The Bi-Weekly Substance Challenge (Challenge 5 & 6 & MeetMat17)
interpolator
Hello everyone, welcome to the Bi-Weekly Substance Challenge.
How it works
- Every 2 weeks a new reference will be shared to be recreated, you are also encouraged to find your own reference, as long as it still fits the topic for that bi-weekly substance, for example if we share a mud reference, you can gather you own mud reference.
- You can be creative with the reference and make it look either realistic, stylized or hand-painted, it is up to you.
Guidelines
- Try and give constructive criticism, giving and receiving this should help each of us artist grow.
- This is a Substance Challenge, so stick with that software.
- You can post Wip's and finished materials here.
- Don't share work you've done in the pass, if you done a material before, try it again don't try pass old work off as something you did for this.
- Final material renders to be rendered in a real-time engine (Marmoset, UE4, Unity, Cryengine, Stingray, IRay is also accepted).
- This is a learning experience for all, don't be afraid to ask questions.
- Learn all that you can and do share what you also learn.
- Please be respectful to each and everyone.
Discord
We have a channel in the Allegorithmic Discord group where you can share WIP's and final images as well as discuss the challenge, Click here to go to it.
If you have any other questions in relations to any of the Allegorithmic's software, use the appropriate channels for those questions.
Resources
- Allegorithmic Youtube Playlists
- Substance Academy
- Rogelio Olguin Substance Designer Gnomon Tutorial
- Joshua Lynch Substance Designer Tutorials
- 80.lv Substance Designer Articles
Previous Challenges
Challenge 5 - Raise the roof! (Roof Shingles) || 1st April - 15th April.
Click the pictures for bigger resolutions.
Challenge 6 - Taking a swim (Sea Foam) || 16th April - 30th April.
Click the pictures for bigger resolutions.
MeetMat17 - Allegorithmic Contest || April 27th - May 11th.
Click here for more information.
The guys at Allegorithmic wanted me to hold off on the challenges for now and I think as a cool challenge it would be good to see people trying out Substance Painter also and trying out this new contest.
Now for those who may not be interested in doing this, you can either push the other challenges that have been done and make them better or if you never got the chance to do one, go back and give it a try. So this gives people time to catch up with the challenge.
I will be posting a new challenge between the 8th and 11th however if you want to wait.
Hope you guys are okay with this.
Replies
Here is what I have if I cheat using the curve node but it limits my options.
But I tried to just plug a Gradient Linear 2 in both and I don't have your "little black line"
(excuse some funky warping around the edge due to pom)
Based upon:
Crit welcom!
Changing the mipmap mode should work as N4amless suggests
@capone : Good start! Though if you want some disorder between the roof tiles I'd suggest that you start with one tile and use it as input in a tile generator node instead of generating every roof tile with one gradient.
@Dickie : Looking good! Though it doesn't look like your tiles are overlapping like they do in your reference, they're sort of just placed next to each other:
Heyo any of you guys have a solution to layering tile correctly in SD?
I've got close ish but it's a bit crappy/not quite there. Explanation of problem
gif of current solution in action
Here is mine, I think it's just about there as a standard generic shape. I don't know what all the weird lines are about, some kind of banding. I've tried blue in a few places but it just makes it worse.
update, fixed (hid) the corner issue
and as @Daniel_Swing pointed out the tiles didn't look like they overlapped
dl - https://drive.google.com/open?id=0B5hEIzkZgS6EUVl6Tm94SmVUVHc
@capone that's a lot of stepping you've got there, are you using a curve node? if so want to screen shot how you've produced those artefacts?
I'm going for classic Spanish shingles. Took me way too long to figure out how to get a good tapered shape. Still not too happy with it.
I'm having a lot of trouble with the edges of the slates - comparing what I got here to my prior image, you may be able to see that I have reversed the "fibers" in the slates, making it look like they're chipped away from the outside of the roof:
This heavily conflicts with the edges and makes them really ugly. So I'm trying to find a smart way to mask the slate-edges away from the slate-surface, so that I can blend two different tiles together. As of right now, I'm using Edge Detect, which I'm not too happy with at the moment:
Any tips?
First pass at setting up the heightmap. Will work on a couple more tile variations and get them to layer on each other better, then I will move onto secondary-read forms.
@capone : That's a shame
https://s-media-cache-ak0.pinimg.com/564x/e8/82/e4/e882e422a1c38a318cee55e23f2dda89.jpg
I would consider making a couple more tile variations with some more dramatic wear and tear and place those around rarely.
I think you could also make a couple more tile shapes that vary a fair amount from the ones you have, like they have been replaced years later.
I think your surface noise, is just too noisy right now, look at references I don't really see that texture represented in them.
Alot of the detail will come from the grim though!
Added much more disorder to the tiles and started to iterate on the secondary-read forms.
Not a fan of the "halo" of edges I am getting on some of the tiles tips. I need to re-jigger the mask there.
Thanks to @Dickie who got the tiles overlapping correctly. Crit welcome
@Quack! : thank you for your feedback!
I'll be dropping by tomorrow to give some feedback myself!
I spent some time figuring out how to make the main shapes of lichens. Zero details added yet, just basic forms.
Secondly I'm getting banding again (though perhaps this is also related to the low resolution?). It is 100% being cause somewhere in my selection box in the pic. If I hook that first blend direction into my outputs that is where the banding begins. The blend is just a copy, if I hook any of the inputs into the output there is no banding but combined there is a sign of banding. ALL transforms in the scene are set to manual mipmap mode (as opposed to default of auto).
Aha fixed, thanks. I wondered those values meant ever since updating to SD6. Curious, what's the general purpose of it? If you're just doing portfolio work I guess it's ok to just ensure they are all 16bit but if calculation times are vital you can go in there and enable 8bit on nodes that do not result in banding?
I'm still confused about the locked out size input at the top?
the parent size thing is only usable on relative to parent. But you're probably better off having it set to absolute and then setting the 'Output size' to be higher, try setting it to 1k and see if you still get banding
@Zombiecrackers : I think your tiles are a bit too thick and the perlin noise is a bit too intense.
@Craig_Rogers : You should make some variation between the tiles (I'm mostly thinking color ) and also look over the tiling in general; I can see the vertical seem very clearly.
@Quack! : Nice start on the lichens! I'd really like to take a look at that graph later! The lines and crevasses on some of your tiles makes me think you're going for a stylized approach but your edge wear says the opposite. It could be just me but I don't think they mix well.
@suspectlogic : Really like the cracks, though you should break them up so they don't overlap from tile to tile (I think someone posted a pixel processor in the first challenge that took care of this very effectively).
@suspectlogic your height information looks very clean! The overlapping looks cool too.
@AzzaMat Like that stylized look. Maybe the concrete underneath could be a bit less glossy?
Quick update from my side, worked on the color a bit. No work on the roughness yet.
Well i think you should separe your two tile (curved one and flat one ) on two tile generator, the problem you have is juste that when you rotate your shape you break the tile, and the tile make your height look good since it's on the middle.
If you separate both you can rotate both surface and offset the flat one so you hide the break under the curved surface. Hope i'm clear ^^ i can show you what i have in mind if i'm not clear.
And here is my take.
Thanks for the info. Attempt 5 is coming along better, trying to do something closer to what you have just for the purposes of learning rather then going off and trying to do something that nobody else is doing. Here as suggested I have 2 tiling generators, I've yet to offset/distort them etc but I did a quick test and it seems to be working with the shingles. The 'backtiles' are currently just long strips, I'll need to turn them into separate tiles if I want them to have slight rotations. Looks like you did the same from what I can see.
I think I have the big shapes where I want them to be (curve node was super useful). The challenge I'm working on now is figuring out how to get a good mask to change the color per tile.
I've noticed with some shingles you have a double overlap. Similar to what SuspectLogic had,
I like it but I've not seen reference of shingles that do this? What's the idea? There was a shingle, it broke and got replaced by just placing a new one on top of an old one? or it's a crack that goes straight across? I'm wanting to try it but don't quite understand what it's supposed to represent.
Happy weekend!!
In the past I've discovered two reasons why this happens,
1) In transform2D nodes were set to automatic (default) instead of manual. In most cases it doesn't matter but every now and again it will fix similar issues.
2) In the tiling generator there is a setting called Image Input Filtering, by default it's set to Bilnear when it should be set to Blinear+Mipmaps.
In my case below none of these are the issue it's something else. My slope blurs are set to full resolution (32), I've tried adding blur nodes but those don't help. I've ran out of ideas. Tempting just to move on but would rather only 10% complete a challenge and learn something than fully finish a challenge and continue making the same mistakes.