Done with the Pond

Haven’t updated for two weeks since I was writing the dissertation and, creating new model & animation assets that don’t really deserve a post..

Now everything in the pond part is finished and I grabbed a video like I did for the circuit here. There’s also one controlled by scripts here and this time it’s an interaction:) Some environmental particle system details as well: sun shafts, falling leaves, some shining “dust”, and bubbles – exist in the painting I think they should be bubbles but who knows….

One Problem of the video is the blur on the right of the screen. I always get this in game mode if I have camera effects (even the build file has this too), and, this was tested to be the problem of my mac because others’ don’t have. No idea how to solve it at the moment..

And the third person controller model is changed now! Made some animations for him cos I thought I use different mesh and rigs with the Unity Ethan but, the fact is that it’s just super easy with no use to make any animation. Just import the humanoid model and do the configuration. Brilliant.

Found an easier way for couple animations, that I can export them in one fbx file, drag into the scene and delete one of them, and give the left one a clip with the original avatar. This would break the prefab but doesn’t really matter and it’s a lot easier to control the position etc.

Very proficient with the modelling and texturing now feel a bit sorry for those made at the very first lol.

Come the gifs!

The interaction is this hugging couple controlled by a boolean. When the player comes into a distance they’ll stop to be as the interrupted  pose in the painting, and when the player goes they’ll continue what they were doing 😛

Less than two weeks left dunno if I could finish everything… Go back to work!!

More Humans Start moving in

Humans are moving into the garden to make it (because it is) a better place -.-

I’m making all those human moves these days. Basically several clips that are in the same crowd or sharing the similar poses in one fbx file. Like I now have 10 clips in the ‘sit’ model, 11 clips in the ‘standing’ model, and 4 clips in the ‘carrying’ model. Although they use the same armature, since in the Unity scene they would have the identical pose which is the first frame of the first clip in NLA editor, then it would be very hard to adjust their transformations and parent-child relationships if I just used one model.

Making armature animations is fun. (The disgusting part is weight painting though – since I had more poses than just riding now, I’d found the weights were not that right at the hip part and it took me a whole afternoon to repaint it, to ‘basically right’. See the upper parts of the thigh is still quite weirdly swelling) But if you want to make them moving naturally, then you have to take care of every bone. Walking is obviously much more difficult than sitting, and like how you walk, every bone have to move to make the whole body move naturally, even the toe. Also the centre of gravity is important! The GIF below was initially for carrying a burden, but for quite a long time I felt it had been more like dragging something. I think the most important part for the centre of gravity should be the hip bone (the cyan below) – position as well as rotation.

Below is a combination in the real scene of several animations. The branch is set parent to one hand, and to make it move but not rotate, I have to be very careful with the hand’s bone and set it to the same rotation of every key frame. That was not very easy because the hip the spine the chest and the arm rotates all the time to look natural!

And another combination: the same person with the left one above but this time he is dragging a frog.

This one below contains more parent-child relationships. Actually I quite like this kind of thing because although it seems more troublesome, but once you got clear who will be the child of whom, then after dragging them in, the whole animation would seem quite good because everything seems moving along with each other and that makes the natural sense.

And a sitting crowd here. For a crowd of people like this as well as those trying to catch the red ball above, the differences between each other are very important. People can’t move with the same rhythm and the same movements. So one thing is to make more clips with maybe just slight differences (like they are all sitting but some are supporting the body with one hand or two, some shaking legs or looking around, etc), and the other thing is to use different speed for each animator controller.

Another thing except from the animation when importing these humans is that new models and textures as most of them have their own stage properties… (That could be another ‘default’ factor of being natural cos it looks that everyone is really doing something..) And here are the big birds in the pond.

I really enjoy the time restoring these adorable creatures by painting new textures, and I try to make them more like birds by adding thin strokes to create a feather-like material but not a colourful plastic. But, well, it’s not a good habit to take screenshots because I always feel that I’ve saved it after I make the screenshot. How much should I pay for my stupidness before I can get a degree?

The Circuit is ON!! XD

Literally four whole days (and nights) working on the components including separating models, reallocating actions, resetting smooth shading and UV unwrap, painting textures, sorting and cleaning animations, dealing with the hierarchies, and the codes…..

Animation

So now I eventually became quite familiar with Blender’s animation system and figured out how to store multiple clips for one file to Unity. In NLA editor I drag all actions I need into one track and delete others so that they won’t show up again in Action editor (become “0”, precisely, and disappear when opened next time. Cos I’d been confused why they were still not “0” although I checked all objects and no one was using them). And also, Unity’s clip list become clean now. Note that you have to delete the current action of the armature to <No Action> or all the clips would play the current action.

Therefore as all humans are using the identical rig, I can then append my prefab one (shape keys & modifier problem fixed) into each model file. I also found that it was because the parent-child relationship things that when I imported the models into Unity the components were always messed up in transformation. It seems that if I had more than one objects have parent relationships with a single object/bone, then it wouldn’t work properly. Like if I had both the hair and eyes set to the head bone, they’d fly away in Unity but if only hair, and eyes set to the eyes’ bone (moving paralleled), then everything’s fine. Neither .blend or .fbx. The same problem in complex animations such as something was set to a bone of an armature, and it had it’s own armature deforms or something set to it’s bones at the same time, then… well that’s confusing because the original models of human on horse animation work in the blend file’s default take (human has its own animation and is parented to the horse’s back bone); but when I had more complex things, I wasted hours on trying to figure out where went wrong, and failed.

My workaround is just separating things into single clips. The mount in one blend file, the human (if multiple humans, store them into clips in one fbx file, cos fbx can have shape keys animations), and birds on the head or something held by hand etc… then in Unity’s hierarchy, unfold their components, place the human as the child of the mount’s back bone, and place the others as the child of what it should be. Not really troublesome and very safe.

Hierarchy

Allocate the animator controller to everyone who has animation. Then, each human(s) with the mount and the related objects is set to an empty game object. That game object carries the moving script and a box collider. There’s also a “circuit” game object being the parent of everything here and it’s at the centre of the pond, so everyone can have a local position around the pond which is easier for the script.

As there had been more and more objects in the circuit I decided to sort them into crowds, and actually they are in crowds in the painting as well. I used numbers as in the clock then I found there were exactly 12! Brilliant.

So in the code I call the transform.localPosition to get the angle and radius (it’s an ellipse though so z divides 0.7 :P). Then angle+=speed and localPosition goes to its cos and sin. And finally use a Quaternion.LookRotation(Vector 3).

So that’s about the important things this week, easier said than done! How painful it should be to learn something…. I love painting textures though! Although Blender always has awful unfilled seams.

OK it’s too much for gifs this time so here’s a extremely unclear screen grab!! Forgot to say, it’s still even not everything in the circuit actually T___T

 

A Makeup Artist Should Be Able to Make Contour to Not Only Human

But also horses…

And after I replaced several models today, I found that I can just append the fixed human model to the same place and simply delete the original one, and give the new armature the original action because they use exactly the same set of armature. Brilliant! However at the moment I can’t figure out how to export multiple actions to one fbx file so that I don’t need that many human models which should save great space; I’ve checked the “F”, but in Unity there is just one clip. Doesn’t affect the appearance though.

Also found a very nice resource of a unlit shader that casts shadows: https://gist.github.com/pigeon6/4237385 because the the light and (even diffuse) reflections are sometimes too strong and the texture can’t be seen clearly, while unlit texture won’t have shadow that’s too weird for a real material.

Keep working!!

The Architecture 5: Keep Modelling…

Come to update the latest progress of the project.

Now just finished the last two giant buildings on the top right of the painting. Seems quite exciting but still great many of the small pieces everywhere ohhhh hope I could finish all of them…

Anyway. Kind of getting bored with these models of the same kind of vivid colours and I don’t feel in this effect they look as good as the ‘sketches’ in Blender. By the way that inner organ like thing was really crazy to model…

But the screenshots in Blender in the unfinished editing scenes look very cool and technical, especially with the blank and missing-data shapes! I suddenly feel that it would be a great modern version related card product or something.

 

I can actually choose the Unlit texture shader in Unity so that it would look as flat as in Blender windows because they wouldn’t accept the lights and shadows from the environment. I’ve talked about this before and then felt it’s still better to have some more 3D ones. This is really contradictory maybe I just don’t get the good way to balance these knotty needs. Who knows.

A small ‘documentary’ here!

The Architecture 3: Painting Texture & Physics

I found I was being so stupid to paint the textures using just Photoshop based on the exported uv “blindly” before, it was so slow and painstaking and there were quite a lot flaws because of the non-contiguous seams. And I learned from the Night Cafe Project that he used Mudbox to paint the bitmap directly and turns out nice so I am considering to get one too.

But I used Blender’s texture painting this time for the new building. It’s much easier than use only Photoshop although the hardness of the brush is unchangeable and it turns out the vague edges which is quite unpleasant.

I also tried the unlit shader as used in the Night Cafe which creates an amazing painting-styled environment. It’s kind of a 2D-like 3D, but I have to say at this stage the flat colour is not really compatible with others and everything needs to be changed if I wanted them to fit. I’m going to change the skybox, but I have no idea if I can manage the terrain and plant. Also the style of Bosch is more realistic and not like Van Gogh’s, and there’ll still be reflective materials like glass and metal so, I’m not sure now but maybe it’s not suitable for mine. But it’s good to know this technique.

Anyway I’ve just finished the second big building and I tried more with the joint physics which was used on the hanged red ball on the above pink building.

This time I tried to make the branches waves so I gave it a try to separate them into pieces and add joint physics. So as how people make rope physics, the tip one is the child of the former one which is again the child of the one before it… But if I still use hinge joint then they’ll collapse like a hinge will do and a branch is not gonna be soft as that. So I tried fixed joint. It turned out nice to wave like a branch but the problem was it would soon stop even though I turned of the drag coefficient. And then my solution is to tie something transparent as hinge or add some constant force to the top ones, as well as add their mass.

For those grow upwards I unchecked the “use gravity”, just use hinge joints and add an upwards constant force to the top one, so it moves like a balloon – maybe they seem a bit too much “enchanting” for a branch though.. Just because fixed joints don’t wave so much and would stop soon even if I added force. There might be some solution to the fixed one though….

Hopefully it wouldn’t seem “so much” after the other moving elements (creatures) are added later.

The Project 3D Night Cafe

The Night Cafe

Today I came across a project on Made With Unity about a 3D restoration of Van Gogh’s Night Cafe by Mac Cauley. It was just surprisingly great to see that really somebody is also doing the same kind of thing at the moment, and has quite already achieved the kind of sense that the 3D restoration of a painting should be like in my very initial imagination.

For The Night Cafe, I wanted the viewer to experience the feeling of stepping into this back alley cafe the way Van Gogh might have. I also wanted to express that feeling of wonder when you see his paintings, of being swept up by his exuberant colors where every brush stroke feels imbued with his lust for life as much as with his inner struggles.

I’m also so thankful that the artist talked detailedly about the methods in this project which is of course very useful and inspiring to me dying for the shading and texturing! Although it’s like hey it’s impossible that no one is thinking to make something as you do, I now feel there’s the motivation again and the visible aim I can head towards. Anyway, people are fond of Van Gogh who has been getting all kinds of creative tributes, but not the other genius of painting!

The Architectures 2: Start with the Giant Building

Finally finished something again to post a blog! I was making one of the most complex buildings (I hope it will be the most, but…huh), and the texturing, occlusion / all kinds of maps of the complexity drove me even crazier than modelling, placing and the combination, which seems just alright from this not really perspective view of speciosity, while it did become like impossible when it came into 3D… and also don’t ever forget the terrain and humans to see if they will be able to fit in.

Anyway this completely pink squid building has now appeared in my scene. Seems I’m getting more familiar with drawing textures baking lightmaps setting materials things but, this is only one of the five equivalent complexities… so good luck and hurry up I can’t afford to spend as long as this for them…

There is one thing special I’ve done to this building. I just took the attempt to search for the rope physics and there are plenty of tutorials so this became something a bit more interesting in this scene so far… I hope the tree can wave too but I don’t want to bother with the tree system now and there is nothing in my assets looks similar to this.

Some other small things before I started with this pink building:

 

For things with decorative elements of multiple materials, I duplicate the model, make it just a little bigger and give it another fade / transparent shader or rendering mode, with a height map. Like the bump circles in the left pic and some metallic rust in the right. There is a subtle difference in reflections when you move around which looks quite nice.

And these are just simple rigid bodies so that the berries are rolling around, and when you walk there to kick them away.

The Architectures / Artefacts

(Now my plan is to use a third person controller in a nude human’s shape to go through this place. I haven’t made it and I’m just using the default one at the moment.)

This post is about the architectures or artefacts – not really sure what I should call them – that I have done so far.

The working process is I create everyone in Blender, export UV, paint in Photoshop according to how they look in the painting (not edit the painting to fit the UVs), and make a suitable material in Unity. Sometimes it would need to go back to Photoshop, change the patterns and reapply them again and again to reach the satisfactory effect.

There are always times that I hope I can write my own shaders when I find it difficult to get the effect I want, especially for the cutout/reflective things. Actually I don’t know well about those so many kinds of maps in the shader’s parameters so the learning process has been numerous experiments (well I think I should find some tutorials to go through these systematically). But so far things turn out not bad.

I didn’t want to bother with the plant branches and vines so I used cutout/fade shaders on some planes rotated to all angles to make sure they can be seen from everywhere. I hope they can wave like grass, but terrain grass waves like the whole thing moves (the root should be fixed) so I gave it up. The planes don’t seem as good as the real models though.

The glass ball has some cracks. But I didn’t find a proper setting for both the semitransparent  and the solid crack, so I duplicated the ball and allocated them two different shaders to compose like a single object.

And the orange stele, with the ball on it, has 400 pearls placed manually… A single texture would not have different materials, and the cutout texture would not look solid and reflective even with height map. I think there must be a better professional way to do it, and it’s like I just forced it very stupidly. It’s like a manually made luxury, isn’t it..