Day 40

We all came back last Saturday to put in extra time for shot 5. I managed to render the demons falling and the debris, but in 640X480.


I showed the renders to Mr Douglas. He told me to render more debris at different sizes and noises. I found that the orbit node system I'm using can make what seems to be a tornado. I just played around with it to see the results.


                                    Trying out a tornado effect

I'm pretty happy with the debris effect I've been working on. I'm just using lots of particles with sprites. I made placed greyish brown colour onto the sprites. I also added an alpha ramp using the same colour node, but I'm not sure if it's doing anything.


                                           The falling debris

For today, I managed to render a HD version of the debris and also a HD version of the demons falling.


For now, both of them seem to render pretty well. The debris takes a little more than an hour for 240 frames. The demons falling take slightly less than an hour for 240 frames.

Day 39

Today i refined the speed of the particles. Made them move slower, still using the multiply by percentage method. Now it plays out better.

Mr Ron said that the particle trails will be for dust and stuff. I tried out using sprites. Now it just looks like smoke instead of dust and debris.

Coming back tomorrow for awhile to work more on the dust.

Day 38

Well, yesterday was a public holiday,so we didn't come in.

Tried to figure out how to make the particles slower,but I couldn't figure it out for myself. Then Mr Ron came, and solved it in like a minute. All I had to do was to multiply a part of an expression, which was using "LIFE" by 0.4 or something to decrease the percentage,so it seems slower. This resulted in the spiral to be really short and not reach the end. I had to play around with the other parameters to get the size right.

Wanted to reference Joyce's tornado for the scale, but she wasn't done with it. Her tornado was really small and I tried to scale it up, but I couldn't figure out which part should be scaled. I tried scaling the base circles and curves, but nothing worked.

Day 37

I refined the spiral effect digital asset. Added a few more parameters like number of particles, turn speed and birthing of the particle trail. I tested using the reference geometry onto the spiral effect to see how it looked.

I continued processing more mocap data. I processed like 30 of them but converted only 13 of them into hip files. The other files were captures for other scenes, which are not important right now.

Mr Douglas came in to day and I showed him the spiral effect. He thought that the particles move too fast, so I tried to make them slower. He told me to try to use the speed limit node. I tried it out but although it could slow down the particles, the particles do not turn around anymore and instead of looking like a tornado, it looks like a teardrop forming. I need to fix this

Day 36

First of all today, I tested making the spiral effect into a digital asset and placed it in shot 5. It worked correctly. I added 2 more parameters, the length of the particle trail and the life of the particle trail.


                                     Spiral effect in the scene

I also managed to process and convert 105 mocap data into FBX, then into hip files. Very boring, but it has to be done.


                                   Some of the files for proof

Day 35

Spent the whole morning and part of the afternoon at a school event. Wei Jie and I were helping out there. Still informative though, there were talks by Lucasfilm, Dougle Negative and Ubisoft. I was in charge of looking after Ngee Ann Secondary School. Toured them around the school too.




                                    PICS OR IT DID'NT HAPPEN

Credits to We Jie for the pictures


When I went back, I started to edit my spiralling effect. I managed to get a spiralling down effect,which is needed to show that the demons are spiralling and falling downwards due to the tornado. I made some custom controls to control the size and height of the spiral so I can export the system as a digital asset later on.




                                              Spiral effect

                                             Using the controls to edit the spiral

                                             Using the controls to edit the spiral

Mr Leong finally helped me install Vicon Blade on my computer in our FYP room, which is awesome. Now I can multi-task and process while I do work on Houdini. I tried again to batch render, but it crashed again. The problem now obviously lies with the software,not our computers, I'm on a 24 core PC...............


           
                                                It works

                                         Processing the individual captures work

Day 34

Today a few things fixed.


First of all, the wavy terrain. Initially my problem was that the particles would move down too fast from the peak of a wave. The problem was with the ray sop.I was using the minimum distance attribute to make the particles follow the  terrain. The problem with that is that it uses the normals of the terrain to determine the minimum distance,hence, the particles will kind of skip to the nearest normal referencing to the birth place of the particle. I was painting on the areas of where I wanted the particles to be birthed. The grid I painted on was already wavy and so the placement wasn't on even ground. 


After much help from Mr Ron,Mr Ng and Mr Douglas, I eventually used the creep sop instead of the ray sop. With the creep sop however, it had all these crazy positions and I had to rotate certain rotations at 90 degrees. In the end, the particles move as expected, without speeding up or slowing down, which is the effect I wanted.


                                   Particles follow the terrain

                                         Instead of ray sop, I used the creep sop

                                         


Secondly, the spiralling effect. After letting Mr Ron and Mr Douglas see the spiralling effect I was working on, Mr Ron suggested an easier method,compared to the method I was using whereby I copy stamped boxes to make a spiral. Mr Ron instead used mainly just the popnet, using an orbit node and a few expressions to make the spiralling thing. It's so much easier and cleaner than what I was doing at first. 


                                               Spiral effect

I'll have to tinker with the new setup to get the desired effect.

Day 33

So I started out tinkering with the spiral effect thing. Yesterday, Jia Bao suggested to make the spiral using the point SOP and then scattering points on it. Then I would add normals to the points and have them face the direction of the spiral. I managed to get up to adding the normals, but I couldn't figure out how to change the direction. In the end the particles don't follow the spiral and just drop down.


                                              Spiral Shape

                                               Normals

                              Normals not facing desired direction 

                                        Particles going down

Wei Jie passed me the terrain wave thing. At first the one he passed to me was quite high poly, and as a result things got very slow and I couldn't test the particles easily. He then passed me a really simple low poly version which has a few large waves. I noticed that when the particles are coming down from the peak of the wave, they kind of just slide down real fast. I don't know if Mr Douglas wants this effect,but personally I don't really like the look of it. I always imagined for the particles to just be at the same speed and velocity regardless of where they are.


                                    Particles on the wavy terrain

Day 32

I tried different ways to do the spiralling effect. At first I created a curve , but I couldn't make a  nice spirally curve, I had to manually translate the point of the curve to get my roughly desired shape. The particles ddo follow in a somewhat spirally way, I like the chaos it has, but there's a little too much chaos.

                                            Using a circle

 

                                                      Using a poorly drawn curve

The second method was using a geometry and a copy sop. I tried different geometries, and in the end settled with a rectangular cuboid. It made a really nice spiral, by offsetting the translation in Y and rotation in Y. Mr Pang thought me how to do this. I assigned the particles to follow it using attractors and it roughly followed the spiral for a while, but then the particles fly out too much and instead of going down at one point, they start to go back up.


                                         Using the cuboid

I'm waiting on Wei Jie to finish his wave effect for the terrain so that I can test the crowd on the animated terrain. I'm curious to see the effect, I;m sure it'll be pretty cool. Hopefully the particles will react as expected and not give me any problems.

Day 31

I spent the day just refining my crowd sim, making my network run better too. I added a volley of arrows just to test the attractors and stuff, just for testing though. The particles try to avoid the arrows as I expected, but they avoid them only when they are about to land, well because the size of the attractors are pretty small, but if I make them too big, they'd move too much.


                                              Arrow Volley

Mr Douglas came in today and told us that its game time. We're going to start on shots. He did shot blocking for 1 shot with Wei Jie's mountain and a temporary sea. I'm supposed to make a particle system for the falling demons, they'll be spiralling down a tornado. I roughly tried to do it. At first I thought of makign a curve and let them follow it, but I couldn't figure out how, I'll probably find out by tomorrow. Right now what I have is the particles, instead of spiralling, they are just making a tube-like shape. I used the turn node. I'll figure this out though.


                                        Spiralling attempt

Mr Leong installed Vicon Blade in a computer on our floor. I transfered the raw captures on there and tried to batch render. It crashed,again, even with a super duper computer. The problem now has to be with the software itself. If I have to eventually just manually process the remaining captures, then I'll have no choice but to do it.

Day 30

I messed around with my crowd sim file today. I tried to make the red particles "interact" . They're just moving around using noise. Now, instead of the green particles moving around trying to avoid the red ones, the red ones are kind of moving out of the way, since the green ones are charging through the battlefield. I think this works better rather than the green ones slowing down and avoiding the red ones.

I removed certain nodes in my popnet because they were just unnecessary, such as a property node that hard a charge and uniform scale attribute. I deleted that and my particles were more stable. Don't even know why I had it in there in the first place. I added a soft limit node to kind of control the boundary of the particles.

Mr Ron said that Mr Douglas will be coming in on Monday for shot blocking. So, I guess we'll be starting more on the shots next week, at last.

Day 29

So we tried out batch processing with the new computer, it still crashed miserably , just trying to batch process 5 files. Mr Leong said he'll fix it by the end of the week.


So I processed like 70 captures today. I managed to open them up and convert them into HIP files. Spent quite some time manually processing each file in the mocap studio. I still have not processed all the captures from the latest mocap session.




                                                 More Files

Messed around with the crowd sim. Not much difference.

Day 28

Today I managed to load up and convert 35 FBX files to HIP files. Some of the files are from the earlier mocap sessions and some are from the one we had yesterday. The files seem to be working okay. The ones we captured yesterday seem to load up faster. I wanted to transfer more files over, but they are changing the cpu in the mocap room, that's a good thing actually. Hopefully I can batch process the remaining files tomorrow. I updated my mocap file list, now it's at 387 captures.

                                                                     Lots of captures

                                                                          Lots of files

Mr Douglas said that the particles skipping when going up a slope could be due to a resolution problem. So i increased the subdivision count on my grid. It did help a little,but there particles still skip to the top.

Day 27

Did mocap for the whole day. We managed to capture everything we needed. Hopefully, this would be the last session. The new computer for the mocap room will be coming in tomorrow,so hopefully I can batch process the mocap data. It'd make things much faster for me. I did export a few captures to test them in Houdini.

Day 26

I somehow figured out how to  copy the color of points onto geometry, finally. It was so easy,don't know why I overlooked it. It had to do with the copy sop. I had to tick "Use Template Point Attributes" and typed "Cd" at the "Set" column on the "To Primitive" row. Well that was easy.


                            The particle colors are transferred onto the geometry

I also refined my network. Added other paint sets and a switch node so I can switch between different types of paint templates. It does not however change the behaviour of the particles.I have to manually change the settings for those.


I also spent some time today preparing the other 6 cameras for the 18 camera setup in the mocap room, for tomorrow's shoot. Hope everything goes fine tomorrow, and we get stable results.

Day 25

We decided to have another mocap session to capture Lu Sheng's actions next Tuesday. We figured it'd be easier to do new captures on him rather than fixing the numerous problems we have with the current captures. With the new 18 camera setup, at least we know we can do clean captures. Moreover, it'd be faster to do new captures since we now know how to operate Vicon Blade smoothly and it's only 1 person, we can quickly do it. Now I recognise the mistakes I've made during the previous sessions and I now know how to prevent them and be more alert to each thing during the capture session.


I tested out collisions and stuff with the new crowd sim. The collisions work. I also used attractors and they work out smoothly too. This solves the problems I had with my old setup. I'm also using less nodes in the POP network, so things are much more simple and clean. I did notice one problem though. The particles do follow the terrain with the ray node there,but, I found that when the particles go up or down a slope or something, the movement isn't smooth. The particles kinda just jerk up to the top real fast.




                             Added a few boxes to test collisions

                                    The particles stick onto the collision object

                       Can't see the attractors for some reason, but they're there

Day 24

Mr Douglas and Mr Ron came today. He tried to fix Lu Sheng's crazy hands in Houdini, but Lu Sheng's hands were way too crazy. I guess the captures were too unstable. We were using 12 cameras instead of 18 when we were capturing Lu Sheng's actions, so there was less space for him to move around. With less space, he was more prone of going out of the boundary where his markers can't be captured.


I tried using advanced processing instead of basic processing in Vicon Blade for Lu Sheng's captures. Advanced processing had a few more processes like "Fill Gaps", which I thought would help fill in the missing markers. Nope. The result was still the same.I'd have to edit it myself, but I'm not sure how to in Vicon. Mr Douglas showed me how to edit the joints in Houdini though. But some of the captures are too crazy to fix manually. 


I exported like 20 more FBXs today. When I batch exported the processed files to FBX, it crashed too! Just exporting FBX files and Vicon crashed. Well at least I managed to get 20 out. My mocap file check list is just about done, I think i'm missing 1 or 2 files, but 99% of them are on the list. The list made it easier for me to keep track on what I've processed or exported. 


                                              The long list

Mr Ron introduced me to the ray node today. Why didn't I discover it earlier. Now the collisions and stuff work,because of the ray node. I don't have to use that slide collision node I've been using in the POP network anymore. I guess I've been focusing too much on fixing everything in the POP net. The ray node is connected to my terrain and POP net outputs. So now all the particles just follow the terrain, just like that. I tweaked my network to be less cluttered too. Instead of using two seperate terrain to paint on, I painted on just one, merging two paint nodes. I now use one with red paint, and one with green. I just changed one of the particle source's Birth Probability to "$CG" instead of "$CR", since it's green now.




                                          Overall Network

                       
                                                       Red and Green Paint

                                                               Particles 

                                                             POP network

                   
                                                                 Ray Node

                                          Characters placed correctly on terrain

                                          Characters placed correctly on terrain

Day 23

I spent the whole day opening the FBX files (that load really really really really really slow), fixing the bones rotations and head placements, and saving them out as HIP files. The HIP files sure do take a lot of space, I did 48 so far and they take up to about 7 gigs all together.


                                            Lots of HIP files

I still got a lot more files to fix and convert though. I made a list to keep track of what files I've processed, converted to FBX and converted to HIP files, also noting the captures that are messed up. The list isn't completed yet, I'll need to go down to the mocap lab and list down the others.


                                                  The list

I also played around with the crowd sim setup. After a bit of tinkering around, the particles from both sources seem to collide and avoid each other. Now they also charge at each other,instead of moving around blindly.

Day 22

I figured out how to do a batch export to FBX files in Vicon Blade. So now, I don't have to manually export the Vicon Blade HDF files to FBX,and make the computer do all the work for me :D. But the batch processing the files still do not work properly.......yet. So, I haven't processed all the captures yet.

I imported some of the new FBX files I exported today. Some bones did not appear like before. I then changed the "Look At Up Vector" for all the bones to "Use Quaternions" instead of "Use Up Vector", to stop them from rotating. The rotation stopped, and the missing bones also appeared magically. I guess the up vector was some crazy number or something, or got divided by zero.

                                         Dislocated Head

                                      Dislocated head is not dislocated anymore

                                                              No Missing Bones

Another problem with the FBX files is that, they load way too slow..........

Since they load really slow, i spent quite sometime trying to load them up one by one and fixing the rotations and Nick's dislocated head.

Also messed around with the crowd file Mr Ron sent me yesterday. The particles don't seem to directly charge at each other when I hit play. Some of them go other directions, I guess it's because of the noise function thing.

                                     Start at opposite ends

                                     Some particles stay behind,go other places