Still Trying to Finish ArtStation's T.L.O.K.A. Challenge
sublime tool
Hello everyone,
So I participated in the Legend of King Arthur challenge on ArtStation. However, I could not complete it on time because I had a lot going on those two months and chose a larger scale project than I probably should have. Oh well. But I am still plugging away at it since I really like the concept and would love to put this scene on my portfolio. Anyway, I decided to move my old and new work in progress updates over here now that the challenge is over.
So the original concept is "Merlin's Cave" by Erik Nykvist from Stockholm Sweden:
So I participated in the Legend of King Arthur challenge on ArtStation. However, I could not complete it on time because I had a lot going on those two months and chose a larger scale project than I probably should have. Oh well. But I am still plugging away at it since I really like the concept and would love to put this scene on my portfolio. Anyway, I decided to move my old and new work in progress updates over here now that the challenge is over.
So the original concept is "Merlin's Cave" by Erik Nykvist from Stockholm Sweden:
Replies
Lighting Test:
I have replaced almost all of the BSP and block out meshes. I reworked the lighting. I built in some things that were not in explicitly in the original concept like a sea entrance, a stairway entrance and a secondary stone circle. I should mention that I did not make the fire particle system. It is from UE4's starter content. I threw it in to see how it made the smaller stone circle feel.
I still feel like there is a long way to go, but I am determined to finish it (foliage, stalactites, boats, waterfalls, etc.) . I am open to critiques if anyone sees something that needs to be fixed.
Also, I could use a some opinions on what sort of base for the statue I should go with. Right now I am leaning toward a simple design, but I'm not sure.
Reference/inspiration:
If anyone has any suggestions or critiques, let me know.
The central island looks awesome, has a mystical feeling to it, fitting in line with the original source material!
Grass/dirt needs more work and geo detail (for a first pass it's cool)
BG ruins closest to us starting to look good but it's darker in the original. Depends how accurate you wanna be there but it amps up the focal point. Also pretty roughed up in the original
Foreground is gonna need more rubble.
I'm not an env artist, take this with a grain of salt.
It's definitely shaping up.
@pistachio Yep, those are all on the to-do list. Ha ha. I'm just doing things a bit out of order because lighting was starting to get fun. Thanks for the feedback though.
I have sculpted the candles and textured the statue. And I'm noticing some thing weird. My textures look great in substance painter, but not great in Unreal, and I'm curious if anyone has any ideas as to why. In Unreal it looked splotchier and noisier. I had to remove a couple of the grunge maps that looked really good on it in SP. And I noticed that some of the UV seams are showing up in unreal in the light map. I have a fairly dense light map resolution for it though. Maybe It would look better if I do a full 'production' build. Maybe Iray will always look better because of the way it renders things. Or maybe I'm just nit-picking and it's all in my head. I don't know. At least the candles look cool in unreal with the subsurface scattering, even though the flames are over exposed.
Substance Painter's Iray:
Unreal light build on 'high' with grunge maps removed:
The boat docks and rope all share one texture set that tiles horizontally in some parts like a trim texture. The Dock is also built in a modular system.
I also started working on foliage. I decided to pack everything except the tree into one material. Parts of this were created by modeling in Maya, baking in Mudbox and Substance Painter, painting in Photoshop and designing in designer. The workflow for each plant type ended up being a little different. Really based on what would be the easiest way to do each (like the easiest way to do the hanging moss was to paint a height-map in Photoshop, and the easiest way to do the moss was to model and bake) I did the final compositing in Substance designer so I could see how all of the maps looked together.
I think at this point All I have left is the tree, and a little more polishing. First though, I have to decide if I should try to build the tree from scratch, or download the free trial of speed tree and try to learn how that all works.
Let me know if there is anything that needs fixing or improving.
First off this guy has a great volumetric fog presentation if you haven't seen it yet. https://www.youtube.com/watch?v=Xd7-rTzfmCo I pretty much followed his example for the fog. I even followed his break-neck-speed mini tutorial on particle clouds he does half-way into the video. I then played with all of the settings in a trial and error kind of way until I got what I wanted. You can kind of see the clouds behind and around the mountain in my scene.
The cavern is pretty much lit by 3 main lights. A sky light with an HDR from HDRI-skies.com plugged into it's cube map slot, and two over-sized spot lights mimicking the sun. I originally tried to get an actual directional light to be the sun, but it just wasn't working the way I wanted it to.
The smaller spot light is the sun (see image), the bigger one helps out the sky by adding more fill light (or indirect lighting, or bounce light) and giving me more control over the fill light.
In order for either of the spot lights to be visible I ended up having to change the intensity units to unit-less for both of them. Otherwise I would have to figure out however many billion lumens or candelas they would need to be to even reach the ground. Unit-less seemed to just fill the entire length minus the falloff. I set the Sun spotlight falloff to almost nothing, since sunlight's normal falloff is like out past Pluto. I wanted a lot more falloff for the Fill spotlight so I could mimic bounce light getting weaker the further into the cave it got.
I then turned the indirect lighting intensity way up to make the light bounces a lot stronger and effectively fill the entire cavern with a soft ambient light. Then I turned the indirect lighting intensity down again when I realized how much the 'Preview' bakes were miss-representing that particular setting's effect compared to the 'Production' level bake.
I then changed the volumetric scattering intensity to control how much light would scatter through the volumetric fog. I wanted a lot for the Sun spot light to get cool light shafts, a lot for the sky light to fog up the ugly corners of the cave, and almost none for the fill spot light keep the docks and hillside fairly visible. I made the Sun spot light yellow, the sky light blue, and the fill spot light green which gave the fog some interesting color gradients. But I had to knock down the indirect lighting saturation because bounce colors were getting a bit out of hand on some objects.
Oh, I also had to turn the source radius and source soft radius on both spot lights way up (2200-2500) to get really soft shadows. Basically to the same diameter of the cave opening. Otherwise I would get 2 shadows from one "Sun". Ha ha.
Once I filled the cave up with the main lights, I added extra lights around to help match the concept (like the mountain's back-light) and add some things what I wanted for the scene (like the sea entrance, camp fire, and candle statues). The mountain back light is kind of interesting actually. It's two massive point lights with almost no light intensity (unit-less 0.1) but a ton of volumetric scatter intensity (38) and a falloff of 5. Those point lights don't really light the scene much, but they blow up the fog and particle clouds like crazy, and cast the mountain in a silhouette like in the concept.
Wow man, I cant thank you enough for that breakdown! Thats a really creative way of solving the issue of lighting such a big scene. I think my problem is that im relying too much on lighting everything with just a directional light and so im losing a lot of control over the overall lighting on the scene, so i will definitely have to try out your method of using spotlights with unitless intensity. I also recently discovered UE4's localized volumetrics tech and have been toying with it in another scene so I'll play around with that too to add some depth to my scene.
Again, thanks alot for the response, it's really a good example of thinking outside the box. 🙌
Here are the final screen-shots from UE4. I decided to add some extra movable lights for each camera to highlight and enhance each shot. The extra lights do things like boost the reflections in puddles, help visually separate objects from the background, increase the SSS in foliage and add volume to the fog. However what looks good for one camera looks terrible for others, so I had to turn them off/on for each camera. Ha ha. Totally wouldn't work for a playable level, but might be ok for cut-scenes or something. I also ended up having to completely re-bake the lighting with different settings for the henge at the top of the level for those last couple of screenshots. Filling the main cavern with enough light to look good a sea level blew out the lighting at the top. So much that even post process effects couldn't fix it.
But at any rate I got my screen shots to look the way I wanted.
While putting together these breakdowns I started to realize how inconsistent the poly densities are. Some assets are super optimized while others are pretty dense. This is for a few reasons. When I first started this project, (months ago) I went into it with the plan of not holding back so much. Spending polygons more generously to get better looking assets. As time went on though, I started to fall back into the habit of modeling things as low poly as I could. Also the assets that I sculpted first and retopologized in Mudbox were much higher poly than the assets I modeled Maya. Mudbox's auto re-topology tool gave some terrible results if I set the target polycount too low. So decided to leave things like the candle cluster at a . . . medium poly density, set up LOD's in Unreal, and called it good enough, so that I didn't have to spend tons of time hand retopologizing things. And sometimes I just forgot to keep an eye on the polycount because I was having fun modeling, like the large and medium ferns.
But now I'm seeing these poly totals side my side in a new way. like how the 6 inch candle cluster has a higher count than the 3.5 meter boat, I'm second guessing some things.
Now, on to the burning questions...
Thank You.
Feel like a "fly" through with some ambiance might go along way if you post anything to A.S.S (art-station site). (if you haven't already, i didn't look) as for some of your concerns i think you've done a decent job.
(i am no environment artist so idk what is good or bad there) but by the looks of a few of those counts you are correct in thinking some art too high compared to others(the rocks look lazy, and i can confirm after attempting rocks, oh boy!
This is a portfolio piece i hope that a viewer interested in your work would understand this immediately and if you do not do this in production then it should be fine, ('do not overload with triangles.').
Mentioning that you agree somethings could have been done better should be enough when speaking about the triangle counts. Just next time keep that in mind and or do what you do not want to do 1st or when you feel like it might help more than forcing yourself. It is what i do, when i feel like it i will when it comes to P.P's. (portfolio pieces XD )
If you have posted this already how was the response?
I wrapped up this project a few months ago, and already posted everything to Artstation, my website and Facebook. Before I finished it, I did reduce the poly count of the candles, the statue's face and the large and medium ferns by like 25% (I know it's not much, but it made me feel better at the time, ha ha). Yah, you're right the topology for rocks could have been a lot better, although I didn't want them to be too low poly, since in the environment they are about 4-8 meters tall. Maybe someday I will take the time to optimize everything better, but I have other things on my to-do list for now.
There wasn't much response when I posted it on Artstation, I got a few likes, but not a single comment so far. I got a few comments on Facebook, but that crowd doesn't know much about topology, so they were mostly just talking about how they liked moody lighting, ha ha. So far you are the first person to provide the advice and constructive criticism I was looking for back in January.
Regarding your polycount, it doesnt matter, as long as you dont use milions of polygons for really small objects. Its a portfolio piece, not an art test. I think you used polycounts in a decent way, using uniform distribution and adding more where was needed, and also in hero assets like that statue.
So, good job man!