21 November 2011

Fire 03

It has been nearlt 2 weeks that I try to find out the perfect way to make flames. The pitch will be in couple of days now, so me and our team are eager to see the positive outcome from me. To tell the truth, I think I succeeded to make fair quality of flame.

I found the tutorial making candle light with softbody. Then I thought softbody wouldn't take too long to render since it isn't fluid and also the colour is from the object not particle itself. If I make the flame with softbody and glowness with the fire that I made earlier in Fire 01, it could be quiet convincing.

So I made an flame shape of nurbs object, and turned it into a softbody. I fully followed this tutorial. I linked a turbulence field to the softbody after I added 'spring' to the softbody, and I played animation and the flame was so.. abnormal.


So I linked a drag field to hold the object. I thought maybe the turbulence field could be the problem, affecting the force to around the object. I changed the attribute as cone shaped volume so it can only affects to the overlapped part of the object.



Now the overlapped part of the object looks like melting down along side of the volume area.
I increased the stiffness in the spring attributes.


and the result was..


 Even I increased magnitude of turbulence, it didn't even get shaken or moved anything. It just melted down.


I almost gave up making individual flames after continuous failing. As reading through the tutorial again, I realised the only stage that the object acts funny will be when I added spring to the object. Before hand, all i did was making nurbs object to softbody. So I started up new scene and skipped the adding spring and it worked so well! I finally succeeded to make a flame and it was perfectly moved by the fields.



 softbody flame + particle fire



Final render for the pitch




I'm so glad I could finish it before the pitch. I never thought it would be that difficult to make gas stove fire before I failed so many times. However, I could learn how to distinct which will be effective way for the scene and which won't. Plus it was a good experience that I could experiment so many various ways to create one asset. The result in this post isn't the final one for the film but I personally think I achieved 70% of job for the completion.

16 November 2011

Fire 02

I wasn't happy with the default fire in maya dynamics. I needed more organic and subtle type of flame. It could get away f it showed only a bit during the short time in the scene. However, the dynamics in our project takes huge part. It should be perfect and even it was only experimentation, I want it to reach at least 70% of final quality.

So I spent days and days to research to make perfect flame. I found some examples on the youtube but those were made with very heavy methods like maya Fluids, Smoke, Houdini etc. I am interested in houdini anyway so I tried to see how it looks.


 

This image is the screenshot of houdini. I only set up the default pyroflame. I didn't modify any numbers, apply any nodes. It was the most basic fire that anyone can make. The quality of fire on the test render looked good but it looked slightly too much but I thought it would be suit for our scene after some twig. But I HAVE TO turn it down. I rendered 250 frames and it took SO MUCH time to render. I had to stop right after it rendered 75 frames after 2 hours. It wouldn't be worked if it took that much time to render. Also we need at least more than 10 flames around the ring that emitting gas.








After the attempt for houdini, I tried to see what maya Fluid fire would look like. Like the pyroflame, I imported a fluid flame from maya internal library. Again, it took TOO much time to render.



 After failing few times now, I'm getting more anticipate to find way to make decent looking flame but can be rendered relatively quicker than fluid and pyro flame..

11 November 2011

Fire 01

So, I started to make fire for the gas stove first. I used the basic fire option in dynamics menu in maya.
With the model Arpit sent earlier, I created a nurbs cylinder as surface emitter, placed it where the flames come out on the stove and hid it. Below image is the screenshot of first animation.



The colour is set as the default so I played around with colour, transparency, and incandescence.

1) Colour - I set the ramp of colour that I want the flame would look like in final outcome.



2) Transparency 

Default setting for Transparency




I wanted to make the fire starts with transparent blue and ends with a hint of orange but it didn't work very well. I didn't really like the transition when blue becomes to orange. It got too bright in the middle which doesn't look organic.

3) Incandescence
Default setting for Incandescence





Playing around with incandescence was very tricky because when I change little bit of colour, it directly affects to the flame colour and glow.



So the first attempt to creating fire wasn't very swell. Also I thought the fire is too blobby and moving too much when the real gas stove fire looks smoothly comes out and more stands still.

I played around with blob map option under the transparency tab, and the result was getting more horrible.. The blob map is worked by a texture map which I have never used before and I found it's really difficult to control. I got the sense it makes blurry depends on the shaker value of the texture map but it didn't really help for whole look of flame.







At this stage, fire looked too flat whereas I wanted to make flames look transparent  in the middle and getting coloured as it reaches to the edge. After few research, I found the Translucence might can help with this issue.





HOWEVER, despite my expectation, translucence couldn't help it like above screenshot. I reckon translucence would be useful when it's applied to solid object, but it didn't work properly this time cause the fire glows itself. I was getting tired..

Then I started to modify figures of fireshape, reducing the glow, try to make fire solid and lower the height etc.

Fire so far with default Goal UV setting
changed Fire Scale 1.0 to 1.5

Reduced the Fire Speed to make it more stable so that it doesn't look too blobby
Increased Fire Lifespan that particles can stay longer for the same reason as above. Less blobby.

To sort out the height issue of flame, I applied drag field to the fire. As increasing the magnitude of drag field, the fire came down to right level. I also applied turbulence field to give a little sense of shaking by emitting gas from the inside of stove. After modifying many figures, it now has more subtle look.


 

08 November 2011

Fire reference

As we have to present our formative pitch on 24th Novemeber, Valentino asked me to make fire and steam for kitchen scene. So I started to find some references. I have recorded different types of flame.

1) the stove flames


2) lighter flame

3) Candle light


 The main aspect of flames from stove is that there are each individual flames are coming from the holes, and you can see the glow around the flames. Also the emitting point of colour is fairly transparent, the side edge is blue and top of the flame appears a hint of orange.

26 October 2011

Creation expression VS Runtime expression

Over the past few days, I was studying controlling particle using expressions and while I was following a tutorial, I reached a deadlock few times. It took 2 days to sort out the problems. It has completely messed up my head so that I had to study the concept of creation / runtime expression from the beginning again.

The original particle movement I wanted to create was that randomly created particles revolving around the Y axis in a limited area. Below video is the final succeeded example.





At first, I wanted particles emitting from an emitter and revolve around the Y axis. So I created an emitter and typed the expression in particleshape like below. The expression means "the particles' position = (From the position of a point, rotate around coordinate <0, 0, 0>, angle of radius 0.1)".




However, the result seemed the expression didn't work at all as you can see from the below video.




It looked like emitter created only one particle and it just vibrates itself. I checked the emitter and particleshape attribute but there was nothing unusual. So I looked up on the internet forums and scripting books, and found the answer. It was because I didn't initialize the position of particles with creation expression.

There are two types of expressions which are runtime and creation. You can apply creation expression and runtime expression to the particles. It can be used separately and both together at the same time. If you connect expression to ordinary object, you can set expressions to each different attributes, however, you can set creation expression and runtime expression to only one particle bearing no relation to attribute.


Creation expression evaluated only once for each particle when particle is born. So it doesn't affect to the particle after it has been created.
Runtime expression evaluated at least once per particle per frame but not at particle birth. For the same instance as creation expression, particles from emitter will be affected by the expression even after their birth.

I found this summary from my blog that i made for tutorial unit last year.. [http://pnsamelie.blogspot.com/2011/03/creation-expression-vs-runtime.html]
I should have been aware of the difference between two types of expression since I cleared up the definition. I am SURE I will never forget these.

So I added a creation expression which initialise the start position as <3,0,0>.



And it didn't work as i wanted it to be again. The result was like this video.




The emitter worked and particles were created. However I had to ask myself why the particles are only created but don't move at all. As now I use both runtime and creation expression, they should have rotated around the axis. According to the expressions, a particle is under the control of runtime expression after it's created, which makes it to rotate around <0, 0, 0> angle of radius 0.1.

I was confused again, I couldn't understand why particles don't move. I looked up online forums with every possible keywords. Then yesterday, I finally knew how i can figured out the situation while I was working on different example. I could easily check it by changing the particle render type to numeric. It shows the creation number of each particle.

It wasn't the matter of expression, but only an eye trick. They are actually rotating in a circle. It just looked like staying on the same spots, because the particle movement was so regular that newer particles created on the same position as older particles'. Below video is rendered in numeric but it's a shame it doesn't show the numbers clearly due to the size of video..




Now I changed the creation expression to gives random creation spots in the space within coordinate of <5, 5, 5>.






The final result looks satisfactory. The first problem can be seen as a mistake because I couldn't remember the attributes of expression, but the second one was much harder to figure out without knowing the "numeric" render type. I'm sure there will be another way to find it wasn't the matter of expression but still it would be the fastest and easiest way to check individual particle's behaviour.



18 October 2011

Inspirations

This clip is what I watched in FMX festival, it has great quality of effects and gave me sense of interacting with animation. Also I could get hints for fog and mirage for Synesthesia.



Loom from Polynoid on Vimeo.


Making



Loom Fluids - Making Of from Andreas Felix Gebhardt on Vimeo.



The below one can be a good reference for fluid, although it's a combination of actual camera shooting and 2d, 3d animation. It's made by turkish motion designer Onur Senturk.






Making


Nokta . Making of from Onur Senturk on Vimeo.


The third one is made by german cg artist Oliver Vogel. Personally, I think it's really interesting when it's turning from 2 dimensional to 3 dimensional. Also it's brilliant making 3d sequence with 2d elements like a dot and lines.


17 October 2011

Plans for this year

For the graduation work this year, I'm going to help out other teams who need special effects and dynamics. Also I'm making a personal showreel specialised in VFX alongside of helping out other teams. 

Firstly, there is one team who wants me to help their project, Synestheia. I have their script and animatic video but hasn't been made into solid storyboard yet. Below video is the animatic footage of the story.



Briefly, they need mainly 5 scenes of vfx so far.

1. Fireworks and spray paint burst from a door hinge sounds
2. Sparkle and mirage from a kettle
3. Aura around a dog / people 
4. Colours come out from each piano keys
5. Different colours in red when the main character sees the colour.

I planned out which elements of special effect would be needed for each scenes. 

1. Fireworks and emitting different colours of particle
2. Fog and particle. Also developing the idea of making bubbles.
3. Fog which needs different type of texture and colour from normal fog.
4. In this scene, I suggested to make particles surrounding her after they come out from piano keys. This will need curve flow and particle instancing if particles have certain shape.
5. I was thinking it would be nice to make the colour looks like hologram so it can be seen changing from looking in different direction. I think it need some random function for colour of particles.


I will figure out what the most effective way to create these effects will be by research and experiments.

Besides, there are other teams who want effects in their film but they haven't fully decided yet. So I will just practice first and help them later whenever they decide to use effects.

Secondly, for the personal showreel, I'm thinking of making 30sec - 1min sequence of a star in the space from the birth to the death. It will contain all kinds of effects and dynamics, mainly fog, collision, fire, lightning. I still need more research for these and will keep upload the experiments on them.