right of the hero mesh. For sven's case the weapon is sticking from its hand as well as it being tilted. Is there something wrong with the rig i got (i got it from the default) or do i just have to start rotating my mesh to fit the default mesh and rig it like that.
On another note, did anyone see the color/normal map for svens weapon. I only saw the mask1 and mask2 for the blade.
The rig should be right. Haven't tested it here, but I remember rigging some stuff that were similar to that (as in, not straight and in the center), such as Bloodseeker's. You just have to rotate it and move in order to match the orientation of the default sword.
I rigged it there is some clipping and I know its not done. But here are updates anyways. It looks a little bright to me though, Would changing the base colors be better than using a layer modifier such as levels?
I think I'm done. Now to do presentation promo stuff.
Question: It may be off topic but I was wondering about toned backgrounds for promos and concepts. Should the promo backgrounds just be simple and low saturated? Or should I create motion lines and/or gradients to lead the eye?
Replies
So yeah the vertices are not connected.
Here, each color indicates a separate mesh
How did he do that exactly? I get these weird groups when I grouped by normal.
On another note, shouldn't this thread be stickied like the regular competitions?
Sculpted the blade wrap as I need practice in that dept, total noob there. I think the result is OK, just mediocre. I'm happy with it as my first cloth wrap sculpt, but i wouldn't use it for a bake I don't think.
I could probably do a much better job just modelling and sub-ding it, a lot quicker too.
I guess that's one thing I need to learn is when to sculpt and when it doesn't need it.
That's a great start! I'd only advise you to be careful with some of the geometry. I see that you don't have anything under the straps near the blade, which is pretty obvious where the straps meet the metal, or rather don't meet it, because when you smoothed it the mesh got "smaller" and you can see holes there. Same thing on the handle, when smoothed you can tell that it's not connected to the blade, so I'd recommend you to try and merge these two and extend the mesh on the blade. Also, there's some weird smoothing going on at the circles, can I see the base mesh?
Here's a quick paintover with some of the problems:
Even with those problems, it's a nice start!
Post 54
Please tell me if there's anything wrong
I'm surprised I got a decent bake onto the LOD1, there are some weird parts but I figure being LOD1, people would not notice such details whilst playing. Any advice on making LOD1 though would be great, I worked from LOD0 & removed edges & moved verts to keep the same silhouette.n The automatic reduction tool was just too messy.
@quockhanhlk - it looks like you're going in the right direction, but be careful not to add too much geometry when it's not needed. You currently have a whole lot of subdivides on the handle, for example. If geometry isn't defining silhouette, it probably means you don't need it! Of course, those can later be used to create the handle strap, depending on your own work flow.
What did you bake that diffuse from?
Thx for feedback, About holes they just appeared after I smooth.I wil post tge low poly to show.
As for the texture, I coloured my high poly with basic coloured shaders & baked it down to my LOD0, this provided my base colours to hand paint into using CS6. I used my AO bake on a multiply layer above the base colours to create depth, & then painted in some extra detail. I also used a metal texture overlay from CGtextures for the scratches on the metal.
When polypainting, should I merge all my subtools into one subtool?
I made some little changes to color (its probably not noticeable)
@cag : how many tris is the low poly? Don't forget to do the Lod0 as well as the Lod1
No need for that.
Your sword is looking pretty good! I'd recommend you to give some more love to the handle straps though, they look pretty barebones now.
Since I decided to finally get into some modelling I find this a great opportunity to start out with my very first ever finished HP model and my first ever attempt at sculpting.. I work with Blender for maybe 2 weeks or so and this is my first output:
edit: I'll probably start all over again because it was pretty much a tryout to see where my current modeling "skills" will get me to..
Nice start! I have a few suggestions though. The blade seems REALLY thick, I'd recommend you to maybe go for half of that or even less. The shape looks spot on though. Be careful with those cuts and stuff, right now they look too detailed, something you would probably get from rocks, not steel. Try to check out some of Motenai's videos on YouTube and check how he does those cuts. Apart from that, some extra polish on the straps could really help.
It's a great start! I'm glad that you're joining the fun
also you can try some custom alpha with the dragrect stroke.
after all is a good start
My lowpoly baking mesh with UVs
Are the places where I split UVs fine? I always wondered the extent of splitting the UVs.
The method I used was that I used the base mesh (before sculpting) as a low poly base, then I tweaked the topology on how I changed the sculpted topology.
yeah it looks good. No reason to split the top of the blade imo. The distortion you'd get on the little tri up top won't be noticeable anyway.
You're only creating 2 more verts but it's 2 more than needed. Not really an issue but something to keep in mind.
(It's hard to give feedback here because everyone is doing pretty good anyway)
It does appear that you still have tris under the blade wrap, deleting those will give you more for rounding out the blade a bit)
Bigger is good, but efficient is best.
The only thing he really split was handle from blade (different material anyway, good place to split), and the blade in half... Less distortion than bending it around a corner. (Square map/long rectangle blade). He floated the blad geo so that's already split.
I dunno, I guess the blade wrap's verts could be combined to those of the blade once those extra polys are gone.
That would save a few faces, a few verts and make 2 islands one.
----
If you do go that route ghost I would leave the circle uv's intact and put the distortion in the wrap. The circle is much more noticeable if it's distorted.
It doesn't make sense why the handle is taking as much space as the blade. You won't see the blade in game, so why is it so big?
At the very least, it would be worth it to have the blade be a single island in the same way the fuller/back part of blade is. It will be much easier to get precise texture then having to combat with un-natural seams.
I didnt scale the hilt it was just like that when I unwraped and I just jigsawed the uvs.
here are the groupings. Black indicated under the mesh or hard to see.
So here's my point. With minor uniform scaling, you can get the two pieces of the blade together in a diagonal that will keep you from getting a seam where your islands would meet. Heres a mock up.
The only thing I scaled was the hilt, I'm not exactly sure how the blue/green fit together, so I just over lapped them, while taking into account how they uvs "might" stack up. Hope this helps/explains what I mean.
If you want maximum uv space,and are willing to accept a little distortion (honestly I don't see it showing much in the blade anyway as it's mostly shiny steel) I'd go this route.
You need to get rid of those hidden tris anyway (That's 48 tris, can make the blade nice and round with that). If you can't see it it's wasted (it can be useful for baking, editing but not useful for in game).
hahahah YESSS.
Honestly, uvs are my weakest point, just try to get that blade together somehow. You will be a happy man later in life.
Though now I need to start worrying a little less about distortion since everything in Dota is baked from high (a lot easier to bake around distortion than hand paint around it)
The blue and green is one piece of the blade but I just split it at a corner area so I can maximize space.
I noticed that the yellow area is distorted a lot, what are the limits of the distortion? (I'm thinking not to have long stretched tris or quads)
The hidden area was just going to be the lod0 and I was going to delete the hidden areas after the lod1. I'm still kind of confused on how to create the lod1 from the lod0, I saw valve's highpoly/lowpoly models and they shorten the mesh.
Back to UV mapping, generally how should the proportions be? I see on heads the proportion is like 2 for the head and 1 for everything else.
Generally it's best to have 100% no distortion and equal pixel density. But that's impossible without every poly on it's own island.
So I try to put most of it where it's less likely to be noticed. Like the handle, since in game it is under the hand.
That's why heboltz3 said to shrink that island, give yourself more space for the important stuff.
Of course the paintover I did was just pasting your bits together. You'd definitely want to relax the uv's some so the distortion is spread evenly across the entire blade. It'll be less noticeable then if just a few polys have all the distortion.
(Still, you won't get that much distortion over what you have
That blade is fairly even colors. If it had text on it the distortion would be harder to hide.
Do you use a checker map to check uv's? If you use the checkerboard pattern (max has it in uv editor, I suppose most programs do now)) you can see in the view how even those checkers are in shape and size. Best way to get proper scale.
======
Still, on high or low poly (not talking high res bake obj) you don't want those polys that are hidden. They do no good. Wasted. And in game they are bad because they are still drawn, but they are overdraw.
The engine has to draw them in front of terrain/characters, etc..., then it has to draw the cloth on top of that too. One model might not make a big difference but if every model in game was like that it would stack up quick.
But splitting the circle and wrap apart work too.
I've noticed Valve's Dota2 uv's really aren't very efficient. A lot of the items I have looked at have large areas of blank space.
Basically when uv mapping you are looking at 3 things:
1:space usage
2:distortion
3:seams.
Space usage needs to be optimized by arranging the uv's so they pack tightest. No wasted space.
Distortions, it's always more obvious on things that are seen easily and things that have crisp details. Text would be easier to see distortion than on your gradient painted blade. The gradient is already kind of like blurred pixels, so distortion ddoesn't show well.
Some text would be hard to read and blurry if stretched a lot.
Seams: Seams are hard to paint out, so obvious in game. They also split the verts on your mesh, making more verts, making your model more complex to render. Best to split on hard edges, between materials (like blade steel and leather handle) and in places like armpits that the player won't see much.
You just have to find the best balance. Use as much space, hide seams and distortions where they won't be noticed and try to use as few islands as possible so you're not creating extra verts.
If you put this material on your model (sorry it was quick) you can see how even your uv's are. The squares will all be even sized and square if your uv's a really clean. If not you can adjust them until they look as good as possible.
I am still improving my UVing ability, I used to waste upwards of 45% of ym UV space, & now I'm averaging around approx. 25% ~ 20% depending on what I'm UVing of course.
On main topic, I've been really busy this week, but I plan on implementing my sword into game later tonight. What a fun challenge! Definitely an easier scope creating 1 great asset rather than creating a set. I'll leave sets alone for a while.
I don't know about other programs but in blender shows the distortion by a stretch feature. It color coded by gradients and it shows how warm the UV spaces are. Blue being the coldest (not distorted) and yellow being the warmest (high distortion).
You can see the worst distortion is on the 'blade spine'. It has a few very long squares in the middle. shorter ones by the wrap. If you adjust those verts by hand you can minimize that and even them out some.
The larger squares on your handle mean it is getting less pixel density, but that's alright as you've chosen that (since it will be seen less).
As far as wasted space, you just have to eye that. Use up space as good as possible and that's all you can do, but be aware of it.
I only did that as an example to illustrate it better. I'll probably never do it again
Thanks very much for the constructive feedback! I've completely remodeled my pre-sculpt HP and tried to implement your suggestions. Heres my outcome:
Slow progress but I made the diffuse texture (color map) by just using the Hi-vertex as the base colors and adding the cavity/ao/normal map channels on top of it. I also added some scratches and grunge from cgtextures but I made it really transparent. Is it too sutle?
@FlashL Very inspiring high poly model considering it looks like you've modeled it strictly within one modelling package.
@Baddcog Nice model! Practice makes perfect as they say. Your colours definitely catch the eye.
@rock Nice to see you taking constructive criticism and producing a great model afterwards.
Here's some progress shots:
Base Mesh
Sculpt
Low poly up next followed by some baking.
In other zip files (the more recent ones i guess) the mesh is usually straight up and down to the left or the right of the hero mesh. For sven's case the weapon is sticking from its hand as well as it being tilted. Is there something wrong with the rig i got (i got it from the default) or do i just have to start rotating my mesh to fit the default mesh and rig it like that.
On another note, did anyone see the color/normal map for svens weapon. I only saw the mask1 and mask2 for the blade.
Question: It may be off topic but I was wondering about toned backgrounds for promos and concepts. Should the promo backgrounds just be simple and low saturated? Or should I create motion lines and/or gradients to lead the eye?