And on top. It does get harder to find the good stuff. Google search is favouring the info with the most clicks and not the best one. A lot of good tutorials and info is lost in the void. I really dont like this snack posting and fast clickbait wisdom. I search for something in know its there and all i get is "make awsome asset in 5min". Solution is to ask some Coworkers.
@LouisMarshall There's a few different topology strategies for sharpening those corners while also minimizing the visibility of any smoothing artifacts. Which approach makes the most sense will depend on how accurate the shapes need to be, how visible the area is and how much time was spent creating the rest of the model.
If the subtracted shape is relatively small or won't spend a lot of time in-front of the player then the easiest solution would be to use the existing geometry and adjust the topology routing so the adjacent edges create a support loop that flows around the outside of the shape intersection. After the topology flow is resolved it should be possible to select the edges of the shapes and bevel to add support loops. The bevels in the corner can be resolved to flattened diamond shaped quads that connect to the geometry of the underlying curve.
Vertices at the top of the inside corners may need to be moved downward to compensate for the smoothing stress when the subdivision is applied. Using the existing geometry of the curved surface as the outer support loop and using the transitional area between the loops to make up the difference between the shapes should prevent the corner support loops from running out into the adjacent shape and causing undesired deformation when subdivision is applied.
Here's an example of what this process and topology layout could look like.
If more shape accuracy is required then it may be necessary to increase the geometry density of the underlying curve before subtracting the shape. Adjust the number of segments in the curved surface until the corners of the intersecting geometry lands between a shared set of edges. These radial edges in the curve can then be used as support loops that tie into the corners of the subtracted shape.
Adjust the topology so the edges that run across the curved surface act as outer support loops for the shape intersection. Add the rest of the support loops around the shapes using a bevel operation and merge down any stray geometry to the outside support loop on the curved shape. Join through or add support loops as required. Any difference between the curved surface and the subtracted shape should be taken up by the outer support loop. This will help constrain any potential smoothing deformation to a smaller area. Which will help minimize any potential smoothing artifacts.
Here's an example of what this process and topology layout could look like.
In subdivision modeling, it's generally considered best practice to use the existing geometry of the intersecting shapes as support loops and to try and resolve most of the topology flow issues at the lowest possible level by matching the segments of adjacent shapes. Both of the topology strategies shown above take advantage of the existing geometry in the curve. The only major difference between these two strategies is whether the support loops for the corners need to end on the existing geometry of the curved surfaces or whether these support loops can run out into the existing geometry of the curved surface.
Here's a close up comparison of the topology layouts and the subdivision previews. Both topology strategies should be viable at intermediate view distances. The subtle difference in surface quality really isn't all that noticeable unless the player is viewing the model in an extreme close up. So whether or not it makes sense to add in all of that extra geometry really comes down to how the model will be used and whether the tradeoff in modeling efficiency is worth the slight loss of surface quality.
Recap:
@wirrexx thanks!
Started thinking about some cobwebs, tried playing with substance designer but wasn't really getting the 'old-tangled-cobweb' look that I wanted so yesterday I tried Houdini. After an hour or so I got something that I was pretty happy with.
This evening I baked a few variations to a plane in Painter and tested them out in Unreal. They need tidying up and placing with a bit more thought but for a first pass, I'm pretty pleased :p
Almost done with this, might make a few color tweaks. any feedback will be appreciated
https://www.artstation.com/artwork/X1dWQY
Game-Ready, Tom Hardy Likeness for an upcoming (nearly finished) fully fledged character. Heavily inspired by his James Delaney look on Taboo (as I feel it suits what's coming quite well). I did change the beard according to other references (PureRef) and what I feel looks best.
It was a great exercise to understand that It's a hard thing to make a likeness but it's an even harder one to average 100 different images to get to a neutral expression that can be rigged and animated.
Sculpted in Zbrush from scratch, textured and rendered in Marmoset Toolbag 4. Ornatrix to generate the hair for the cards, manual placement in Maya with the help of the amazing GSCurve Tool.
Hair is 106k tris, beard is 10k tris and eyebrows 3,8k tris. Optimized and game-ready.
There's also a bit of fan art for shits and giggles at the end of the post...so stick around!
A thank you, to my brother @nimlot26 Georgian for his patience, support and guidance. His Patreon mentorship: https://www.patreon.com/nimlot
Hey all! Glad to see that this thread is still running after all this time (if only somewhat) - it was one of the main sources I used early on when learning and it ignited my passion for low poly with pixel art texturing years ago. After all this time I thought I'd share my latest work to where it all started: a low poly Shotgun in what is my weird style mix of retro aesthetic with pbr. It is a little expensive (triangle-wise) at 996, but with all the moving parts I wanted for animating and posing (e.g. the stock folds up and the shells need to be taken from the side), it's fine. The count ignores the strap because it's designed to be deformable for physics and positioning, making it a little more poly heavy, but hey. Anyway, this was created with Blender, Aseprite, Marmoset Toolbag and Substance Painter, and the textures are obsessively pixel perfect; it's not the most efficient process, but still satisfying to see come together. If you contributed to this discussion in the early days and you're seeing this, thank you so much for the years of inspiration!
Hi guys, the final variant for the AR project is done.
Check my thread for more work in progress:
https://polycount.com/discussion/225845/finished-works-work-in-progress-weapons#latest
or my artstation for 4K res. Thanks!
https://www.artstation.com/artwork/d0LAa3