Day 56 Genesis

Day 55 Genesis

Today was the so-called the final day or render day. The tweaking must stop and proceed with the rendering. To really confirm the "best" look, I did flipbook of sequences with different attributes. This is to compare the attributes like which are better, which are more suitable. To wait for the cooking of the sim alone was tedious. The process goes on till I get the one with the maximum effect. I put in the flipbook sequences in After Effects to compare. With Mr Michael help, I got a better result.

 flipbook sequences in AE

There was some problem with the caching although the oversampling was set to 10. Mr Ron showed me a few method to play with the converting.

 different volume conversion

switch to test out different method

After that, I had to rop the bgeo sequences to save the calculation time when rendering. It took me about an hour to rop it out. Then I left it to render after I input in the lights and desired shader color.

Day 54 Genesis

As the final hours getting closer and closer, I am feeling the heat even though Mr Michael told us that we need not show the final renders. I wanted to work to the fullest to come out with the final renders to impress everyone. Hmmm..... we shall see.

For now, I am still tweaking. Opinion/suggestion from Lu Sheng by Mr Michael that there will be no end to tweaking, that's true but for a perfectionist like myself, I wanted to show the best, but come to think of it, there will be subjective views and " What is the best??? "

Despite those, I carry on tweaking. However, there was still some particles stuck at the emitter. To fix this, a fan node was place in position slightly behind the emitter to blow out all the stuck-particle. To get the position of that specific place, a vopsop to normalize the normals and a polyextrude to get the position.

polyextrude for the fan position

 expression to use the position to put fan

Day 53 Genesis

Over the weekends, I did some testing and lots of tweaking. It was a tedious process with all the cooking after every small adjustment of number but it was worth it. I converted the smoke to particle to see how was it like. I did a render to test. When I went to school and asked Mr Ron why does it looked like that and he told me that there was not enough sample rate of the popnet at the particle-fluid node which I used to do the conversion.

 test render

I then took this opportunity to refresh my memory of the flood system as I had to complete everything by this week. I kinda remember how it works, hope everything kicks in. I also did some test render and added motion blur to one of them, it looks decent and I quite like it. Hopes this works well.

 flood render without motion blur

flood render with motion blur

Monday goes by with lots of tweaking. As the emitter emits puffy smoke, I had to tweak other attributes as well. Temperature was one of them. As temperature is high, it acts like real-world physics where high temperature = smoky/steam, lower temperature = condense liquid. Went to work on this area but nothing seems to happen. Consulted Mr Ron, he suggested Viscosity. It kind of works now as it produces more "tube-ish" flow. Everything started to come together.

Day 52 Genesis

After much tweaking, the gush of the water was still kinda weird. The water still goes upwards no matter how I tweak. Went around the network to see what happen until I remember about the point normal in the water emitter. As expected, it was pointing upwards. A quick shift of node fix the problem.

 weird normal with xform at top

fixed normal with xform at bottom

Also there was something about the creation frame, as the dragon will only spit water at frame 115, I adjust the start frame on the DOP node and creation time on various nodes. However, it did not work for some strange reason. I posted question on the forums regarding this. Mr Ron's suggestion was instead of playing with frame time, why not play frame 115 at the first frame by using a time-shift node.

After tweaking, I tried to convert the smoke sim into particle form to start testing renders. There was weird movement and I had to solve it fast.


Some test renders I did.

Test Gush 3 from Tanner Caelum on Vimeo.

Day 51 Genesis

Mr Douglas came in for his final progress check before he leave for his oversea trip. He wanted to see the progress of the shots and check that everything is ok. I showed him that I have integrated in the gush system into the shot. However, there are a few problems where I had to solve like snapping smoke container and the direction of spit .etc. Now what I have left to do is to tweak for the best look and correct physics of spitting out the water. Spent quite long to tweak the parameter like the fastness of gush, the constant emission and so on.

Flip Book Test 1 from Tanner Caelum on Vimeo.

Flip Book Test 2 from Tanner Caelum on Vimeo.

Day 50 Genesis

The problem was fix using write file directly and use a rop geometry to rop out the sim and then bring in back using file node. So now I had to render out the bgeos whenever I tweak the slightest of attribute. I am able to come out with a test render. Afterwhich, I finish up the commercial version and did a little documentation on the hip file.

Test Gush 2 from Tanner Caelum on Vimeo.

I also receive the shots from Mr Douglas. I was wondering how can I start to put in the stuff. As Mr Ron help to fix the caching problem, he stay on to help me kick start the work. As he already started halfway, he tells me about Murphy's Law. Murphy's Law, states that "Anything that can go wrong, will go wrong"

So Mr Ron continues to guide me throughout. After filing in the exported geos of the dragon and the mountain, changing the emitter to a cone shape makes the spitting more natural and to drive out the normals and use them as velocity direction.

convert primitive to polygon 

polyextrude to cone

point node to add normals to add velocity

For the vector field, it was the same as how I did in the gushing system. This time, as the dragon is moving, I needed a way to "parent" the field. The emitter with all the movement/animation was brought in using a object merge node. A circle was put into a copy node with the emitter as to copy all its moving/position data to the circle. A polyextrude was added to create the other end of the cylinder with that as the position of the vector field. A sweep node for copying the spiral normal circle on the points of the backbone which is is the polyextruded out vector field. The rest of the step are similar to the previous system.

 polyextrude for vector field position

 sweep for arranging

network of vector field

For the dopnet, it is the same as the previous system except that needed to bring in the terrain as terrain object and sourceapply to source the smoke.

network of dopnet

Day 49 The Dawn of Age

I fell ill on sunday. It was quite serious. Wanted to go school but have not fully recover. Somehow it gotten slightly worst. Got to laid on bed for the whole day.

Day ??? The Dawn of Age

"What if after doing the commercial version and the problem still lingers? Maybe the current file got corrupted? Why does this problem happens now?"

All this questions swim throught my mind. I couldn't just sit till monday to continue the commercial version, so i decided to re-do on my computer. While re-doing, I encounter some questions, like "why this must use this". Almost finishing with the re-do, I was hoping that it will work and that the current file was corrupted. Just as I finish with the last node, I drag the pointer of the mouse towards the play button and click it. Here goes.

NOOOOOOOOO... Its the same thing. Why Why Why? Looks like i got to resign to fate.

However, this does not stop me for doing things. I continue for test renders but was slightly more tedious then usual.

I play with impulse birth rate keep the vorticle strength the same but tweak the magnitude

Although I did some test, I am not sure if it is accurate since that problem occur.

Day 48 The Dawn of Age

While doing test renders at home last night, I encounter a problem. There was something wrong with the playback. I tried all ways to solve the problem but nothing work. This morning, the very first thing I do was to call Mr Ron for help. As this problem was kinda serious to me as it slows down the speed and the progress of my work. When Mr Ron came, he quickly point out the problem. It was about the file out node as the cache produce weird result. After hearing that, WOW, seems like life ain't easy after all. So it seems like before I can do a test render, I need to do a manual cache check to see if the sim is the correct one. I was hoping that by re-doing the same system, it can solve the problem, as I need to convert that file into a commercial version anyway. Spend the day re-doing and at the same time understand it better. It was not long before the day ends.

Day 47 The Dawn of Age

Mr Douglas dropped by for the weekly routine progess check. Everyone's work was almost there in terms of "ready-ness". He was satisfy and impressed by us. I also showed him some test I did which only a few option was tweaked out of more possibility. After the progress check, he stay behind to block out the scene for our final shot. Here are some test I did with tweaking of vorticle strength. At first I tweak the magnitude in the vorticle. After some quick renders, I realised it did not change anything until I had confirm with Mr Ron if the vorticle strength in the smokesolver will affect the sim.

Here are some test.

Different magnitude but same vorticle strength

Kinda like the one with the render of vorticle magnitude 0.8, though.

Day 46 The Dawn of Age

I spend mostly the day doing test render. With many areas to tweak for nice and efficient results, I did a handful for tommorrow's review. I wanted to do a sequence but did not have enough time. There are many areas to tweak.

E.g.
> placement, size, number of, the speed of the rotation of the vector field.
> magnitude, radius and the number of vorticle seeds in Vorticle geometry.
> bouyancey, vorticle strength of the smoke solver.
> velocity for emitting speed under initial state and physical in rbdobject.
> impulse birth rate of the source in the popnet
> the shader properties for its look.

Here are some of the test render,

Decreasing step size while keeping Point Radius Scale

Decreasing step size with Point Radius Scale at 1

 Increasing Point Radius Scale while step size remain same

Trying out various numbers

Decreasing step size

Increasing Point Radius Scale while keeping step size at 0.05 but PRS number was too low for caculation hence the time taken was weird as supposingly the time taken will decrease if PRS increases.

Increas PRS while keeping step size at 0.02, same behaviour as above results

Day 45 The Dawn of Age

Before continue working and tweaking for a better look of the gush, I had a few clarification regarding the direction I need to head and some question about the system. In the system, I had talked about new nodes like the up-res smoke object. However, there was some problem that the up-res smoke is not working as somehow it cant read/find the low-res source file. I consulted Mr Ron but couldn't solve this as well.

up-res error

In the upressmokefield, the input in the default object will affect the visual in the viewport. E.g. in dopnet, if the sim is low-res, and in the visualisation its up-res it will not work. Some quick caching and placeholders for reading file knowledge was taught.

caching/file out

After understanding and clearing of my doubt, I also did a test render gush. A test render of 80 frames and it takes 2hr 30min to render out the whole thing. Here is the link of the test render in Vimeo -> http://vimeo.com/23531802 .

Or to see the video here,

Day 44 The Dawn of Age

As Mr Ron gave me the test system late in the afternoon, I needed time to go through it. Over the weekend, I spend awhile understanding. I did a "fake scenario" test where using NULL nodes as substitute for the real nodes. Using legends to mark out the befores and afters nodes after each "shelf-tool" clicks. I drew out the network which was quite messy at the first look. Sorted it out and begin to understand. However, certain areas or particular question which I don't understand I will also write it down to ask. I also did some test rendering. I had most of my questions clarify.

actual system

understanding test

test 1 -- 3 Mins 18 Secs

test 2 -- 7 Mins 42 Secs

Day 43 The Dawn of Age

Mr Ron was busy till the late afternoon so meanwhile, I play with the same system for awhile until Mr Ron came in with a test he did. Here is the link to his test video --> http://vimeo.com/23256903 , this is the one with motion blur but it takes quite long to render so Mr Ron remove the motion blur and play with colors to achieve another look --> http://vimeo.com/23256928. The test was kind of similar to the ones I am testing/playing with. A tube emitter, a simple network with 2 OTLs e.g. Tangent and attribute transfer, to create the vector field. In the dopnet, seeds vorticle was included in the usual smoke set-up, a sopgeo to bring in the vector field, gasparticletofield for applying the field but the new things about this setup is a resizing bounding container which will resize according to that particular frame/size of smoke. The other new thing is the up-res smoke where the user can chose which res to work with to give greater detail.

For the particle fluid look/characteristic, object merge in the smoke node as a object mask and using a popnet with source and a voppop inside, converted smoke to particle which allows basic liquid shader to be applied. Attribute delete was implemented to remove unwanted data so as to reduce the amount of bytes for faster and take up less memory. Mr Ron also tweak the blur, density, shadow tab in the basic liquid shader.

vector field

new system

new additional nodes

particle fluid

test

Day 42 The Dawn of Age

Mr Douglas dropped by for the usual "progress-checking" day. Showing him the vector-field driven water gush system and a few renders, Mr Douglas was fascinated and seems intrigue by it. With Mr Ron topping it up with something he wanted to show, a seed vorticle as noise driver in the same network. I try playing with the vorticlegeometry attributes like magnitude, the amount of vorticle force to apply on each vorticle and the radius, the radius of influence for each paddlewheel-like force. 

normal Dopnetwork

dopnetwork + seed vorticle

vorticlegeometry attribute

Also, for more dynamic feel, I tweak the oversampling of the popnet and the birth controls on the source itself.

popnet oversample

source's birth control

Day 41 The Dawn of Age

I continue testing the various system. A few of the system makes the particles go the other direction rather then the desired one. I try all kinds of method like reversing the normal's direction, duplicate it and try out different position of the system and eventually, I translate the vector field behind the emitter so to counter the flow of the particles. However, now, the particles move the correct direction but after some time, the particles start to flow the other direction. Mr Ron came by and help me with the flow direction by changing the calculation to multiply and reducing the number of scatter point.

draw back of system

alternative method

calculation to multiply

Reduce number of scatter point

After which, I test render one of the better system and add motion blur.

 without motion blur

motion blur

test render 1

test render 2

Also, the way of the water move was not quite what I wanted. The particles gather to fast and start to clump up, getting a weird shape.

clump