[UE4] Physically Based 3D Art: 2016
interpolator
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Edit: As of June 29th, 2016, this thread will no longer be updated with new art. I will be rehosting its contents outside the site (to be determined at a later date).
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Next month will mark 1 year since I made the first major push into stylized art and despite some pitfalls on the way, I have a better understanding now of what kind of art I want to make.
Right now, I've gone many sleepless nights putting in more effort and detail into some new props I'm working on, so I will be posting them as they come along.
Edit: As of June 29th, 2016, this thread will no longer be updated with new art. I will be rehosting its contents outside the site (to be determined at a later date).
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Next month will mark 1 year since I made the first major push into stylized art and despite some pitfalls on the way, I have a better understanding now of what kind of art I want to make.
Right now, I've gone many sleepless nights putting in more effort and detail into some new props I'm working on, so I will be posting them as they come along.
This discussion has been closed.
Replies
And actually, I don't quite see why a new thread is needed for this; since it's the same as your other thread, really.
But, I look forward to seeing what you have to show!
What does Physically Based 3D Art have to do with hand drawn art? When an artist draws something with perspective they don't call it 3D, neither do they draw a rock and call it physically based.
I look at your other thread, and it's a blend of sketches and experiments in stylized rendering, and that is fine. By using the wrong descriptive terminology, you will get unproductive comments, not to mention the people that might provide relevant feedback won't be interested in the thread based on the title alone.
For your own good.
I get a bit bored when I'm not thinking of cartoons.
At least that´s what I think. Your Sega looks good so far. Would be a shmae not to finish it.
===========
I got around to making a few more kitchen props and lighting them. I'll eventually get around to texturing them.
The props you have here look good, I am interested to see how you texture them, is it going to be similar to your last thread or are you going to do more traditional hand painting?
And WOW, did I get lucky. I just dropped my textures in and the image came out just how I imagined it. No post processing, no lighting tweeks, nothing. Everything was perfect.
This thread now makes much more sense then your last one. The last two assets have a stylized look.
The grain will almost always (in practice) run along the long axis of the piece of wood. Think like a telephone pole, how the grain runs up and down; and translate that into pieces of a chair (yes, it's true even when the wood is cut into boards). The reason is, the wood is more structurally solid this way. Look at real chairs, and look closely at the grain, and learn through observation.
Also, the front of the chair's back rest, and the back of the back rest, have grain running in entirely different directions. Which just isn't possible. And if the grain was running vertically, the back of the chair would break pretty quickly, I'd imagine.
- The flower pot could use some variety in leaf sizes/variety in height in adjacent to each other too.
Even though you've jumped off to another project again... I do have to be honest and say this stuff is at least a clear improvement than the stuff you were putting together before... Get your ass back on the Genesis and finish it soon. It's nice seeing some progress with you now at least. Keep at it Jordan.
Enjoy some cute hyper pink in the mean time.
some pointers to improve this.
- Firstly, I'd say this is more like a dining area not a kitchen... so correct the name I guess haha
- The table has these weird squiggly lines, I feel like they don't really do it much justice so maybe remove them?
- The room is really bleached out so drop the high intensity lighting, maybe work with some natural lighting instead.
- The white crevices in the floor tiles seem a little too thick, I'd definitely decrease the thickness on them OR... scale down the over all tiling so the tiles are smaller (scale down would be my guess)
- The blinds don't read as blinds very well, they looked like a wooden window bar frame at first so I'd spend more time defining that model (add some beaded chains to adjust it and a segment that helps it hang) just to sell functionality for realistic design. Also widen the flaps and try apply a slight SSS translucency to it.
- The room is empty, some wall clutter and floor clutter would be good... you're going with a stylized feel so why not implement some of that creativity you love so much from your cartoon mindset e.g. a stylized mouse hole, mouse trap, etc... something to populate the scene since right now it's just a handful of props.
- Again to aid the assembling of the wall... maybe add a wooden ledge to the bottom of the wall which divides the wall and the floor better.
That's all I have for now, keep rolling with it Jordan (Y) You're doing better.
In addition to what's been mentioned, I'd look at the shadow being cast by the chair on the rightmost wall - it sticks out in particular to me
Correct me if wrong but I'm assuming just increasing the lightmap resolution of your wall will help with that. Using modular walls instead of one big, welded single asset helps with lightmaps too I've read
I personally would also make the gaps between the tiles smaller and remove the squiggles on the table as I do not personally think they work as well as the wood you currently have on your chairs.
I agree with some of the points above, I think you need to tone the lighting down a bit and decrease those gaps in the floor.
So while it may be hard to see now, I plan on making more scenes that take on these ideas (for example, here are some initial drawings I did for the kitchen, as well some future ones).
I definitely wouldn't do this if all I could make was realism.
I have drawings that go faaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaar crazier and venture into Cubist or modernism. But that's for another time.
This was suppose to be a middle ground of taking an existing artstyle and combining it with slightly more realistic everyday objects. I didn't decide to go back to completely inventing my own objects since I remember complaints of it being unrecognizable or failing to look 3D.
Now that I have the technology, I can finally set out to finish the scene with how I had it planned in mind (i.e I wanted to have the cars completely reflect each other as they pass by).
Edit: Corrected colors and glass reflection.
It's up to you in the end, but having a big screenshot of each, with color being the only difference, feels unnecessary and will only make the thread unwieldy and slow loading in the long run. Plus it kind of feels like padding, like you want it to seem like more pieces than it is. I'd say post a big shot of one color, and then a compilation image with the rest of the colors. But again, it's up to you.
Toronto Transit Commission Street Car
I feel the level of graphics I'm aiming for is further ahead than what current game technology can actually support. I'll still use UE4 for other projects.
This is more about a realization I've reached. Because I've been using lightmaps, it demands a huge amount of memory. However, I don't believe the memory cost outweighs the actual visuals I want. There are other things like scene complexity, texture resolution, lighting accuracy etc that the amount of lightmaps can't solve without becoming monstrously huge.
Love the Cars, looking really good, keep it up!
Or do you simply mean that it's quicker to render a frame than it is to bake lighting for a scene? Well yes, yes it is. But that's all you get : a frame, rather than a scene you can choose to present in a pretty much infinite number of ways, including actual realtime, with all cameras and shit.
the best 3D artists in the world are generally using realtime to present their work. That includes highly stylised stuff. They're doing that for a reason, think about what that reason might be.
I'm also positive that those same 3D artists wouldn't agree that what you're aiming for couldn't be achieved in realtime. Certainly not judging by that last image you posted. Its a question of knowledge and ability within the toolset, and a clever eye that's able to seamlessly connect tech and aesthetics. At this stage, you can't say that it cannot be done, only that you can't yet do it. Not the same thing at all. It'd be extremely beneficial for you to practice so you can do it.
"Or do you simply mean that it's quicker to render a frame than it is to bake lighting for a scene? Well yes, yes it is. But that's all you get : a frame, rather than a scene you can choose to present in a pretty much infinite number of ways, including actual realtime, with all cameras and shit. "
It's something I definitely took into consideration and is one downside to this. However, lightmaps are static as well. Once it's made, you can't do anything about them without having to re-render again, which I now find worse. Game engines are suited for real time, yet why am I using a [non-real time] system to do it? I rather go full hog and just use an offline render that gives better quality then.
"the best 3D artists in the world are generally using realtime to present their work. That includes highly stylised stuff. They're doing that for a reason, think about what that reason might be. "
When it's suited for a game, sure. But I've decided I wanted to aim higher in which case, it can't be real time.
"I'm also positive that those same 3D artists wouldn't agree that what you're aiming for couldn't be achieved in realtime. Certainly not judging by that last image you posted."
The G-wagon is mostly reusing the same UE4 assets so it does look similar. However, it's doing a lot of things that the game engine would struggle with. I'm getting far better anti-aliasing and lighting precision that games can't do right now. I've also talked about scene complexity.
My next scene takes place outdoors which is a much harder thing to render than a small room like the kitchen was.
" It'd be extremely beneficial for you to practice so you can do it. "
I'm not abandoning UE4. In fact, I still have other props that are close to being completed with it. But I've always treated my cartoon projects different, so that's why I feel the change needs to happen.
Baking lighting is like rendering a scene from every angle. Therefore, yes, it can take longer, but the results are more flexible. You can rotate around it all, for the time of a single bake. Where, to rotate around an object using offline renderers, you have to wait for each frame. It quickly adds up. If you're doing a single still frame, offline may well be a good decision. But if you want multiple angles, or some animation, an engine will almost always be faster, even when baking lighting on high settings. And assuming you know how to do it well, it should look, more or less, just as good.
On other points :
engines like ue4 and unity will use a mix of static and dynamic objects, and are set up to support both while still making the realtime frame look good. That's on a basic level, but readily seen in many AAA titles. On a more advanced level, check out some of the realtime showpiece demos, particularly the Adam video released by unity recently. Cross reference that with your point about static objects. As I say, knowledge and skill with the toolset. Don't talk in absolutes about what can and can't be done, unless you're a tech director at the engine creator, chances are you'll be wrong.
if you're going to claim to be aiming higher than the Adam demo ... Well, looking forward to that.
"As I say, knowledge and skill with the toolset. Don't talk in absolutes about what can and can't be done, unless you're a tech director at the engine creator, chances are you'll be wrong. "
This is ignoring that game engines aren't infallible. You don't have to be a tech director to know games can't crunch a million polygons, or use ray traced lighting, if the technology is not there to do it.
But where my ambition changed is when I realized I didn't want to lightmap an entire outdoor scene, while still being unsatisfied with the current lighting or detail fidelity that current game technology can't offer that more powerful offline rendering can.
Your second point about "a million polys" and so on shows up some fundamental holes in your knowledge. You'll probably want to fill those in before making grand statements that others will be easily able to rip apart.
I mean,whatever, it's up to you. It's just while realtime rendering tech moves on apace with great leaps and bounds, and whole communities of artists work together to get the very best out of it, choosing to work with offline seems totally regressive. Also, making big choices for the right reasons, rather than some plucked from thin air twaddle, is an absolutely vital skill to learn.
As I said before, it's not just about lightmaps. Everything about games takes on lower quality compared to what several hours in an offline render can do. That's natural when games only have a couple teraflops to work with, compared to the thousands that gets pumped into every frame of CG.
What you refer to as "quality" is relative. Just as with any rendering if you are not competent to pull it off then it will looks bad regardless of if it is offline of real-time. Putting stuff into v-ray or whatever you are planning to use is not a magic machine that will automatically make your stuff look better than it does in Unreal 4. You are also forgetting that this is a game-art forum so of course real-time rendering is favored. Maybe you should take Robomegas question into consideration.