Abegayle Elmes

This week I was definitely struggling with this project. I felt like I hit a wall and got kind of lost through a mind field in a spider web. I felt like I wasn’t very good at this project and I was starting to compare myself to Diana who was doing the smoke and fire simulation. It was hard because the model had so many more parts to it where as the smoke was secondary effect. This made me feel really inadequate. I did have second thought if whether I wanted to continue in Houdini but I think that this struggle was something I needed to over come and learn that you are always going to come to points where you don’t know how to continue. That is the same in graphic design and will be the same in FX. I think it wasn’t that I couldn’t understand it it was just i needed an explanation in a different way.

I did feel like I was letting the team down which made me really guilty, I was worried that Medhi and Diana thought I wasn’t trying but I was it just took me longer to figure out.

I was changing and testing the different settings for the glass to make it more realistic but seems it was breaking the glass model. I think I needed to go about it in a different way as it just looks really bad.

This is what it looked like when I pushed the settings too far so clearly I wasn’t doing the right techniques.

Above is a screen shot of someone else’s break down of their car model when they were smashing it in Houdini. As you can see they made a lot of tiny segments which I needed to do to create a similar car glass style.

I added in more detail to the glass so that when I played the simulation dropping to the floor it would break off in a lot of little pieces. This seemed to reflect a more realistic constraint system.

Questions for Medhi:

1. How to connect the glass to the metal? (they break differently)
2. How to send through to the animated car to check proxy?
3. Slow but cant use the poly extrude in the compile loop
4. Bending of metal doesn’t look like metal
5. Initial force of sphere?

• edit node is not better to use transform
• use the vellum solver instead
• initial velocity to remove
• timeshift
• don’t animate in sop
• animate keys in attribute
• file cache needs to be saved somewhere to access simulation easily

After this lesson I felt I really needed to focus on how to fix it and make sure I was caught up. I had some good tips from Medhi but I think I needed to go away and rethink my method. I was getting quite nervous as it felt like I was still trying to figure out how to proceed from the first stage and it was well into the 10 weeks of work.

Due to extenuating circumstances, I had to miss this tutorial because a close family friend passed away. I rewatched the tutorial to catch up what was said with how Diana was getting on. It seemed like she was progressing really well. She was trying to figure out how to make the smoke simulation to drag and follow the car in a natural way with out breaks between the smoke so it was a continuous motion.

I wanted to make sure all the breaks in the car was working accordingly. I used the exploded view to try and see if the model was broken. As you can see below, there are darker shaded on the model which is where I assumed the model wasn’t working. I was trying to find a way to fix this with different nodes like poly fill node that Medhi had used in tutorial 6 where he was fixing the building mode. I wasn’t getting anywhere with it though.

This was the for each loop I was using that I got to work finally, the issue was that I didn’t have the named attribute correctly. It seemed to work when I started using the name as ‘piece’ though.

Here I was trying to get the glass and the model to work together. Later when I spoke to Medhi about this issue of now having the glass and the metal interacting properly. He said that it would work better if I made the simulations separately and then joined them together after so that the simulations were working to how I wanted them to look.

Here I am testing out the constraints between the pieces of the metal frame to try and change the glue constraints so that the metal would bend and not break. I couldn’t really get a result that I wanted so needed to try the vellum solver which would work better than the rbd solver.

• unpack – convert to polygons
• polyextrude to close mesh
• or divide, remove shared edge
• rbd material fracture need close polygon

So with this meeting I was having issues with the model. I was getting a bit upset with myself because Diana seemed to have progressed nicely with the smoke simulations but the car was getting more and more complicated. I was getting very frustrated with myself. I was using the rbd constraints as the simulation tool because I was following what Medhi had done in the smashed building tutorial.

As you can see the constrains were working with the glass. The white lines show the constraints. But my issue was the the whole of the glass was just falling out of metal frame. It wasn’t holding its shape like my references.

When I change the collisions from a moving sphere to just using the ground/gravity. I can see that glass would never respond this way. There needs to be an initial hold before there is a breakdown.

Here is the workflow I set up for the glass. I thought that if there was more points that it would be more realistic. Unfortunately, more scatter points just makes it slower and didn’t resolve my issue.

I then thought that if I fractured the glass differently to look like glass then it would make more sense. However, it didn’t make much sense as car glass doesn’t shatter like normal.

Shattered glass references:

Meeting with Medhi

• assemble has more options than connectivity node
• volumes to collide rigid bodies – just use sphere
• same names throughout – don’t create name
• come pieces don’t have a thickness – use poly extrude
• fracture first the poly extrude
• clean each

I was having more issues again and turns out I was making a new name in my workflow top many times so that the system couldn’t identify the pieces. I was getting very confused but the assemble node seemed to assemble everything back together again. I think I needed to go away and read more about the assemble and connectivity node. This week was definitely the hardest for me as I felt so behind compared to Diana and I wasn’t seeming to understand how to resolve me issues. Felt like I kept making the same mistake and wasn’t good enough to continue. I kind of lost the enjoyment at the stage. Hopefully I get it back.

This was how I separated my model so that I could start working on fixing the model to then distract it. I needed to get the parts like metal, glass etc. This was because these parts need to make sure the model isn’t broken so I can fracture it.

• Blast – primitives @path=/…
• unpack – transfer attributes, iteration -1
• convert – polygons, quadrilaterals
• assemble – inside group – inside piece, create name attribute, connect inside edges
• blast @name=pieces
• polyfil
• polybridge
• null

Meeting with Medhi & Diana

In this meeting we spoke about how I was going to bend the metal in the correct way. In order to use the vellum solver technique I would need to have an extrusion on the metal frame.

• Ask about animation car and cameras -very focused, so need to simulate.
• bend metal simulation – 45 mins into week 8
• extrusion – add thickness
• look at car reference
• technique – vellum solver.
• for each
• make car wheels move
• unpack
• point velocity
• poly soup – convert
• convert node to polygons
• assemble node to create packed geometry
• add node, create packet geometry
• scatter points, randomness
• remove relax
• pointjitter
• vdb from polygons – dnesity mode
• attribute noise @cd<0.5
• do by hand set up
• task memopry VMWare
• inside a loop – too many points.
• Meta import node
• detail iteration
• building destruction – go inside nodes
• Flipbook needs to be JPEGs not MOV
• FMMPEG-MPEG

After my meeting with Medhi I was able to fracture the glass accordingly in the way he suggesting with the point scatter.

I was getting a bit confused with which method I should be using. The vellum solver was something you used for soft material bit I thought it meany more sense to use the rbd constraints.

We had planned a meeting with me, Medhi and Diana to talk about cleaning up the model. Due to some extenuating circumstances I was not able to attend the meeting. I did watch the recorded meeting on Teams to catch up with what has been discussed.

This was the model that we were working on but we knew it was going to be a different car but this was what we could use to figure out the process of how we would tack separating.

• Polyfil node will work to close the polygon shapes.
• Alembic, transform, blast, unpack, convert, polyfil, divide node.
• Use man test model to see if the scale is right.
• Keep paths and UV’s – What does this mean?
• Options to fill in shape: assemble, polyfil
• Group names: inside, piece, class, name
• Isolate the problematic part with selection deleting which makes a blast nide
• Could it need remeshing to work?
• Select two edges that aren’t closed and create a polybridge node.
• Polybridge then select the edges then press enter – need only one division
• Remove polygons that have no use.
• Different numbers of polygones meant the shape couldn’t close?
• Remove extra points
• Might be faster to remodel in Maya – ask Luke
• Source + Destination in a polybridge node
• Blast for window to make them glass.
• Update the scene otherwise we will have to re do it – messy model.
• We need to find out what materials we need.
• Clean only parts that are going to break

Here was my process of breaking down the parts of the car and separating them so that I could treat them as different materials.

Questions for Luke:

1. Is this the final model?
2. How to close shapes of model in Maya?

I messaged Luke to ask how to make the model better as when I watched Medhi and Dianas meeting it looked as though the model wasn’t made very well. Luke replied saying the there has been a new brief. The building wont be smashing but the car will. This means we will have to section the car parts and not the building. This is annoying as I know Diana worked hard on figure some of this out but we have a better idea how to separate this now. I will be the one attempting to split the car up on Houdini.

Meeting with Diana 2:

We wanted to catch up since I had been unreachable for a few days because of extenuating circumstances. I informed her about my message from Luke that there was a new brief which was to only crash the car not building so we need to find this new brief and animation.

Elements of the car that will need separating:

• Glass
• Tires
• Metal frame
• Engine
• Car seats

Lecture with Luke to update us on the Indie Film Collab:

There is a new brief:

The driver goes too fast and loses control, with the resultant skid turning into a full-blown crash. It’s actually foreshadowed by a line of a song “I think I’m about to crash this car…” and so almost acts as a punchline to that. It’s a lack of observation and turning the wheel too hard whilst going too fast. Any spectacular flips, etc. are very welcome.

• driver is singing and so is the passenger – it’s a musical.
• Losses control of car because the driver is going too fast – skids and crashes.
• “I think i’m about to crash the car”

Luke is expecting to hopefully get the car rig by the end of the week. There has been some hold ups with the film director because of weather issues in filming.

How to clean up a model in Maya?

I was told by Medhi to clean up the scene in Maya would be a lot easier than doing it after. I didn’t know what this meant as such so whats a tutorial video.

• optimise scene size that is not undoable
• clean up the outline – get rid of groups