Are there any problems with doing an exploded bake? If there are no significant disadvantages I can't see why you wouldn't explode the mesh for baking because you're going to get cleaner results, the gun I recently did required to be exploded for the cleanest bake. If there are reasons however to avoid exploded bakes then I'd like to hear them.
For the general baking stuff you pretty much want to read this.
Your high poly is looking pretty cool so far, you might find when you bake that some of the edges are a little too sharp and don't translate to the nm that well, the high poly edges should (generally) be smoother than they appear on the actual object because the goal isn't to make a good looking HP, but a model that bakes well.
I've actually read that thread multiple times ^^ And thanks.
Though i haven't read that wiki post, Which looks pretty informative. I was curious if i could do multiple bakes like he stated and get good results. So maybe I'll try that.
I'm also curious about edge thickness.
I've heard racer talk about edge width rather than softness or hardness before. And i guess my question is, Should the edges be the same width without exception on any given "Material type" (metal, lets say). Even if the reference has some variable edge widths?
I"m not asking if i should make the model accurate to the reference edges rather. if an edge on the reference is 2 X as thick as another, Should i make mine that way?
example: (Ignore the varying width portion. Had a question about that bit, but pretty sure i know the answer )
I obviously haven't gone over this model yet to even out all the edges. But its still going to concern me later on.
I can't think off the top of my head why you'd try and keep edge width specific to a material rather than matching references. That could be a good way to think about edges though when modelling if you want consistency, but metal can be both sharp and rounded in different parts. So I wouldn't worry too much about making sure your edges are all the exact same width across one material type. So, have varying width, as long as it makes sense.
Some more experienced people might come in and have better info on this though.
For the AO bake, theres no reason you can't have a different version of the explode just for AO. Have the parts you want to shadow all together, and then anything you know might move/animate or anything moved away.
Hmm. Interesting. I'll have to try out some different methods once it comes time to bake. Was really just trying to prepare and gather some feed back on the best process before i go to that stage. I've bake tons of stuff before, But I'm never happy with my results. So i want to do it as well as physically possible
The time while i model I'm trying to be conscious of how "Wide" the edges I'm making are. and how to make them the best they can. And it was just painfully obvious to me how far mine are from perfect.
Try to find the the chamfered edge here XD
I remember that racer talked about abusing chamfers to create nice wide edges with short falloff.
I guess my question ATM is how can you create all these nice wide edges and keep all the edges that chamfer create under control? i can come back with a specific example on the model later today if need be.
Just blew my power supply (working on a laptop) so i figured i'd post a little update seeing as i might be gone for a week or two. Depending on how long it takes get a new unit.
And yes, I know the edge consistency is terrible between the slide and the grip/frame.
Agreed, high poly is fucking beautiful (I'd be interested to see wires as I'm not 100% sure what parts are floaters or not, some stuff might've been faster/easier to do in ndo?), keen to see your low poly and bakes though. Keep up the awesome work!
I tried to use as few floats as possible. I'm fairly new to sub d modeling (been doing it for 7 mo. or so) so i try to force myself to model the complicated and challenging bits as often as possible. as far as projection goes for details that are floating above the surface there is only mag "dots" on the back and all the little nipples on the grip are floating. otherwise, it tired to make the geo as it would be in real life. So bolts are obviously different object etc.
Not sure if everyone will yell at me for such terrible topology. If its a flat face, its spot to dump tons of terrible topology But i hear a lot that the only thing that matters is how the smoothing and visual turns out. So hopefully that's true and i don't need perfect edge flow and what not ^^
the high poly is generally pretty good, plastic has a good feel.
also you will have to change this type of inset screw to be closer to the surface, unless you want some huge skewing or want to model those in on the low poly, which is a waste. do some test bakes and the other areas where you need to increase geo/simplify hp model will become obvious.
edit: also you'll need to make the grippy bumps bigger so they aren't just 4px dots when baked
the high poly is generally pretty good, plastic has a good feel.
also you will have to change this type of inset screw to be closer to the surface, unless you want some huge skewing or want to model those in on the low poly, which is a waste. do some test bakes and the other areas where you need to increase geo/simplify hp model will become obvious.
Nearly done breaking down the high poly into a low poly, so I'll definitely be doing a lot of test bakes here shortly ! Thanks for the advice.
To be clear, This isn't a skewing cage problem right? it wouldn't be fixed by cutting in a vert, or a loop around that detail? then removing it for the final low poly model?
This is more of a detail that simply wont read well in any other view aside from the straight on? Sorry if those are obvious answers. I have read a lot in EQ's threads and all the baking/normal/projection wiki's and all that good stuff. But application is really where i wrap my head around a subject
If the latter is the case, What might you suggest instead of that detail? because the foremost face of the "bolt" (really modeled like a rivet) is as far forward as i can be without going beyond the grip material. Its that way cause I'd assume it would be uncomfortable to grip a hand full of bolts when you grip your gun
Nearly done breaking down the high poly into a low poly, so I'll definitely be doing a lot of test bakes here shortly ! Thanks for the advice.
To be clear, This isn't a skewing cage problem right? it wouldn't be fixed by cutting in a vert, or a loop around that detail? then removing it for the final low poly model?
This is more of a detail that simply wont read well in any other view aside from the straight on? Sorry if those are obvious answers. I have read a lot in EQ's threads and all the baking/normal/projection wiki's and all that good stuff. But application is really where i wrap my head around a subject
If the latter is the case, What might you suggest instead of that detail? because the foremost face of the "bolt" (really modeled like a rivet) is as far forward as i can be without going beyond the grip material. Its that way cause I'd assume it would be uncomfortable to grip a hand full of bolts when you grip your gun
fragfest2012: Thanks!
no, it's a projection skewing issue. if you simply retopo over that screw it will see it sideways unless you cut in geo, which is a waste.
also never remove geometry like that from your low once baked since normal baking uses the mesh normals from your low poly to compute smoothing.
Ah. I see. That's why people raved about being able to change the LP with the hand plane workflow? Makes a bit more sense now.
So that makes me curious about the silencer. I did a test bake and had to add geo to the front so it wouldn't skew the 6 holes around the main hole. Is there no alternative to solve skewing like that? Add to the vert count or have skewed details?
Being as your title says "Walther p22 Based Pistol-Game Model" then I may be wrong, and just ignore this, but if your aiming to replicate a p22, the magazine and bullets aren't accurate. They both look much more like a 9mm.
22lr rounds are rim fire, so they don't have a visible blasting cap, and are much skinnier... also making your mag way too thick; just guessing off the top of my head I'd say it should only be ~1cm thick. We have a p22 in the house so let me know if you want any specific pictures.
Being as your title says "Walther p22 Based Pistol-Game Model" then I may be wrong, and just ignore this, but if your aiming to replicate a p22, the magazine and bullets aren't accurate. They both look much more like a 9mm.
22lr rounds are rim fire, so they don't have a visible blasting cap, and are much skinnier... also making your mag way too thick; just guessing off the top of my head I'd say it should only be ~1cm thick. We have a p22 in the house so let me know if you want any specific pictures.
Either way though, it looks lovely.
Yeah, i did have a bit in the OP about project details. But opted to tell everyone to tell me whats wrong with the model
And you did just that!. But yes, i am aware, And it is actualy meant to be a 9 mm with 16 rounds in the mag and 1 in the chamber. I didn't want the gun to be "puny" in the game but i really liked the look of the p22. So tried to make a a modified version of it that packs a bit more of a punch
Following suit of the modified deign the original p22 mag is a single stack, and rather "plain" magazine. So opted for more of a modern double stack type of look.
In that case, take a look at the Walther p99 for fun, it's basically the 9mm version. There are a couple of design differences, but I personally like a visible hammer on most pistols.
Benton mentioned it looked like a Bond gun, and that would be because Bond has used a p99.
Yeah. I did look at the p99 a bit, But imo, it has a lot less character. I liked the exposed hammer like you said, safety placement, and really all the details all over the gun were more interesting on the p22.
I've been working on baking this over the past few days, and I've reached some smoothing issues pretty early on. I'm hoping with the aid of the pictures i have here (from UE3) we can figure out whats going on.
Under the impression someone will link the thread anyway, Quoted from the hard edge thread:
"""Some general rules with that in mind:
To avoid artifacts, you must split your uvs where you use hard edges.
You do not however, NEED to use hard edges wherever you split your uvs(as some may suggest)
With an Averaged projection mesh you can use hard edges along uv borders with no negative side effects. Because the verts are already doubled at your uv seams, you get it for free. Your verts do not triple or quadruple if you have both a uv seam and a hard edge in the same place."""
So i might be over thinking this, Or truly dummer than i think.
and after reading that quote again, I think I'm both over thinking it and dummer than i think.
So does this mean you make the UV's around the model (SG included in the model) or do you create the SGs based on the UV's? Should my SGs Resemble the HP as much as possible? and i should make the UV's Support that?
I'm having a hell of a time understanding the rules of baking. sorry for all the noob questions that have been answered a thousand times. I'd like to say i get it from all the threads and countless pages I've read through, But its not quite clear yet.
So here are the pics i promised. Tried to keep colors associated with their respective peice. I'm sure its insulting how obviously i mark the error here, sense i know you guys could see them without any paint So sorry for that, Just trying to be helpful.
Also to note, The normals haven't been hand edited or manipulated in any way.
before i say anything, the high poly is looking really good so far! But as far as your artifacts on your low poly, generally this happens due to yes, smoothing groups and edge placements. Honestly, trial and error is your best bet. For the clip, if you have a little wiggle room with tri count, try putting a couple support loops in the middle. And for the smoothing groups, make it all 1 group around all the sides, and see if that helps
before i say anything, the high poly is looking really good so far! But as far as your artifacts on your low poly, generally this happens due to yes, smoothing groups and edge placements. Honestly, trial and error is your best bet. For the clip, if you have a little wiggle room with tri count, try putting a couple support loops in the middle. And for the smoothing groups, make it all 1 group around all the sides, and see if that helps
ok. So that's a perfect example. If i throw some loops around the mag, And make that face 1 SG, I would have to split up the UV's no?
Hard edge on the SG MUST have UV breaks. But Uv Breaks don't prompt anything to do with SGs.....?
And I had a feeling it might come to more trial and error. Just hate waiting for renders ! I'm very impatient
After playing with it for a bit, I think I've worked out the mag. One more set of tris i can see creating issues. Anyway. Progress ^^ just slow. Is helping me understand smoothing and projection a lot though.
This is shaping up to be a really great model. Can't wait to see it textured.
Also, just a hint, consider in your normal map preview screens to save your normals as .tga rather than .dds. .dds has weird artifacts and will make your normals somewhat jaggy.
This is shaping up to be a really great model. Can't wait to see it textured.
Also, just a hint, consider in your normal map preview screens to save your normals as .tga rather than .dds. .dds has weird artifacts and will make your normals somewhat jaggy.
I'm not sure what you mean. The Flat up there of the normal? i save it out as a PNG, from the .tga render, For preview.
This picture. I'm referring to those jaggy artifacts. I believe it has something to do with compression, I'm no expert. It's just I've had this problem before and it's solved by saving your normal map as .tga.
It's not a huge issue for the sake of this thread. But if you were making a preview or saving it for a portfolio or something you might want to fix it.
Oh, You mean those giant smoothing errors ? Or the pixelation there?
Working on fixing the those "pulls" right now. As for the pixelation and kinda spotty highlights, I'd assume that is is compression of the normal map into unreal.
Was on the fence about redoing the UV's. but here we are, mid night, Re did a bunch of topology all over the gun, hoping to get a full "Pre-Final Test" bake done while i sleep.
lemme know what you guys think of the new layout. I'm hoping I've resolved most the issues with the previous bakes. And now I've mirror and overlapped a lot of the matching UV islands. Will update in the morning hopefully with and updated bake
...Feel like there is a lot more wasted space on this second map. Oh well. i'm sure everything is using space a lot better in this one vs the other.
I think that tread mazz423 linked is really informative. I far from an expert in baking, but from personal experience, it always seems to do weird things when the model isn't all one smoothing group. Separate smoothing groups on uv breaks is something I'll have to give a try on my current model when I get there. I haven't done anything with UDK so It could read things very differently.
Also, there are alot of edges that can be removed. Most of them won't effect your final tri count but a few here and there will. I drew on the slide what could be removed, as well as the potential outcome, most importantly where there are more than 2 tris in a face that can be a quad.
Also it appears that you have a few n-gons, especially in some of your round shapes.
Yeah. I added all those stupid edges to the slide trying to fix the errors. The rest of it is in tris cause i accidentally made it an edible mesh instead of poly and found it too late so had to deal with it being triangulated (would explain the the edges connecting quads). N'gons i was under the impression that triangulation would just strip some edges in there? Maybe on the circles, But not the 5 sided ones?
as for the others, It was just my attempt to even out topology. But I'm thinking topology isn't my problem.
The Bake i did last night was a complete disaster. nearly every piece/flat face had some Triangle looking smoothing issue. SO. For the 5th night in a row. I get to do some more test baking and try to solve this.
Also, I've read that thread, And every post/page in it ^^ Didn't know how to say that politely whilst thanking mazz for the suggestion. It is a great thread, and i have read it. And the waviness thread, and most the oddball threads people post about their normal problems. But. Reading has never really driven a point home for me. I learn by doing. So. Here is to doing it again
Reading has never really driven a point home for me
Same for me, I found baking quite daunting when I first tackled it, luckily I had plenty of people around me to hassle for advice. If I were you I'd suggest performing a few test bakes on something far simpler than this model, make a set of cubes, create super simple high polys for them and try baking with different smoothing groups (all smooth or hard normals for example) and then try using control edges. just see how they react differently and what results they produce. Then once you think you have the basic understanding down try applying what you've learned onto this model, starting with this is very much a trial by fire.
Just did a test cube with to make sure my workflow wasn't junk. Worked perfectly. I've played with loops to change cage projection and that. I think my main problem is smoothing groups and Uv's. Edges are baking fine. Good really. And Cages all project exactly how i anticipate they will. Its just once i get the map on the model, its got all kinds of non sense Gonna try what i did last night, except manually doing SG's instead of letting textools do it.
Replies
One thing I was curious about with exploding bakes was AO - in some cases you might want the AO from the unexploded mesh, but not from other parts. Polycount wiki had some info on this where one method involved baking AO from exploded mesh and using the simpleAO tool on the lowpoly (unexploded). Here's a link if you want to take a look http://wiki.polycount.com/AmbientOcclusionMap?action=show&redirect=Ambient+Occlusion+Map#EarthQuake.27s_Baking_Method
For the general baking stuff you pretty much want to read this.
Your high poly is looking pretty cool so far, you might find when you bake that some of the edges are a little too sharp and don't translate to the nm that well, the high poly edges should (generally) be smoother than they appear on the actual object because the goal isn't to make a good looking HP, but a model that bakes well.
Though i haven't read that wiki post, Which looks pretty informative. I was curious if i could do multiple bakes like he stated and get good results. So maybe I'll try that.
I'm also curious about edge thickness.
I've heard racer talk about edge width rather than softness or hardness before. And i guess my question is, Should the edges be the same width without exception on any given "Material type" (metal, lets say). Even if the reference has some variable edge widths?
I"m not asking if i should make the model accurate to the reference edges rather. if an edge on the reference is 2 X as thick as another, Should i make mine that way?
example: (Ignore the varying width portion. Had a question about that bit, but pretty sure i know the answer
I obviously haven't gone over this model yet to even out all the edges. But its still going to concern me later on.
Some more experienced people might come in and have better info on this though.
Thanks for the replies.
Try to find the the chamfered edge here XD
I remember that racer talked about abusing chamfers to create nice wide edges with short falloff.
I guess my question ATM is how can you create all these nice wide edges and keep all the edges that chamfer create under control? i can come back with a specific example on the model later today if need be.
And yes, I know the edge consistency is terrible between the slide and the grip/frame.
Uploaded with ImageShack.us
HP is about done. bit more to work on the back of the magazine, Other than that is about LP and UV time.
And nice model
And Thanks
and a plain shaded view to see how the floats will fit in
I tried to use as few floats as possible. I'm fairly new to sub d modeling (been doing it for 7 mo. or so) so i try to force myself to model the complicated and challenging bits as often as possible. as far as projection goes for details that are floating above the surface there is only mag "dots" on the back and all the little nipples on the grip are floating. otherwise, it tired to make the geo as it would be in real life. So bolts are obviously different object etc.
Not sure if everyone will yell at me for such terrible topology. If its a flat face, its spot to dump tons of terrible topology
also you will have to change this type of inset screw to be closer to the surface, unless you want some huge skewing or want to model those in on the low poly, which is a waste. do some test bakes and the other areas where you need to increase geo/simplify hp model will become obvious.
edit: also you'll need to make the grippy bumps bigger so they aren't just 4px dots when baked
Nearly done breaking down the high poly into a low poly, so I'll definitely be doing a lot of test bakes here shortly ! Thanks for the advice.
To be clear, This isn't a skewing cage problem right? it wouldn't be fixed by cutting in a vert, or a loop around that detail? then removing it for the final low poly model?
This is more of a detail that simply wont read well in any other view aside from the straight on? Sorry if those are obvious answers. I have read a lot in EQ's threads and all the baking/normal/projection wiki's and all that good stuff. But application is really where i wrap my head around a subject
If the latter is the case, What might you suggest instead of that detail? because the foremost face of the "bolt" (really modeled like a rivet) is as far forward as i can be without going beyond the grip material. Its that way cause I'd assume it would be uncomfortable to grip a hand full of bolts when you grip your gun
fragfest2012: Thanks!
no, it's a projection skewing issue. if you simply retopo over that screw it will see it sideways unless you cut in geo, which is a waste.
also never remove geometry like that from your low once baked since normal baking uses the mesh normals from your low poly to compute smoothing.
So that makes me curious about the silencer. I did a test bake and had to add geo to the front so it wouldn't skew the 6 holes around the main hole. Is there no alternative to solve skewing like that? Add to the vert count or have skewed details?
22lr rounds are rim fire, so they don't have a visible blasting cap, and are much skinnier... also making your mag way too thick; just guessing off the top of my head I'd say it should only be ~1cm thick. We have a p22 in the house so let me know if you want any specific pictures.
Either way though, it looks lovely.
Yeah, i did have a bit in the OP about project details. But opted to tell everyone to tell me whats wrong with the model
And you did just that!. But yes, i am aware, And it is actualy meant to be a 9 mm with 16 rounds in the mag and 1 in the chamber. I didn't want the gun to be "puny" in the game but i really liked the look of the p22. So tried to make a a modified version of it that packs a bit more of a punch
Following suit of the modified deign the original p22 mag is a single stack, and rather "plain" magazine. So opted for more of a modern double stack type of look.
Thanks.
Benton mentioned it looked like a Bond gun, and that would be because Bond has used a p99.
I've been working on baking this over the past few days, and I've reached some smoothing issues pretty early on. I'm hoping with the aid of the pictures i have here (from UE3) we can figure out whats going on.
Under the impression someone will link the thread anyway, Quoted from the hard edge thread:
"""Some general rules with that in mind:
So i might be over thinking this, Or truly dummer than i think.
and after reading that quote again, I think I'm both over thinking it and dummer than i think.
So does this mean you make the UV's around the model (SG included in the model) or do you create the SGs based on the UV's? Should my SGs Resemble the HP as much as possible? and i should make the UV's Support that?
I'm having a hell of a time understanding the rules of baking. sorry for all the noob questions that have been answered a thousand times. I'd like to say i get it from all the threads and countless pages I've read through, But its not quite clear yet.
So here are the pics i promised. Tried to keep colors associated with their respective peice. I'm sure its insulting how obviously i mark the error here, sense i know you guys could see them without any paint
Also to note, The normals haven't been hand edited or manipulated in any way.
Thanks in advance!
ok. So that's a perfect example. If i throw some loops around the mag, And make that face 1 SG, I would have to split up the UV's no?
Hard edge on the SG MUST have UV breaks. But Uv Breaks don't prompt anything to do with SGs.....?
And I had a feeling it might come to more trial and error. Just hate waiting for renders ! I'm very impatient
And thank you
u will need all the uv space u can have, its never enough
like the mag, grip.. even cylinders... what ever u can
You're right though, i should go back and rework the uv's to make better use of space.
Also, just a hint, consider in your normal map preview screens to save your normals as .tga rather than .dds. .dds has weird artifacts and will make your normals somewhat jaggy.
I'm not sure what you mean. The Flat up there of the normal? i save it out as a PNG, from the .tga render, For preview.
is .png a type of .dds?
And Thank you! Hope it turns out well ^^
This picture. I'm referring to those jaggy artifacts. I believe it has something to do with compression, I'm no expert. It's just I've had this problem before and it's solved by saving your normal map as .tga.
It's not a huge issue for the sake of this thread. But if you were making a preview or saving it for a portfolio or something you might want to fix it.
Working on fixing the those "pulls" right now. As for the pixelation and kinda spotty highlights, I'd assume that is is compression of the normal map into unreal.
It is rendered out and and imported as a .tga
http://www.polycount.com/forum/showthread.php?t=107196
Lot of answers already in there.
lemme know what you guys think of the new layout. I'm hoping I've resolved most the issues with the previous bakes. And now I've mirror and overlapped a lot of the matching UV islands. Will update in the morning hopefully with and updated bake
...Feel like there is a lot more wasted space on this second map. Oh well. i'm sure everything is using space a lot better in this one vs the other.
Also, there are alot of edges that can be removed. Most of them won't effect your final tri count but a few here and there will. I drew on the slide what could be removed, as well as the potential outcome, most importantly where there are more than 2 tris in a face that can be a quad.
Also it appears that you have a few n-gons, especially in some of your round shapes.
as for the others, It was just my attempt to even out topology. But I'm thinking topology isn't my problem.
The Bake i did last night was a complete disaster. nearly every piece/flat face had some Triangle looking smoothing issue. SO. For the 5th night in a row. I get to do some more test baking and try to solve this.
Also, I've read that thread, And every post/page in it ^^ Didn't know how to say that politely whilst thanking mazz for the suggestion. It is a great thread, and i have read it. And the waviness thread, and most the oddball threads people post about their normal problems. But. Reading has never really driven a point home for me. I learn by doing. So. Here is to doing it again
:thumbup:
Same for me, I found baking quite daunting when I first tackled it, luckily I had plenty of people around me to hassle for advice. If I were you I'd suggest performing a few test bakes on something far simpler than this model, make a set of cubes, create super simple high polys for them and try baking with different smoothing groups (all smooth or hard normals for example) and then try using control edges. just see how they react differently and what results they produce. Then once you think you have the basic understanding down try applying what you've learned onto this model, starting with this is very much a trial by fire.