Unique bake attempt.
3d App: MayaLT 2016.
Baking: Knald
Texturing: Photoshop CS4
Retop: 3d-Coat
Wips will be in Marmoset.
Concept by John Lane.
-Any feedback is appreciated as I push through this. (ongoing)
-Thoughts on setting base textures
-Approaches on creating wear layers with convexity map
-Probably need some explanation on floaters. (done)
-Need clarification on smoothing groups. (done)
-Do not totally understand what dilation does when baking. (done)
Side notes:
-Based on previous work. Some pieces will be left out of the bake and brought back and stitched into the lo-poly.
-Maya LT's polycount limit is 100k for export. I have been averaging around 85k when exporting. This is with 1 smooth iteration. I will try to get additional iterations if I can get it under 100k limit.
Replies
Will make adjustments on pistol grip and stock and then start baking.
Wireframe
Plan is to retop in 3d coat. Some of these pieces didn't bake off properly when reducing to lo-poly.
First pic is of a plan space and then the shape you want in the poly. Rather than having to cut it in get it nice and smooth you create the shape you want and then move it as close as possible to the target area without it touching. (Side view) When it renders it will think its like the final picture and bake in the depth to your normal map.
In critique of your high poly ill try be as useful as i would like to get back. Immediately because its in orthographic/Perspective it immediately kinda widens the pick. So id try and slim your version down slightly.
Red areas are detail you really should try and include in your HP either by floaters ^^^^^ like i showed above to help. Or by actually cutting it in (either by hand or by boolean) I mix it up dependant on the situation. These little details will break up the big blocks of geometry and kinda ground the piece.
Blue is something that you kinda seem to have missed. and especially moving down towards the handle in that area yours is a lot thicker.
Green definately needs some love. Its way thicker than the concept and the handle is making some wierd shapes/views on your image.
Overall its a real good base to just go back on and add in those details. I really dont envy you trying to get that grip right! Im having issues doing a scifi weapon of my own. So i know your feels!
Hope this all helps. Good Luck
The shots uploaded are
Hipoly-Basic Materials
Hipoly-wireframe
Hipoly-clay
Lopoly-clay
Lopoly-wireframe
Lopoly-w/normals and nonbaked pieces
On the last shot there are pieces withheld from the bake. A low poly was made by retopping in 3dcoat. Xnormals did not like the Uv's I did so I tried baking in knald. Knald does not pickup floaters on the grip so still need to figure that out.
You seem to be having a big problem with overlapping rays.
Retopo manually. Don't rush things and use 3d coat for retopo especially for hard surface, it can be fine for organic objects but this bad topology and bake result is evidence of why you shouldn't use that app for that.
With the 'lowpoly' you have currently you should really make sure all the outer edges are matching the highpoly as close as possible as thats the area that matters the most when baking a surface. All these parts like this can be removed. (I could have circled the whole thing honestly)
All you have to do is match the shapes and main curves not make sure every edge is connected to one another.
In what way does it not help?
I've explained why you are getting a bad bake, what topology is bad topology and how you should fix the issue.
Tldr version, Don't use 3d coat to retopo a hard surface model and remove edges from your high poly by hand if you don't have a low poly to work from. Only leave whats necessary. If you have all these unnecessary edges you will get a bad cage because the tangents are pointing all over the place which equals a bad bake.
Another tip would be to break the smoothing where you have UV splits, those of which should be on every edge over 45 that isn't a smooth transition.
The floaters issue is a different problem, and for all you know might be solved by making sure your cage actually projects correctly.
bottom is xnormals
Tested that and no success. It bakes very awkwardly. That is the fastest way to get the lo-poly.
It didn't work period. Pulled the loops and did the Uv's; it just doesn't look good.
I never implied magic here...Stop nitpicking what I say.
Show us what you've got and we can try and help you get a good uv map and bake. Honestly If it didn't work then you have some learning to do, please learn to accept critique gracefully, that will get you far in this industry. Theres no need to be defensive, we all have a lot to learn. Good luck.
I feel you got a great start here....but I feel your proportions need some work.
the front of the gun where you have the heat sink in the concept is flush with the body of the gun in yours it is higher (if that is the right word) also your gun if fatter and shorter than the concept
really look at the concept bring it into photoshop pay close attention to the shapes and how they relate to each other, high light and then figure out spacing and then execute the changes in your modeling package
its such a cool gun and I hope you nail it
-Gonna get back to doing the lo-poly.
-Do have some minor fixes to implement - Lengthen in the x axis by .100.
-correct smoothing issues due to bad edgeflow.
-Side Note: I previously stated some pieces will be left out of the bake and brought back in later. Hence why some pieces are not smoothed in the wireframe.
-Also note Maya LT's limit on poly export is 100k. So I cannot double smooth the piece entirely. I will dig a bit further and see if I can do 2 iterations in certain areas.
-Refer to this post: http://polycount.com/discussion/comment/2474957/#Comment_2474957
i don't know if I understand remaking 3d models from a 3d game but it's a cool gun
-So did a test bake today, decided to do the front barrel.
-Basically need to understand hard edges in a more practical example. Saw the cube example and really don't find it helpful. If you split off every shell because you used hard-edges; you end up with a UV-layout that hard to use for texture painting.
Need some clarification on that
While your proportions are alot better your shapes are still off
really look at the concept and try to match your shapes and how they relate to other shapes
I did a paint over to help out
Summed up:
90 degree edge + harden edge + Split UVs = normal map has less gradients (better results for generating cavity map) and less visible seams.
90 degree edge + harden edge + Sewn UVs = normal map will have seams in most cases because edges will bleed into each other.
90 degree edge + soften edge + Sewn UVs = heavy gradient banding in the normal map, very likely seams present on edges.
90 degree edge + soften edge + Split UVs = see above.
However, I would strongly suggest you take a good read through technical talk sticky threads about normal maps bakes and edges, this one in particular:
http://polycount.com/discussion/107196/youre-making-me-hard-making-sense-of-hard-edges-uvs-normal-maps-and-vertex-counts/p1
Get normals and UVs right and texture painting is going to be a breeze (technically wise) i say this because looking at the concept i see no heavy patterns that could produce heavy visible seams and cause major headache.
Also seconding what @Mossbros said, from what i see on your HP model there are mostly straight edges, but on LP there are some weird very soft shapes so its even harder for xNormal to generate accurate maps. Would definitely take the route @Ged suggested, make low poly by deleting edges from HP, that and you can easily generate cage for baking from maya (if MayaLT has option for baking normals, just select "show envelope" from dropdown and export that)
Cheers and happy baking
The front piece you showed here, has very low poly and you will have problems getting a good bake with it.
At the current engine polycount isn't that big limitation anymore, so a low poly vert-count of ~10k to 20k should be totally fine.
90 degree edge + harden edge + Sewn UVs = normal map will have seams in most cases because edges will bleed into each other.
90 degree edge + soften edge + Sewn UVs = heavy gradient banding in the normal map, very likely seams present on edges.
^
This is what I was talking about, but for a split uv you have a to cut and give padding. It just is not clear how to get the layout in a way that is still usable for texture painting.
I guess I'm just gonna have to post the UV's and bakes as I do it.
-Like I said not worrying about, can worry about it next piece. If that doesn't cut it oh well...sorry.
-could elaborate on the breaking up of shells part, because that isn't clear.
Anyways I will post the test bakes later on today.
I really don't think anyone can iterate more upon whats already been said for you, at this point it just seems most are wasting their time.
If you want people to elaborate on things for you, I think that you should really work on the way you respond as you are coming off as very passive agressive with every response.
kysg said: At least act like you meant it.
-Still a bit confused on it, tried scaling pieces in on barrel so I still had a readable layout for texture painting.
You can make your UV's more readable by adding secondary edge loops, this will allow you to make things a nicer layout for painting in photoshop, but without the supporting edge can lead to the normals displaying wrong.
Ben Bolton is pretty much a pro at this, and has shown in various examples how and why.
http://polycount.com/discussion/166058/ak-337-modular-rifle-system/p2
That being said, from what i can see in normal maps UV islands are at random angles if the edge is straight in 3D it's generally better to have it straight in 2D
But that isn't the main issue here, the main issue is, for starters you need to go and read up a bit on unwrapping and some general guidelines/rules related to the concept, mainly texel density, and cutting up UVs in a way that minimizes texture skewing, take a look at that and then go over this model again, i think its gonna make things more clearer.
And having a bunch of strips in 2d reduces your texel density but that is a whole different discussion.
-got some seams on doing matId bake from knald, not sure how to correct that.
-still got some issues in the bake overall.
-Is there a way to pull out dilation once your in photoshop?
in order:
-Low Poly with Normals and Base Color
-Low poly with Normals
-Low Poly
-Low Poly wireframe
-Hi Poly with placeholder mats
-Hi Poly wireframe
-Base Color texture
-Normals
-UV layout
And regarding bake, there is some skewing going on with bits on front grip and trigger handle have square bits a little skewed i would say its because of low poly and normal rays, and i think there are some additional squares on LP not present in HP on the stock in the back, farthest tube has normals flipped or no normals... from what i can hunt down scrolling up and down
Apart from missing/extra details in normal map you should go and fix the UVs since i bet they are causing this, also the UV layout looks like it either has some major inconsistencies in texel density or missing parts because i cant believe all those polygons managed to fit in like that ... may be wrong here but definitely look into better unwrapping.
Cheers.
Seams on color will have to be hand painted. I'll just manually pack the Uv's, auto-packer removes control so will just have to do it until I get it.
Texel density will not be same across the whole mesh.