You want to cap the "hole" off with a nice fat bevel, so the baker catches some detail there, then you'll just paint the inner bit black in spec/diffuse.
Simple extrudes never really register with bakes - if you're seeing it head on and there's no beveled edges to help catch the light, you're not going to see it. Same goes for the normal map baker.
I would think AO would work either way though (but it could depend on how you set it up)
Perfectblue: No, because the rays aren't hitting anything. If you want the look you're talking about you need to make the hole gradually narrower (a chamfer). But that's not what you should do. What you want it simply a black circle shape, no more than that.
yep, as regards the skewing problem (superhappy, normal map shading problems fixed by SG's is another issue)
Perna/EQ thank you both for your responses. I just made a quick test object thats similar in nature to the silencer problem I encountered with the flat gray for the inward extrude.
Did the same thing here. The LP is just a cap, yet this time it picked up the extrude from the HP unlike the issue with my silencer.
If the rays are not hitting the inside part I pointed out, why are they hitting the rectangle extruded details highlighted in blue in my original picture?
editFurthermore, on the original HP if you look at it straight on I did bevel it in a little bit. I think the rays should be hitting it no? Instead of producing a flat black.
re: small areas marked in blue, because these are just slightly inset, the AO is pretty weak here, but the inset cylinder goes way back, and settles on a dark gray color. IMO: these should have some nice fat bevels too for better readability
Thanks will give it a shot. I just edited in a picture to the above post, but ill post it again here. The extrude of the barrel is not at a straight 90d, its beveld some. Thats still problematic and results in flat grey.
Beveled "a little bit" is the key here, you want nice fat bevels. Remember to zoom out! You'll never view the end of the silencer this close up, so you shouldn't model assuming you will, again you're also limited by texture resolution.
[edit] *Anal retentive hat on*: Your indents aren't evenly spaced!!!! =P
Anyway, you're learning the absolute most important lesson of all here, never be afraid to edit your HIGHPOLY to fix bake issues, people so often think everything that could go wrong is because the low, and will refuse to touch the high after its "done".
I removed the inward extrusion on the gun as suggested so its just a flat cap on front. However, this is creating a new problem. For whatever reason the inward extrude isnt registering with either the normal or AO map. For instance, the inward extrusion for the barrel of the gun just becomes a flat grey. It should at least have the appearance that its going inward much like the holes on the side of the silencer.
Did you float the hole and forget to add it to your projection modifier?
e: no, you didn't, it just looked that way in the second image you posted.
Just read through this entire thread and I have a bit of advice for you. TRY doing stuff and experimenting. Don't just rely on us to fix every single problem, try and use your noggin to think exactly what you're doing. Sure, baking's cool, but try and figure out what the baking process involves.
I learnt loads of stuff just by experimenting with simple shapes and bevels etc, and looked to see what baked well and what didn't. Experiment and explore things by yourself, you'll have a much greater sense of achievement having figured it out by yourself.
Sorry guys a few of my replies came directly after you made some responses so it may have looked like I was just posting some stuff rapid fire without reading. That was not the case, I just posted directly after some of your responses before refreshing the page. And yes I probably did read through to fast.
@PoogoP and perna, yes I agree that is what I do. The vast majority of what I know so far is experimenting on my own or googling things when I get stuck. Sorry I gave off that impression I am not trying to have you guys spoon feed me every answer. If you struggle through something to learn it, you will come out on the other side with a much better understanding of what factors are at play than if you are just told to XY "just because".
Thanks again perna and EQ, I didn't mean to rapid fire questions like that. I didn't mean to aggravate anyone, my apologies.
Here was the fix anyways. It wasnt so much the bevel amount that was the problem, it was how far the extrude was going back the rays were not picking it up. Even though by my own judgement I thought the extrude back was relatively small (probably went back 2/10 the length of the cylinder), I guess not small enough.
Heres a fixed copy of the AO with the result I was looking for.
@Perna LOL, yah man I know im still working on it I just wanted to show that I fixed the original problem I was bitching about . Just messing about with random shapes and baking them, I wanted to git rid of that gray cap tho and find out how/why and I did so I posted it. Ill work on the "final" one now I promise!
EDIT: With black of course.
Well in this case, his supporting geometry can be a bevel at the end of the cylinder that improves the look of the mesh, and he's already cut out a bunch of geometry from the inside, he could also cut down on the number of sides if it was really an issue, but honestly, you should always plan for some corrective geometry in your polygon budget.
I really have no Idea what you're saying about SG stuff, you should post some examples. Chances are if adding SG fixed skewing type details with your bakes, your bakes weren't set up properly in the first place, and your SGs also caused nasty seams from missed detail. Again, with a cage/averaged projection mesh, your lowpoly mesh normals are ignored, so it has no effect on ray direction, the only thing it would help with is smoothing errors. I put a lot of time into writing a big ass thread about this, you may want to give it a read. =P
@Perna, did a diffuse pass with a black material and used that to mask out the holes on the original lighting.tga, but I am a bit confused as to why. You stated the black within these indented holes is needed otherwise it may give off the impression there are plates behind it. Obviously that is not desired, but the indents in the reference occlusion is not solid black, more so a very dark grey, at least thats how it appears to me. Also should I gradate these indents a little bit so they are not just solid black, like I did with the bevel on the front gun barrel of the silencer?
Do I need to worry about making other objects have solid black in the AO. Such as the indents on the grip (shown below)? Or should the default of whatever the lighttracer produce be sufficient and the silencer was a special case with the black and AO. Trying to understand what circumstances warrant this solid black AO approach.
Umm the solid black is because the hole at the end of the gun barrel goes right into to the gun, and thus is very dark. You don't have to make everything black, only where the hole is really deep!
Umm the solid black is because the hole at the end of the gun barrel goes right into to the gun, and thus is very dark. You don't have to make everything black, only where the hole is really deep!
I hope I've understood you properly.
Yes that was my understanding, only a few key pieces need to be pure black. It makes sense to have solid black at the end of the gun barrel because it fakes the extrusion of the barrel without having to provide the geometry that no one will ever see.
As for the many indents around the silencer, they are not that deep, so I am not certain why perna is suggesting they should be black.
PerfectBlue, if they're not holes but just indents, why would you colour them black?
Just use common sense.
You have a reference pic for the silencer you're making?
Right, thats why I was scratching my head when you were telling me to make the indents on the silencer black. Its a mix between a few different styles but this is basically what the indents on the silencer look like. I will remove the black mask on my previous post for the AO. Thanks
Alright thanks. Also as far as LP construction goes is it similar to HP concerning tris and ngons. e.g. Its ok to have tri's and Ngons in your HP as long as you can't see any errors/artifacts. Does the same thing apply to LP, tris and ngons are fine as long as it looks fine/doesn't produce any render errors/artifacts? Or is this problematic for game engines.
Also just found the original references for the silencer. Its a mix between the two. The indents in it are relatively deep compared to the reference above, thats why they are modeled that way. Mine may still be a bit too deep however.
Why would anyone not just paint the black holes into their AO map afterwards? It seems awkward to try to adjust your bake, and it's not like it takes any time at all to make a new layer above the ao and paint in the black holes, so even if you update your AO bake you still have black holes where you need them.
WIP of the low poly at 1,200 tris so far. Since ive read that tris are ok to have in the LP im frequently using them to get rid of control loops, although when I do have tris I try to prevent them from being too thin and narrow, as I vaguely remember reading thin and long triangles on your mesh can be problematic for game engines.
On the right track? Still have to make the LP of the mag, front tac rails, and the laser sight.
Just some general advice here: Try to make sure that you remember to give a good amount of detail to the areas that will be up close in FPV, make sure none of that stuff is noticeable blocky or low poly looking, but instead nice and smooth, and looking as close to the highpoly as possible, and then use less geometry for areas that are less likely to be seen.
@Perna, awesome insightful post thank you. My target polycount is anything within a reasonable range of a modern FPS. Since you said I could get away with 4x the count I guess about 5k tri budget for the total LP gun would be reasonable?
edit: also thanks for the insight EQ, appreciate it.
You need an RGB mask or matId mask output anyway, so why not make one for the holes? Painting them by hand takes time and will be inaccurate.
I suppose it might be difficult or inaccurate to paint out some wacky ridiculous holes but for the barrel of a gun or something similar it's usually dirt simple. I've literally never had it take more than a minute and I'm not sure I save any time versus selecting geo/making black geo to fill holes.
Obviously I'm baking a diffuse material mask (I'd hope everyone does this by this stage in the game) but it just seems unnecessary to mix in fake lighting in the bake rather than painting it in unless you have some seriously weirdly shaped holes.
Already eating up the budget 8k without mirroring over big parts like the body or the other half of the handle. With the laser sight taking up a huge chunk around 2k. Trying to leave the parts noticeable in FPV with a higher count. Also what would be a acceptable amount of work hours to create this guns LP? I would like to remove the actual geometry for holes where I dont need them and just use black and mask them where possible. And keep the holes in the LP where you need to see through it. I think the laser sight may be candidate perhaps for removing the LP holes but using black mask, but it is somewhat noticeable in FPV.
I have merged these two threads together(p90 + first bake) as its all the same topic. Try not to create so many threads, esp when its all discussing the same topic, for instance I looked at your post history and you've created 5 threads on uv mapping in the last two weeks. Try to keep all of your stuff one on place, it gets really confusing when you create more threads.
I'll try to get a reply to your latest post soon, just wanted to make a note here.
Your biggest problem is just, you have no idea where the detail should go. 296 tris on that thing? WTF, it should be like 20 tris, let your normals do the work! The lasersight is also way excessive.
@Perna, Thank you so much, that was very insightful. Who do I make the check out to ? Incredibly appreciative you took the time to do that, thank you.
Rather than post every tweak I made, I'd thought I would just post the areas that had major problems. Still have tweaking to do for sure, need to spend more time on the body and handle definitely, but its a step in the right direction compared to where I was I think.
Also thinking about removing some of the loops at the bottom of the second grip (the one that doesnt have the saftey or trigger).
that sidesaddle STANAG is driving me nuts. that's all I got, other than the fact that I've saving out all the images EQ has posted in this thread because as per usual (or as PER usual dohohohohoho) he's dropping pearl of wisdom on the swine
Correct me If I'm wrong but usually models tend to look better if you don't force those triangles. If you can afford it in polycount make it quad. Usually normals look better. Since amount of polies isn't such big issue today its better to use quads. No poles or long triangles. Ofc there are places where triangles are good choice. Here I would stick to quads mostly though.
I'm confused by how many parts your mesh has. You'll want to merge all the parts you don't specifically want to be removable. That'll help you work on resolving your loops and so on.
Do you mean apply symmetry, and collapse the stack to make it one object? e.g. on the handle and body
You mention about not having mirrored the pieces yet. Don't ever do that. You're working in max. Use a symmetry modifier. It'll show you the "mirroring" realtime as you work
Sorry I meant symmetry, I know they are not interchangeable my apologies.
Part of the reason I was hesitant to remove the loops around the grips is because my first pass at the low poly body was so low on tri count you guys told me to up it. So I had been focused on preserving the tri count and seemingly ignoring topology and edge density. But the layout of the edges was terrible, and most of those edges were not even meaningful after taking a closer look at it.
Anyways tried to adjust the curve density especially around the body grips, trying to keep only what I need for the shape. I have your notes tacked up on my wall to study, thanks again for providing them. Thanks for being patient with me, I have a better understanding of what needs to be done now.
*note, on the handle (below) there are a few separate objects so geometry may not appear to line up in some areas.
total count with symmetry applied to everything that needs it so far is around 4.8k. One of the rails needs to be copied around 2 more times which is around 300 tris at the moment I believe, as well as the center tac rail for the silencer.
EDIT Update : I think the LP is getting there, the whole thing with every piece modeled is now complete. Im sure there are some optimizations I can go in and look for though.
Still, its like someone took a futuristic, lightweight gun that was well balanced, short, and maneuverable and dicked around till you can barely hold it anymore.
Good job with fixing the low poly, there are a few more things I would change, but for now it is looking good enough.
EDIT: @ perna, just posted this right after yours. Going back to read your stuff now.
Doing some test bakes but I am having issues in some areas such as the handle in 3DS max 2011. The bake results for the handle show multiple artifacts and a very low appearing resolution even though I am using a 2048x2048. Using 3Point Shader
Handle Bake Test
RayMiss Check
LP + cage
The Normal Map was rendered with these settings:
2048x2048 normal map
Renderer: Antialiasing enabled, Filtermaps Enabled using Mitchell-Netravali. Global Supersampling Enabled using Hammersley
Advanced Lighting: Off
The AO map was rendered with these settings:
2048x2048 Lightingmap
Renderer: Antialisaing Enabled, filtermaps disabled.
Advanced Lighting: Light tracer active with 200 ray/samples and 0 bounces
Yes the handle needs to be reduced further its roughly the same amount of tris as the body at the moment. Yes I do use the symmetry modifier, I have just been posting the tri count for half the mesh up until now, I will no longer do this. The 6800 for the full LP is accurate, with all parts instanced and symmetry on. About working with the instances copied around, point taken. What I had been doing is just keeping track of it in my head, knowing what the tri count for the tac guard was, but it makes more sense to go ahead and instance parts such as this.
Yes, I have a far better understanding of what is required of the LP now and what you can get away with when baking normals to fake the detail. Major credit here goes to your paint overs they really helped me see what was going on, so thank you again. And no I am not just blindly copying the paintovers, that teaches you nothing. I am trying to take the time to analyze how and why. e.g. Every edge I try to look at it and think, "What purpose does this serve" it does serve a purpose "can the topology with it be made more effecient" etc. Things like that.
Yeah I really think it's much too soon to bake. You say you get low rez looking results despite rendering at 2048x2048. how much of the 2048 are you using for that piece though? Is this piece going to be detachable?
About your settings:
-I wouldn't use any of the weird fancy filters, stick with area at size 1.0
-Hammersley is ridiculously slow. Only ever use it once, when you are 100% sure that you have the exact result that you want. Until then stick with max 2.5 star
-same with AO.. you don't want to sit around waiting for AO to render when you aren't even getting the results you want out of the normal map. That's just a waste of time.
You turned off advanced lighting, so what are you using for AO?
Anyway yeah I think you're ahead of yourself here and need to go back and finish your lowpoly, then smoothing groups, then UVs, then bake setup, THEN finally your bakes.
I just wanted to do a test bake to see if I had any artifact issues, and I do, much like I did with the silencer. So I am trying to troubleshoot whats causing these problems with the test bake of the handle at the moment.I know I am not ready to bake the whole thing yet. But from the small test I did, I am not sure why its producing such errors. For the UV space for the handle its not optimized to take up all of the uv space, but I scaled it up so it takes up a good portion. Perhaps taking off some of those filters may fix the problems with the bake.
For AO its light tracer with 200 ray/samples and 0 bounces with advanced lighting turned on.
Make sure you go to preferences->viewports->configure driver and choose
the highest texture settings
2x "match bitmap..."
anisotropic
linear
Done.And changed to recommended render settings for the test bakes, additionally scaled down the UV space for the test handle bake to a more appropriate size.
Errors are very pronounced, not sure what might be the issue. Going to fiddle around with settings and see what happens. Heres a screen shot, using the same LP and cage as before. The normal definitely does not look correct
Unfortunately I tried experimenting pushing the cage varying amounts, all of which produced a similar result. Below is a link to the .obj with the two objects without any modifiers on them.
Thank you very much for taking the time to look at this. Perhaps you will spot what is going awry.
@Perna, thanks for taking the time to look at it. Yes I did try to encompass the LP over the HP since thats what the majority of tutorials I read over suggest to do. Thanks for clarifying the myth surrounding this issue.
As for why its not baking, might this be some sort of Max 2011 bug, because I am still unable to get this particular part to bake at all. I would settle on just getting it to bake with all those errors for now, rather than not bake at all.
Since the HP and LP handle are not designed well for baking, what would would you suggest for a bake friendly version? Making more of these smaller parts/ intersecting peices one mesh like you talked about earlier, such as this handle?
Depends what you mean by "look better". If the aim is to make the wireframe look cool for pimping on polycount, then it's as pointless as gridmapping UVs.
It's easy to see in the original mesh the extensive use of triangles makes it very hard to read, and so those triangles are placed in all sorts of awkward configurations which will lead to poor normals and be hard to work with.
There's no need to triangulate planar surfaces.
By "look better" I mean how normals look. Don't know if its maxs fault (considering it has issues with displaying normals) but usually meshes with quads are looking better with normal maps. Maybe its my lack of experience but some times I changed triangles to quads in low poly, rebaked and usually normals looked better.
Update. Here is my first unwrap. I tried to scale up the large parts that would be up close in FPV, and scale down unimportant parts while not wasting UV space and keeping enough padding.Some of the less important parts I had to scale up a bit even though I wanted to leave them smaller, as they were so small they were seeming to have issues picking up any information from the bake. In addition, I tried to keep parts of the same object together and keep the placement of UV's logical. The final low polygon is around 5.2k tri's I believe. Anyways heres the flat, how can I improve?
I removed the flashlight and laser sight, for two reasons.
A. The laser sight HP was no good (as shown by perna) for LP creation, I would have to redo a good chunk of it.
B. This may be going into a mod and the flashlight/laser can be obstructive in FPV.
@ Perna Thanks for another critique and for all your contributions and time in this thread they have been most helpful. Please don't be tempted to do any more paintovers, I don't want to be a nuisance.
You need to try to even out the pixel density on the butt, your going to see a nasty resolution transition to the rest of the gun.
Sure. My reasoning behind less resolution for that part was that its obstructed completely in a FPV. I figured the resolution it had would be fine when viewed from other players. Will definitely need to scale up the part on the far right though, the density is pretty low (the object with the huge numbers).
Your uvs are a mess, too many seams and some in really strange places, like the seam on the inside of the grip.
You need to straighten up some of these lines, soem pieces look rotated at slight angles. Its going to take more pixels to represent a shape rotated 2 degrees than a shape that runs with straight lines. Remember pixels are square, to represent a line you only need one row of pixels. To represent an angle you need 2, so your textures bakes will look lower res than they should.
You need to be much more mindful of how you're packing everything together, you need to leave a lot more padding room, or else your texture will blur together when mipping, some uv elements are so close that you will get bake errors as well.
Your uv density is all out of whack. Many areas that are seen close up have half the density of areas that are farthest from view. You should find a script to unify your uv density, then go in and scale up all the stuff that is clear in FPV, then repack it all.
Before laying out my final uvs, I like to do a quick unwrap, and then a dirty auto-pack and then testbake. This will help you nail down any geometry related baking issues before you finalize your uv layout, which is a good thing.
Oh one other thing, there is a round half circle shape right in your face that has half the # of sides as the barrel, which is further away from view. This makes the gun look lame and lowpoly. Never have lame blocky stuff right in your face, it will always look bad. Always reduce geometry from an off screen element if you are over your limit and need to add it where it counts.
Replies
I would think AO would work either way though (but it could depend on how you set it up)
Edit: I am far too slow.
Perna/EQ thank you both for your responses. I just made a quick test object thats similar in nature to the silencer problem I encountered with the flat gray for the inward extrude.
Did the same thing here. The LP is just a cap, yet this time it picked up the extrude from the HP unlike the issue with my silencer.
If the rays are not hitting the inside part I pointed out, why are they hitting the rectangle extruded details highlighted in blue in my original picture?
editFurthermore, on the original HP if you look at it straight on I did bevel it in a little bit. I think the rays should be hitting it no? Instead of producing a flat black.
[edit] *Anal retentive hat on*: Your indents aren't evenly spaced!!!! =P
Anyway, you're learning the absolute most important lesson of all here, never be afraid to edit your HIGHPOLY to fix bake issues, people so often think everything that could go wrong is because the low, and will refuse to touch the high after its "done".
Did you float the hole and forget to add it to your projection modifier?
e: no, you didn't, it just looked that way in the second image you posted.
I learnt loads of stuff just by experimenting with simple shapes and bevels etc, and looked to see what baked well and what didn't. Experiment and explore things by yourself, you'll have a much greater sense of achievement having figured it out by yourself.
@PoogoP and perna, yes I agree that is what I do. The vast majority of what I know so far is experimenting on my own or googling things when I get stuck. Sorry I gave off that impression I am not trying to have you guys spoon feed me every answer. If you struggle through something to learn it, you will come out on the other side with a much better understanding of what factors are at play than if you are just told to XY "just because".
Thanks again perna and EQ, I didn't mean to rapid fire questions like that. I didn't mean to aggravate anyone, my apologies.
Here was the fix anyways. It wasnt so much the bevel amount that was the problem, it was how far the extrude was going back the rays were not picking it up. Even though by my own judgement I thought the extrude back was relatively small (probably went back 2/10 the length of the cylinder), I guess not small enough.
Heres a fixed copy of the AO with the result I was looking for.
EDIT: With black of course.
I really have no Idea what you're saying about SG stuff, you should post some examples. Chances are if adding SG fixed skewing type details with your bakes, your bakes weren't set up properly in the first place, and your SGs also caused nasty seams from missed detail. Again, with a cage/averaged projection mesh, your lowpoly mesh normals are ignored, so it has no effect on ray direction, the only thing it would help with is smoothing errors. I put a lot of time into writing a big ass thread about this, you may want to give it a read. =P
Do I need to worry about making other objects have solid black in the AO. Such as the indents on the grip (shown below)? Or should the default of whatever the lighttracer produce be sufficient and the silencer was a special case with the black and AO. Trying to understand what circumstances warrant this solid black AO approach.
I hope I've understood you properly.
Yes that was my understanding, only a few key pieces need to be pure black. It makes sense to have solid black at the end of the gun barrel because it fakes the extrusion of the barrel without having to provide the geometry that no one will ever see.
As for the many indents around the silencer, they are not that deep, so I am not certain why perna is suggesting they should be black.
Right, thats why I was scratching my head when you were telling me to make the indents on the silencer black. Its a mix between a few different styles but this is basically what the indents on the silencer look like. I will remove the black mask on my previous post for the AO. Thanks
Also just found the original references for the silencer. Its a mix between the two. The indents in it are relatively deep compared to the reference above, thats why they are modeled that way. Mine may still be a bit too deep however.
http://wiki.polycount.com/NormalMap#Triangulating
On the right track? Still have to make the LP of the mag, front tac rails, and the laser sight.
edit: also thanks for the insight EQ, appreciate it.
I suppose it might be difficult or inaccurate to paint out some wacky ridiculous holes but for the barrel of a gun or something similar it's usually dirt simple. I've literally never had it take more than a minute and I'm not sure I save any time versus selecting geo/making black geo to fill holes.
Obviously I'm baking a diffuse material mask (I'd hope everyone does this by this stage in the game) but it just seems unnecessary to mix in fake lighting in the bake rather than painting it in unless you have some seriously weirdly shaped holes.
Image dump incoming
Overview
Body
Nitro Tac
Misc Items
Mag
Laser sight and tac rail
Container thing? I dont know what this is
Front Grip
I'll try to get a reply to your latest post soon, just wanted to make a note here.
Your biggest problem is just, you have no idea where the detail should go. 296 tris on that thing? WTF, it should be like 20 tris, let your normals do the work! The lasersight is also way excessive.
Rather than post every tweak I made, I'd thought I would just post the areas that had major problems. Still have tweaking to do for sure, need to spend more time on the body and handle definitely, but its a step in the right direction compared to where I was I think.
Also thinking about removing some of the loops at the bottom of the second grip (the one that doesnt have the saftey or trigger).
Your rail, i would use floaters instead of keeping it as an solid mesh. It'll save allot of triangles! But you'll loose some texture space.
You still got some vertices that doesn't help the shape:
Do you mean apply symmetry, and collapse the stack to make it one object? e.g. on the handle and body
Sorry I meant symmetry, I know they are not interchangeable my apologies.
Part of the reason I was hesitant to remove the loops around the grips is because my first pass at the low poly body was so low on tri count you guys told me to up it. So I had been focused on preserving the tri count and seemingly ignoring topology and edge density. But the layout of the edges was terrible, and most of those edges were not even meaningful after taking a closer look at it.
Anyways tried to adjust the curve density especially around the body grips, trying to keep only what I need for the shape. I have your notes tacked up on my wall to study, thanks again for providing them. Thanks for being patient with me, I have a better understanding of what needs to be done now.
*note, on the handle (below) there are a few separate objects so geometry may not appear to line up in some areas.
total count with symmetry applied to everything that needs it so far is around 4.8k. One of the rails needs to be copied around 2 more times which is around 300 tris at the moment I believe, as well as the center tac rail for the silencer.
EDIT Update : I think the LP is getting there, the whole thing with every piece modeled is now complete. Im sure there are some optimizations I can go in and look for though.
Not saying its a bad thing! :P
[ame]
Still, its like someone took a futuristic, lightweight gun that was well balanced, short, and maneuverable and dicked around till you can barely hold it anymore.
Good job with fixing the low poly, there are a few more things I would change, but for now it is looking good enough.
Doing some test bakes but I am having issues in some areas such as the handle in 3DS max 2011. The bake results for the handle show multiple artifacts and a very low appearing resolution even though I am using a 2048x2048. Using 3Point Shader
Handle Bake Test
RayMiss Check
LP + cage
The Normal Map was rendered with these settings:
2048x2048 normal map
Renderer: Antialiasing enabled, Filtermaps Enabled using Mitchell-Netravali. Global Supersampling Enabled using Hammersley
Advanced Lighting: Off
The AO map was rendered with these settings:
2048x2048 Lightingmap
Renderer: Antialisaing Enabled, filtermaps disabled.
Advanced Lighting: Light tracer active with 200 ray/samples and 0 bounces
is this a problem with my UV seams possibly?
Yes the handle needs to be reduced further its roughly the same amount of tris as the body at the moment. Yes I do use the symmetry modifier, I have just been posting the tri count for half the mesh up until now, I will no longer do this. The 6800 for the full LP is accurate, with all parts instanced and symmetry on. About working with the instances copied around, point taken. What I had been doing is just keeping track of it in my head, knowing what the tri count for the tac guard was, but it makes more sense to go ahead and instance parts such as this.
Yes, I have a far better understanding of what is required of the LP now and what you can get away with when baking normals to fake the detail. Major credit here goes to your paint overs they really helped me see what was going on, so thank you again. And no I am not just blindly copying the paintovers, that teaches you nothing. I am trying to take the time to analyze how and why. e.g. Every edge I try to look at it and think, "What purpose does this serve" it does serve a purpose "can the topology with it be made more effecient" etc. Things like that.
Thanks again
I just wanted to do a test bake to see if I had any artifact issues, and I do, much like I did with the silencer. So I am trying to troubleshoot whats causing these problems with the test bake of the handle at the moment.I know I am not ready to bake the whole thing yet. But from the small test I did, I am not sure why its producing such errors. For the UV space for the handle its not optimized to take up all of the uv space, but I scaled it up so it takes up a good portion. Perhaps taking off some of those filters may fix the problems with the bake.
For AO its light tracer with 200 ray/samples and 0 bounces with advanced lighting turned on.
Done.And changed to recommended render settings for the test bakes, additionally scaled down the UV space for the test handle bake to a more appropriate size.
Errors are very pronounced, not sure what might be the issue. Going to fiddle around with settings and see what happens. Heres a screen shot, using the same LP and cage as before. The normal definitely does not look correct
Result
RTT settings 1
RTT Settings 2
Normal Map Settings 1
Normal Map Settings 2
AO Settings 1
AO settings 2
Projection settings
Thank you very much for taking the time to look at this. Perhaps you will spot what is going awry.
http://dl.dropbox.com/u/25806698/HP_LP_Handle.zip
As for why its not baking, might this be some sort of Max 2011 bug, because I am still unable to get this particular part to bake at all. I would settle on just getting it to bake with all those errors for now, rather than not bake at all.
Since the HP and LP handle are not designed well for baking, what would would you suggest for a bake friendly version? Making more of these smaller parts/ intersecting peices one mesh like you talked about earlier, such as this handle?
You'd be amazed the number of problems that simply making your meshes match up more closely can fix.
By "look better" I mean how normals look. Don't know if its maxs fault (considering it has issues with displaying normals) but usually meshes with quads are looking better with normal maps. Maybe its my lack of experience but some times I changed triangles to quads in low poly, rebaked and usually normals looked better.
I removed the flashlight and laser sight, for two reasons.
A. The laser sight HP was no good (as shown by perna) for LP creation, I would have to redo a good chunk of it.
B. This may be going into a mod and the flashlight/laser can be obstructive in FPV.
@ Perna Thanks for another critique and for all your contributions and time in this thread they have been most helpful. Please don't be tempted to do any more paintovers, I don't want to be a nuisance.
Sure. My reasoning behind less resolution for that part was that its obstructed completely in a FPV. I figured the resolution it had would be fine when viewed from other players. Will definitely need to scale up the part on the far right though, the density is pretty low (the object with the huge numbers).
You need to straighten up some of these lines, soem pieces look rotated at slight angles. Its going to take more pixels to represent a shape rotated 2 degrees than a shape that runs with straight lines. Remember pixels are square, to represent a line you only need one row of pixels. To represent an angle you need 2, so your textures bakes will look lower res than they should.
You need to be much more mindful of how you're packing everything together, you need to leave a lot more padding room, or else your texture will blur together when mipping, some uv elements are so close that you will get bake errors as well.
Your uv density is all out of whack. Many areas that are seen close up have half the density of areas that are farthest from view. You should find a script to unify your uv density, then go in and scale up all the stuff that is clear in FPV, then repack it all.
Before laying out my final uvs, I like to do a quick unwrap, and then a dirty auto-pack and then testbake. This will help you nail down any geometry related baking issues before you finalize your uv layout, which is a good thing.
Oh one other thing, there is a round half circle shape right in your face that has half the # of sides as the barrel, which is further away from view. This makes the gun look lame and lowpoly. Never have lame blocky stuff right in your face, it will always look bad. Always reduce geometry from an off screen element if you are over your limit and need to add it where it counts.