camera animation........done!

I have finished the camera animation. As you can see from the video below there are only 4 characters in it and they are not doing much, they are just to help with camera alignment. You will also notice other than the guy jumping (that was purely to see if the explosion part was going to have the desired effect, I think it works) and the doors opening nothing else is animated. I have already begun with the lights but that’s for another post, once I have ironed out all the problems I will no doubt come across.

During the process of the animation I was rendering out small parts of the scene to see how they look, as a result below you can see the actual quality for the first half and a wireframe preview for the second half.

The animation begins

I have begun the animation process. I have opted for a fairly unusual approach to it however, I am animating the camera around the scene before adding and animating the characters. Why? I hear you ask, well my scene is not completely devoid of characters, I have 4 in there at various points to allow me to position the camera at the correct height. The main motivation behind this decision was that I intend on having a very busy scene with lots of characters, as I already have the storyboard in my head I can animate the camera where I want, then I can see which characters need animated at which time, potentially saving me hours on animating people that are not even in shot.

I have been working in dual screen with the curve editor permanently open on one screen, I can almost read it like a book now. Below is a screen dump of my extended workflow,

The reason I need the curve editor so much because I am making my animation quite quick and all the keframes flow smoothly in to one another, as if the keyframe was not there. Max has a habbit of adding a slowing curve to the animation on either side of the keyframes. When I am animating around something I add the first and last keyframes then add the ones in the middle to curve the arc a bit, Max would then slow the animation down when coming in to the middle keyframe and speed up on exit, this makes for a very ‘stop start’ feel to the animation, not very smooth flowing.

Another problem that has arisen because of this is that, on the odd occasion, Max can not handle the curves without being told how to do it, what do I mean? In the picture below I was animating around the 2 characters whilst looking at their face, when adding additional keys to curve the motion Max added a massive Bezier curve causing the camera to look at the floor, by using the curve editor you can easily see where the problem lies and quickly correct it.

Because I am animating around a lot of things in smooth circles I have adopted the technique of animating the camera along a spline to keep things smooth. I have previously found manually trying to animate a camera in a smooth circle is both time consuming and tricky, it often ends in a very jerky camera motion. Because I am using the same camera for the whole scene I don’t want to constrain the camera to a single spline, instead I have been using the method of converting motion paths from splines for a pre defined part of the animation. Below is the before shot with the cameras trajectory.

Here is how to achieve this.

1. Draw your spline and adjust it in to the shape and position you want your animation to take.

2. Select the camera (or any other object you want to follow the path), open the motion tab and select trajectories.


3. Under sample range set the start time and end time and decide how many you want it to have. Samples means keyframes, if you select a start time of 0 and of 100 with end time 20 samples you will end up with 20 keyframes evenly spaced along your spline. Its important to realise here that of you do not have enough samples on a curve your animation will not be able to follow the path exactly, like below.

4. Now turn on AutoKey and hit the convert from button and select the path, your camera is now constrained to the path but only for the time that you selected and is free to go anywhere else for the rest of the animation. Yo may find that if you can not get the rotation to match as you had wanted you will have to alter the rotation values yourself. With AutoKey still on go to the first keyframe and rotate the camera how you want, then do the same for the middle and last keyframe. Because the constraint did not set any rotation values for the rest of the keys you should find that thoose are the only keys you need to edit, the rest of the keys should inherit the correct values.

I hate sound (check out the video montage though)

But sadly it must be done. I have final completed the sound file for my animation, I can now begin to time my animation to the 1 min 7 seconds it lasts (my animation is going to be 1675 frames long!). There are three songs

1. Shoot the runner by Kasabian
2. Gothic Anthem by Gothminister (thank you Martin)
3. Elements of life by Tiesto

In between the tracks is an explosion followed by a rewind sound effect, here is where I plan for the camera to spin around the dance floor and add lots of motion blur. The reason that I added an explosion was that when I was trying to mix the sounds all together, keeping the animation in mind, it was not working with just the rewind sound effect. It was actually whilst listening to radio 1 that I heard an advert for Tim Westwood that used an explosion cover up a mix between two different genres it worked really well. To mix all the sounds together I used Adobe Audition. In this programme, after mixing all the sounds together I normalised all the tracks so that they were all playing at the same volume.

You may remember me mentioning a log time ago that as a group we are all beginning our animations on the outside of our room’s door, having the door open and entering the room to begin and finishing on looking at the inside of our door. We will then, as a group, animate from the end of each others animation to the start of the next persons animation via the building halls. (Trust me, this will be important info in a bit).

After I had finished making my mix track it came out a bit too long so I began the process of trimming it down to length (1 min, agreed by the group). I eventually had the idea to cut a large chunk off the beginning as it would save me 15(ish) seconds. The problem was that now the music seemed to start a bit sudden and didn’t sound right. The solution I came up with is that it can begin whilst the camera approaches my room through the hall. Now I had more work to do, I had the break point in the track but I needed to decide on the speed that the volume would increase and the start part would need to be muffled until the door opens. After much messing about I have the volume continually and steadily increasing over the course of the first part from 0 to 50% then the second part of the audio, which is where my personal animation will start, there is a sharp jump from 50% to 100% which is where the doors will open. I managed to get a muffled effect in Audition using the Graphic EQ and removing the treble but as it turns out the effect can not be gradually reduced (as the doors open) so its an ‘all or nothing’ effect. To solve this issue I decide that I knew how to do it in Premier Pro, so that’s what I did. I exported the two separate sounds from Audition with the changing volumes and then in Premier Pro I keyframed in and out the treble to simulate the muffled effect then exported the audio again as separate tracks. The result of all this work is that I now have two separate tracks, one flows seamlessly in to the other and the volume increases as you approach and finally when the doors open the muffled effect disappears.

Below is a video montage of the creation of the sound (sped up obviously) with the completed track as backing music, it shold be a delight for the eyes and the ears. Since making it I have moved the break slightly and the bass and treble transition now runs over the break but you get the idea. The break is at 12 seconds. Enjoy.

The decks are finised

I have finished my decks. The final method that I was going on about in the last post where I was going to make an animated material worked. While the material and the glow works the timing still need a lot of work, I think the first 4 or so beats work really well then the rest all fall out of sequence. For the sake of testing if this is going to work I am happy with the results. Unfortunately the video here has lost quite a bit of quality so some of the detail ahs been lost, things like the detail on the record and the detail on the side of the turn table that the strobe is flashing at.

To create the new material I opened a new flash document and made the stage are 15 pixels by 132. I got this size from the opacity map I made for my earlier experiments as when I was cutting the holes in the mixer that is what I used as a guide. I then added the colour and some black bars which would mask out the pars of the colour I did not want showing at the various key frames. I should probably mention that I did not actual mask the colour as in max black will not glow and there for will not show up as a light. Last thing before beginning the animation was add the sound track, this was not going to be in the final animation sequence but it enabled me to see where the music peaks were making the whole process go a lot quicker. On the animation I simply added one key frame at every peak and one in between for the lows then animated the black bars to cover what ever I desired at each frame. For the final animation I plan to actually work out how it should look a lot better.

I decided to use flash to create the material for a few reasons. Firstly I could add the music to a layer and see the peaks in the music helping me to decipher where the key frames need to go. Secondly it is capable of producing an AVI file at any size (15 x 132 in this case), if I had done it in premier pro and asked for an output that size all I would have got would of been a 720x576 file squashed to fit. Lastly because flash is easy to use for this sort of thing.

For the decks I wanted to simulate the strobe on the side. On a DJ’s turntable there are a series of small, differently sized rings of circles circulating around the edge (I know that doesn’t make much sense so here is a picture),

the strobe flashes at a constant speed and as the DJ speeds up the record a different ring can be seen to help you work out how fast it is going. Normally the strobe flashes so fast you can’t actually see the flash but unfortunately with max I was limited to one flash per 2 frames. However I still caught the essence of its purpose here but that is part of the detail that was lost when blogger compressed my video, sorry. One thing I am really please with on the record though is the specular map working well to simulate a shiny record and a not so shiny label.

I have also been working on a lot of my scene, it now has windows (although the blinds are closed) which even have the blind pull chord things, tables and chairs, photo boards for the walls and even and exit sign above the door. I realise that some of the photos are a bit dark but this is going to be a dark scene so I have been creating everything in the dark to make sure that it all looks ok when the lights finally go out.

I think I am nearly ready to begin the animation phase. Wish me luck, I will keep you posted.

the long creation of the decks

Ok, on to my own work now. I have been finishing of the decks and the mixer and all is going well, nearly done.

On the mixer I want the beat bar lights to actually light up and be as animated as beat counter lights are in real life. I have been trying to fathom the best way to approach this and went through a few ideas before finally settling on an idea. So before I tell all on my plans here are the ones I have been through so far.

1. I was going to use Photometric Free Point lights, add a volume light effect so you could see them glow and animate them on and off in sequence, the same as I intend to do with the light above the dance floor. The reason for wanting to use these lights is because they have the ability to glow in a rectangle and in a confined area, matching exactly the lights that would really exist there.
   
   This idea fell flat on it face with a bloody nose very quickly with no chance of suing the local council for wonky paving slabs. Two reasons here. 1) Just adding one of these lights with the volume light effect took the render time up from 2 second per frame to 45 seconds per frame, an unacceptable mark up and I can only imagine what would of happened if I added all 16 that I need! 2) Animating 16 lights on and off in sequence over every frame would have been a logistical, time consuming nightmare.
   


2. Idea 2. I created an opacity map to make the area of the light totally see through. I then put a coloured block underneath so it could be seen through the holes created by the opacity map. Using the material ID properties I added a glow effect to the coloured box to create a simulation of light but not real light, this should speed up rendering time. The box didn’t glow! It turns out that Max is unable to render a glow through an object, even if it is see through. The reasons behind this are complicated but in the simplest possible terms, 3D Max adds a glow after rendering the scene and it adds it to what it ‘can see’. As the object is only see through it still exists and is still in front of the object with the glow, therefore Max can not see it to add a glow to it.

One thing I had to consider was the way that I was going to animate the lights on this box up and down. I thought about using a black box animated in an up and down direction over the light box and use it as a mask. But more on this in a bit.

3. The next logical option from there was to cut the mesh for the individual holes so that the glow could shine on through, and that is just what I set about doing. After slicing a whole load of new faces and deleting the relevant polys,
   
   the glow shone through proudly. Still a problem though, the glow was white and unrealistic. I could have changed the colour of the glow using the effects settings but I would have only got one colour, and there are three on the box so using just one colour would not have worked. I went through the phase of thinking about using three different coloured boxes but this would increase my polygon count and meant adding three glows instead of one, increasing the render time.

So after all this I came to my final solution, to use the glow function in the video post. In the video post you get a lot more control over the glow. I set the glow up to glow the same colour as the colour on the box and this worked perfect. The polys that remain in between the lights on the mixer block the glow creating the illusion of individual lights. I have managed to create light without using any lights, thus speeding up my render time. This means that I will have to set up to render my final animation through the video post but I have had a lot of experience here so I don’t anticipate it to add much to the setup times.

Just one thing remains, the animation of the ‘on off’ of it all. I was still considering the mask trick previously mentioned when I thought “that will look fake, I will have to animate both channels separately and the mixer is actually on an angle making the animation up and down slightly more complicated.” “I know, I am going to use an animated material!” This is something I have never done before, excellent, I will learn it then. I plan to use a carefully timed AVI of this image where there are black lines jumping up and down the left and right,

when put on to the box under the mixer it should appear as though all the lights are operating independently of one another, I hope. I will get back to you on that. For now I need to have a quick try, if it works then I need to decide on a track and try and time the AVI to that track.

A small piece about those character problems

I mentioned this briefly in my last post, I don’t want to dribble on here and bore you with this one as I’m sure by now you have had more than enough of the situation (I’m not far off losing the plot over it).

Martin, another member of my group has the rock room to animate and it involves a lot of these characters using various instruments. The drummer’s arms were the culprit here. When raising his arms enough to bash the dead animal skin with a dead part of a tree his under arm would deform in a rather nasty way.

This problem was being caused by the fact that the elbows on the character do not bend and twist in the same direction as a human arm. This is because the mesh is deep but narrow, if the IK (inverse kinematic) was in the same as a human arm there would have stretched too much on one side and crushed in on itself in the other creating a bad pinching effect. In an ideal world we would have solved this problem by detaching the IK, unlinking the character controls (IK’s and bones are linked to them for ease of animation), re-attaching a new IK in a different direction and linking new controls. But it would not end there, no, we would then have had to enter the skin level of the character and changed the vertex weighting of the bones to accommodate for the new direction in movement. As all this is still new(ish) to us it would have been a trial and error process taking up way too much time. To solve this problem we rather ingeniously detached the IK from its elbow control so the mesh could twist a bit closer to what he wanted.
It wasn’t as much control as he would have liked but it was enough for him to be able to complete his work given the short time that remains for this module.

Go on my son.

(After all that, it this post was longwinded, I appear to be unable to write a short blog, and now I’m just making it longer by waffling away here).

Some new updates on my YMCA scene

So far I have spent a lot of my time rigging this Y character and trying to help resolve problems that people have identified with the animation of the model (more on that later).

Today I want to report on how my scene is coming along. SLOWLY! But do not fear, for I intend to spend some serious hours modelling over the coming days then the problems will lie in character animation and animating the lights.

So far I have modelled the light rig, the DJ stand, the dance floor and part of one deck.

The lighting rig was modelled using a series of simple lines all attached to one another once in the correct place. I made one cross then used the array tool several times to create lots more in the relevant directions. I then applied a grey material with a high specular to simulate shiny metal. I figured rather that as it is not an integral part of the animation this would be a better option than adding raytrace material to it for correct shiny metal as it would speed up the rendering time.

I have created a nice shiny dance floor with proper reflection and everything. The dance floor is where most of my animation will take place so for that reason it was important to me that it be as realistic as possible. When the lights are up you can see the reflections in it, some noise bump (to simulate it being slightly worn but well used) and a high specular to give it a nice polished look.

I have begun creation of the decks that the DJ will be using. They are in their early stages but as you can see they are coming along nicely. I have already created a lot of materials for these but I only plan to show them for a few seconds in total, I think I may have gone a bit overboard on them (as usual, it really is a curse). The main element of interest behind the DJ booth will be the use of volume light to show what is going on.

My original deadline that I imposed on myself to have the room modelled by was the 14^th November. I am going to miss that deadline due to the complications with the character and I now hope that I can have most of my scene animated by next Wednesday the 19^th. I’m fairly confident that this is a reasonable deadline and will do everything within my power to reach that goal.

Motion

Over the past few days I have been sorting out the YMCA characters motion.

I have been having trouble making the motions seem natural. Even though the model is rigged very similar to a human there are still several areas that are different making it that little bit harder to rig. One of the areas is the shoulder, here the shoulder and arm are not connected the neck and spine so have to moved slightly differently. Another area being the knees, due to the ‘A’ having the bridge across the middle I had to devise a way for them to correctly bend. I eventually decided the knee should bend below the cross bar.

To solve several issues of the mesh deforming in an odd manner I have been adding vertices around the joints to allow the mesh to deform a lot smoother.

Another of the problems was adjusting the weight of the effect each bone has over every vertex. Some of the shoulder vertices in particular were deforming in an unnatural way and I had to do several tests of changing the weight between the forearm and the shoulder until it began to look correct.

I think the model is now ready to be animated. The best test the model could get is for it to be used in a real animation and see if any issues arise so I am going to deliver it to the rest of the group and see how they get on with it. I am confident that if any further issues arise I can fix them on the fly in the individuals animation and then use information gathered there to update everyone else’s animations.

Now on to the YMCA project

On to the YMCA indeed. Its been a while since I actually went in to any detail on this so here is the story so far. My group have decided to each create a separate room in the YMCA and create our animations around a single stylised theme using an animated YMCA character.

I originally created this character a few weeks ago, since then I have rigged a skeletal structure for it, I just need to finish tweaking the rig.

We have also visited the YMCA and measured the building from which we have created a scale model in Max for everyone to work from. This way everyone’s characters will be in the same proportions with respect to doors and windows etc.

Last semester I actually created a few separate scenes each with character animation in it and came across some serious complications when trying to merge the animated scenes together. For this reason we have decided to create our own animations and then merge them in to one seamless animation using premier pro. However this is not without its own set of complications to consider:

1. Camera settings – every one must be using the same FOV on the camera or the scenes will appear to shrink and grow as we move between environments


2. Character scale – all characters must be the same scale in relation to the height of the room and doors. To resolve this we have agreed on 3 different heights of our characters and once I have completed the rigging I will send round a copy of the model with one of each size character in.


3. Start and finish points – To ensure all the animations flow smoothly we need to make sure one animation finishes in the EXACT same place the next one starts. To ensure this is the case we have devised a clever little system. Each person has the same copy of the whole building and will model and animate in their own room with out moving the room at all. Each animation will start looking at the door of their room, animate through an opening door, around the room and end up looking at the same door (now closed again) from the inside. When all the animations are complete we will as a group animate the camera from the last persons finish point to the next persons start point through the buildings corridors.

To name but a few.

On to my progress and future plans.

As I previously mentioned I have created and rigged the YMCA man. There are a lot of problems with it at the moment though. As you can see the mesh ‘kinks’ when moved too much.

I believe this is being caused by two factors. 1st is that the mesh does not have enough segments around the joints to deform smoothly

and 2nd the absolute weighting of a particular bone over the central vertices is incorrect.

I need to solve this problem soon so that I can email the finished characters in the model to the group so they can begin their animations. Once they have the model I will have to teach them to use it as they have never animated a character using this particular style (I may do a video tutorial for this so people can watch at their own pace as many times as they want). (Self imposed) DEADLINE: Thursday 6th November

For my animation I plant to have the camera enter the room whilst some music is playing and show it for a few seconds before changing to another genre of music and then repeat the change another 2 times whilst in the room to show how diverse the events are. I need to pick some music before I can begin the animation of timing the lights. (Self imposed) DEADLINE: Wednesday 12th November.

I need to have the scene modelled but not necessarily textured soon. (Self imposed) DEADLINE: Friday 14th November.

I need to have the animation ready to be rendering for a completion DEADLINE of Sunday 30th November as this is when the group has decided we will begin the hallway animations.

Learning curve.

I’ve got a few things that I am going to need to learn to complete this module to the standard I would like.

1. Character animation – Last semester I used a similar rigging system to animate humans but this is slightly different. We all know how humans move, we see it every day and with a little study it is relatively simple to replicate movements in 3D. This character however, whilst it has a lot of similar aspects it also has ‘bits’ where people don’t and I am going to need to completely design a movement pattern for it to avoid an un realistic (that’s right, the human brain still expects it to walk in a certain way even though they have never seen it walk before.) movement pattern.


2. Light animation – I dabbled with lights last semester but a 30 second animation took me over 60 hours to complete. This animation is going to have to be really well timed to the music for it to work as the lights help to set the atmosphere. This is going to be a challenge.


3. Choreography – Again in time with the music I am going to have to animate several characters whilst making sure they don not bump in to each other.


4. Multi pass camera rendering – This is something I have often heard of but never got around to learning. It is the process where by the render makes one image, then moves the camera along its motion path slightly and composites the new image over the old one to create the frame and a motion blur look. Whilst it is not essential for me to learn this for this animation there is no time like the present, It should help make the image look more realistic. I intend to animate my camera in a way that motion blur actually forms a pivotal part of the scene and is not just there for the sake of it. It will be like pre post production or something.

Have I bitten of more than I can chew?????

Only time will tell.