[FINISHED] Open Pokeball

aclund3

Hi everyone!  Here's a small personal project where I was experimenting with some different / new techniques.

The critical commentary in this community is second to none, so dig in and let me know what you think!  Anything you think.  I'm here to learn, so don't be afraid to get harsh if you need to, I'll understand

I ended up getting the polys a bit high, but I think I know how to reduce and optimize them if ever necessary.  Here's a link to the 3D model on Sketchfab if you'd like to fly around.

Modeled in 3ds Max, baked and textured in Substance Painter, and rendered with Sketchfab.

aclund3

Bump. Hoping more people can see and hopefully give some feedback

aanselmo

Overalll I think is well done, the only thing bothering me is this sharp corners from the high-poly bake, I assume. How did you made your uvs and you baked the details with what?

, 

Seal40k

I personal would love to see more small details on this piece

ActionDawg

One really glaring problem is that you've given performance killing microchamfers to every edge, and have interior screws modeled out to great detail, at the cost of having larger cylindrical/spherical shapes with noticeable polygon stepping. The level of detail you give to parts should be relative to their size, and microchamfers are something easily captured by a normal map.

Take this for instance:


This is just all sorts of strange and has no real consideration for making the best use of the total polygon count.

aclund3

Thanks for the feedback, everyone!  You are all too kind

@aanselmo  Those corners/edges are bothering me as well, and I wasn't able to understand why.  Possibly the lo-poly doesn't have enough geo in this area, or maybe the lo-poly model doesn't match the hi-poly mesh closely enough?  I baked it out in Substance Painter.  UVs were unwrapped in 3ds Max, not the best usage of available space in here, but I'm a bit unclear on if its more important to keep the texile density consistent, or to fill up the UV space.


UVW Render from 3ds Max


Here's how I split the UV shells/islands around the release button. It may have worked better to just flatten it all out rather than split everything on the edges.

I also have very few smoothing groups in here, not sure if that's an appropriate approach or not.

aclund3

@Seal40k  I'm not sure what kind of small details you mean...  If zooming in/flying around in the sketchfab model doesn't quite do enough, can you be a bit more specific and I'll grab some shots for you to take a look at  I really appreciate the interest!

aclund3

@somedoggy  Well put, and a good point about the LOD needing to be relative to the size of the component. I definitely got locked into detailing without pulling back enough to get the bigger picture.  I also will readily admit the poly count is too high for this model, and I struggled a bit with how to reel that in.  In the case of the microchamfers, I seemed to get better results with them and so added them on any 90-degree edge knowing that this would just be a one object type of project.  Wrong approach for a game asset to be sure.  I'll get in there and eliminate the microchamfers and rebake.

As to the last sentence, I'm not sure what the polys should look like in that screenshot (other than no microchamfers). Is there any way I could convince you to do a quick paintover with where you would expect to see edges?  It would help me immensely, and also I would love you forever, so there's that...

aanselmo

What I learned for hardsurface, is that you need to slip the uvs every 90º and I found that the marmoset is the best one for baking, you can paint skew on the mesh that might solve those errors

ActionDawg

@aanselmo, you don't have to split the your normals for 90 degree angles. It's all about knowing where you should and under what circumstances your asset is being used. I routinely don't do this for certain scenarios.

For example, an incredibly rough hero asset with tons of 90 degree angles could have much more aggressive smoothing than a shiny metallic hero asset.
Or if you have smallish stairstepping 90 degree angles, so repeated over and over, it will probably look just fine to leave it all as one smoothing group.

So taking into account the surface quality and the relationship between angles in the model you can get away with more. This of course also can be affected by mipmapping, LODs, and compression which will make you have to be more conservative

@aclund3, one thing I would do is more closely match the curve above the screws with the scews themselves. You add ~3 more edges to the screw for a 50% increase in area. In addition, the details there don't really read with any depth from the distance you've presented the model. I would make the entire area flat and either remove the screws, or integrate them into the surface with no screw hole.

To be very aggressive:

This could be the geometry for that area and in context nobody would ever know the difference.

aclund3

@somedoggy Thank you very much for your thoughts and the redline.  I will rework the lo-poly mesh to be more in line with your comments and rebake.  Great stuff!

aanselmo

somedoggy said:

@aanselmo, you don't have to split the your normals for 90 degree angles. It's all about knowing where you should and under what circumstances your asset is being used. I routinely don't do this for certain scenarios.

For example, an incredibly rough hero asset with tons of 90 degree angles could have much more aggressive smoothing than a shiny metallic hero asset.
Or if you have smallish stairstepping 90 degree angles, so repeated over and over, it will probably look just fine to leave it all as one smoothing group.

I am just saying what I learned so far, I worked on a gun and had problems like this and it worked well when I split the parts into 90º with smoothing groups from uvs. Before the gun I used to use a single smoothing group and baked the details from the High and most of the times worked. But thanks for the info I will keep in mind

aclund3

Alright, I'm back with an update!  Thanks all for the input, it was quite helpful

I heavily reallocated poly density, taking from smaller details (and killing most microchamfers) and adding to the larger components where poly stepping was super visible. I attempted to entirely remove the slots and screws from the lo-poly but ended up with baking issues that I don't quite know how to solve yet, so I settled on just reducing geo for those small components.

Also, I reworked the UVs to be a bit smarter, rebaked, and re-hand painted masks for materials in substance painter.  Aaaaand, I fixed the weirdness around the release button up top. Apparently I never checked that area to make sure the lo-poly matched the hi-poly as closely as possible. One of those things where you think you obviously already did that and so you don't check it... sigh.

So, yeah. Please let me know if there's any other glaring problems you see, but I think I'm pretty much done with this guy. 



I didn't really intend for this project be much of anything, but I've continued learning new things and am glad I spent the time  At some point I may revist and get a really good render in Marmoset or something else that's super awesome!