Dwarven Architecture

Update 06/24/16:

After two weeks, it became increasingly clear to me that moving toward a rich district was not the right thing to do. So I’ve scrapped the idea of an Indian/Scandinavian style - not to say it will never get done, but it definitely won’t be part of this pack.

However, I’ve been hard at work polishing everything I could on the project and I am so happy with the results! Because of this, I will be able to submit to the Marketplace next week. This mean that I will create a new thread in the dedicated subforum - and as promised, ask what’s your take on the price - early next week.

Some stats for those interested: In the previous beautyshots, there were no dynamic shadows or GI because of FPS issues. Now, there is. The game run always at 30 FPS in 1080p, never dipping below even in the most demanding moments and averaging 50 FPS. Removing the dynamic shadows moves the FPS roughly 15 FPS higher, with the lowest drop at about 38~39 FPS. Now these stats are taken in-engine, so the FPS would actually be higher in runtime. I still have to do some test if I can pull it lower, but the whole level can actually run with 1500mb of GPU memory. For comparison, AAA games run usually well over 2k on PC (Crysis 3 is at roughly 2500, while games like Dying Light are at 3.9k) - on ultra, that is.

Some things to take in consideration is that I have huge open areas in the presentation level. Someone who has more knowledge than I with the engine could easily pick lights to deactivate them after some distance - or with smaller areas, the Cull volume works wonder to boost the FPS. I’ll still work a bit on optimizing the level, but in term of props everything is done (big architecture has several LODs and medium/big props too).

While waiting for the new thread next week, here’s the final official update on this thread, with sweet beauty shots:

http://i.imgur.com/hXhqpvk.jpg

http://i.imgur.com/80zLZ1l.jpg

http://i.imgur.com/NIZr65J.jpg

http://i.imgur.com/tLUC7KE.jpg

http://i.imgur.com/uCj29SP.jpg

http://i.imgur.com/AcRA4Dx.jpg

http://i.imgur.com/jfYdPWG.jpg

http://i.imgur.com/zYuK5iA.jpg

http://i.imgur.com/sWtXNVd.jpg

http://i.imgur.com/OkPLRUZ.jpg


Update 06/17/16:

As I’ve said, I took a week off to refresh my mind and start with a clear view of what I wanted to do with the Rich area. It worked in a way, as I was able to start with a different, yet complimentary view of how I wanted the rich area to be. In another way, it didn’t quite work as I expected - I spent most of this week iterating, scrapping and restarting modules, materials and layout. What I show today is advanced, but not final - for instance, the yellow/blue painting under the arc will change. The drapes may - or may not - be there in the final iteration. I am not satisfied by the visual quality I have right now with them. I will also add several smaller pillars, to add depth to the modelling. However, the rough shape/layout is done for the entrance and close to final.

Pictures (lighting hasn’t been done yet, it’s only reflection capture at this point):

http://i.imgur.com/mW9nkyQ.jpg

http://i.imgur.com/w4RVtpy.jpg

http://i.imgur.com/kKovchh.jpg

http://i.imgur.com/0HCJVRo.jpg

As per usual, comments/cheers/threats/encouragements are welcome!

Wolf1


WARNING: Huge update this week, with a ton of 1080p pictures (06/03/16):

So after last week, I was able to focus on bringing life to the basic layout of the market area. I’ll be honest; I’ve scratched my head more than once about how to proceed. At some point, the lighting would simply not render because there was just so many props. I had to switch to dynamic lighting (considering how wide the area is, it’s surprising it did not happen before that) and was able to achieve something I consider more than respectable with little loss in term of graphics/performances.

I’ve also made quite a few props which should allow people interested in the pack to have whatever basic items required to bring that area to life. Now, because these props are generic, you will probably want to use them as a complement to another pack, but they can stand their ground by themselves just fine. Since they use generic materials used mostly on the architecture, the cost in term of materials is low.

Let’s begin with the beauty shots (taken with the High-Resolution Screenshot tool with a 4k screen, I had always over 30 FPS in full-screen with a GTX 980, which moved over 50 in 1080p) :

http://i.imgur.com/VU3DV2S.jpg

http://i.imgur.com/upTmm85.jpg

http://i.imgur.com/oDrPv03.jpg

http://i.imgur.com/ymqgebw.jpg

http://i.imgur.com/6cHW7iB.jpg

http://i.imgur.com/TtBEY32.jpg

http://i.imgur.com/OM52K7n.jpg

I’ve also made a showroom map where all the architecture/props are put in categories. This will allow users to have a feel of each distinct district along with a big picture:

http://i.imgur.com/ivVNvE8.png

http://i.imgur.com/AQuSlzb.png

http://i.imgur.com/C6j7iGD.png

http://i.imgur.com/vPuQVcy.png

http://i.imgur.com/nADTNTj.png

Now the interesting part in this is that since everything is modular - beside houses and their interior - you can come up with virtually any layout that you want. I’ve made one that was somewhat open, but we could easily have a tight area mainly consisting of houses and not bigger than 6m-wide, or we could have a full-fledged city several miles long.

Because every single material is PBR, this also mean that the set should - in theory - be easily mixed with other packs on the marketplace, allowing even more freedom in how you want to use it.

In any case, I’m interested with your feedback as, in the end, you guys will be the ones to use this :slight_smile:


UPDATE 05/27/16

It’s time for the weekly update! I’ve been mainly working on a new district - the market area - so I’ve been almost exclusively prototyping. Production only “really” started at the end of the week. I also took this opportunity to optimize some textures - all of them were in 4k initially, but it soon became problematic. I’ve almost reduced all of them to 2k without any impact whatsoever on the visuals.

Here are some new WIPs:

http://i.imgur.com/5gAYW7v.jpg

http://i.imgur.com/K794WfD.jpg

Ground texture (tillable):

http://i.imgur.com/6c39iTS.png

Wood Texture (trims that can be repeated horizontally):

http://i.imgur.com/OU2Q7yg.png

Now keep in mind these does not represent the final product of that area. There will be a lot more stuff in that area. Next week should be more representative of the scope!

As usual, comments/feedback welcome :slight_smile:

UPDATE 05/20/16

I don’t usually post updates every two days, but I believe that the map literally exploded in this small amount of time. Here are some new WIP:

http://i.imgur.com/5iSZWF4.jpg

http://i.imgur.com/H3oZAFq.jpg

http://i.imgur.com/a9YUW7H.jpg

http://i.imgur.com/50rVZba.jpg

http://i.imgur.com/oxiZh5s.jpg

I plan to add three different districts (from which the bridge will connect instead of leading… well, nowhere), along with a huge gate which will connect the main bridge to the exterior. There won’t be any exterior architecture in this pack - I believe that it would require another pack and weeks of dedicated work and I believe that this pack itself will be more than enough for the time being.

There’s also quite a bit of tweaking to do, both in term of placement and materials, but ultimately I’m really happy with the shapes/ambiance/overall look. Currently, in 1080p, with everything activated in term of lighting and post-process, I’m at a steady 75+ FPS and never dip below that. Average is 100 FPS and top is 120. I have a GForce 980.

UPDATE 05/18/16

This week has been incredibly productive. I’ve basically spent the whole weekend on this along with the first half of the week. Things are going a lot faster than I expected, so I have a new screenshot ready as a WIP:

http://i.imgur.com/8I9SUSr.jpg

4k picture: http://i.imgur.com/HErxuSo.jpg

Now of course, there’s a lot to do. There’s some things I’m more or less satisfied in this picture, while some others looks really good. For instance, the bridge ground is probably final. But there are still some tweaks I want to do with the support pillars (mostly around the area where the player can go). The basalt cliffs also feel too low-poly, even though they aren’t, so there’s some tweaking to be made there.

And, of course, a whole lot of architecture remains to be done. Next step is to make a carved statue in the middle of the circular platform and then move to make a “main gate” where dwarves would come in and out of the area.

-----Original post below

Hey all,

I’ve started working this week on an architectural set for a Dwarven environment. So this week has been mainly focused on coining the general style and palette for the set - basically making a benchmark.

Here are some WIP pictures:

http://i.imgur.com/sRCjnWr.jpg

http://i.imgur.com/YjCZwRa.jpg

Now, this is really only a small part of the overall scope of the project. The goal would be to have a full, complete architectural set, including interior with different themes and exterior. Of course, this will depend on my velocity and if it’s a realistic expectation; but for now (and considering how well things went this week), it might be something achievable.

The lighting has been baked, but the results could probably be better with real time lighting (and faster!). The whole thing use 4k textures and developed with 3rd person in mind, so while the textures are high-res they might not be fit for a 1st person games. You will be able to play around, but also on top of the pillars, which are roughly 3x3m. It should be more than enough for any kind of gameplay, hopefully!

Please, do let me know what you think! I will update the thread weekly/bi-weekly as things progress. For now, I plan to be through the whole thing by mid-june. I’m incredibly happy to work on this project and I can’t wait to hear your feedback/opinions on the matter!

Well, I see one pillar and one horizontal piece made so far (used several times). The style does not remind me to dwarfs but rather arabic style as its pretty fine ornamented and not rough like you would expect from dwarfes.

Hey Voll,

Thanks for the feedback! I do agree that it does indeed feel a bit too fine visually; The goal would be to distance the set from your typical high-fantasy dwarf architecture and go for something a bit more refined. Think LOTR but more ornamented than say Warcraft or Witcher.

These structures should actually be used in interiors; as a matter of fact, these pieces that you see right now would be - in theory - put inside a cavern complex. They are fairly small right now, but they’ve been developed with modularity in mind. This mean that the pillars’ height can be altered as you want to. It would not be impossible to have impossibly long pillars that can go into the depth of the earth! Here are the modules separated:

I love the look. Very reminiscent of the Lord of the Rings with the Scandinavian influences. If you’re looking for any collaborators, I’d be interested just for the heck of it.

Ahh, yess, that shows the real modularity and its very good. from youre first pic i assumed the colum is one piece as its 3 times used exactly the same (similar for the horizontal use.)
Thats really a great approach as I can even imagine using all those modules vertically as well as horizontally and combining them in differnt variations.

If you add maybe some arches and 45 degree pieces than it would be a really nice set to create such architecture.

Do you work with one master-material that can be changed to create different looking surfaces, e.g. different colour combinations, other ornaments? If so one could even create different buildings or “clan/team” buildings in mathing designs out of the existing static meshes.

@nsanderson @Vollgaser Thank you both for the offer! However, I’m a single dev working independently. I’ve left the studio where I was working two weeks ago to regain my independence as it is something really important for me :slight_smile:

By the way Voll, your work is great!

On another note, I’ve updated my initial post to add a new WIP.

Massive update with tons of new screenshots!

very nice :slight_smile:
what’s your planned price?

Thank you!

It’s hard to say. There is still a lot of work to be done and a lot of architecture to create. Ultimately, with the scope I’m aiming at, it would be between 50 and 75 US. But when the time comes, I’ll make a specific thread to discuss the price with the community, so don’t take these numbers as final.

These look awesome!

Please, never stop doing stuff like this. it looks stunning! thanks for keeping us up to date.

Weekly update on main post :slight_smile:

Thanks to both of you for the kind words, it’s definitely motivating to see that type of support!

Thanks for the update! =D

It’s time for the weekly update. LOTS of new stuff, I’m really eager to hear what you guys think about it!

Awesome updates Wolven! This just looks better and better.

Another great update. Thanks a lot. was the 30FPS the case for running it in 4k? or were the screenshots only taken in 4K?

Also, will you be adding LOD’s? :smiley:

Yes! 30 FPS was for 4k (in fullscreen). Screenshots were taken much bigger than that, then rescaled in 1280x720. That was mainly to prevent the Temporal AA sharp edges.

There is, in fact, already some LODs included in the package! Right now, LODs follow roughly this pattern:

LOD0 - Full Resolution -> LOD1 ~30 to 50% of initial density -> LOD2 ~30% of LOD1 density.

I have yet to do a LOD pass on the props, but they’re not the biggest budget eater. I’ve deactivated the shadows on most lights, as they are mostly ambient rather than actual lights. The biggest budget eaters, right now, are the floor and the basalt cliffs from the first part. One of the good thing about these is that they have extremely efficient LODs.

But I’m at an impasse here, as both baked lighting versus dynamic lighting require two different configurations.

Baked lighting have visible popping if the LOD swap is too close. Since the light is baked, any slight shape modification will alter the final result. Now, this is not a problem since polygon counts aren’t an issue (when used moderately) with baked lighting - so we can actually make the LOD very far rather than moderately close. But when you swap to dynamic lighting, the polygon counts becomes more of an issue since every single polygon has to be calculated in real time, so you want to have as little polygons as possible. Basically, you need two different setup - one for baked lighting, one for dynamic lighting.

I’m not sure if I should decide to focus on one more than the other. Right now, my scene is compatible with both - small to medium sized baked lighting scene and medium to high-sized dynamic lighting scene works. In both case, extra-large scenes will have issues (either because of tech limitations or because of amount of triangles on the screen).

One possible solution would be to offer two types of these “budget-heavy” props - one for dynamic, one for static. But it has to be easy to use and handle in both cases.

Anyway, to answer, your initial question, yes there will be LODs. I aim to hit the 40~50 FPS in my scene in 4K while attaining a constant 60 FPS in 1080p while using dynamic lighting. I’ll keep everyone updated whenever there is a big breakthrough in term of performances :slight_smile:

As for this week, there will be no weekly update. Because the rich set is marginally different from everything else, I need to make proper research and gather as many references as possible. I’d really like to go for an Indian/Scandinavian architectural style, but don’t hold me to that as the end result could be really different depending on different iterations.

As usual, thanks both for your support!

Thank you soo much for such an in-depth answer. Makes a lot of sense.

Good to hear you are doing some research for the next bit. Looking forward to what comes next.

Thanks again. :smiley:

Wow, the detail you’ve added is amazing! Especially with the woodwork. The time and effort you put into this really shows; it looks awesome.

Ah very cool, reminds me of my less impressive version I made a few years back!