Conclusion

I have learnt more than I ever thought I would in this unit.

The book I have been using (Mastering Autodesk Maya 2011 by Eric Keller) has been such a good resource. It has given me little tips and trick and helped me progress. It has also given me ideas about what else I could try even if it has not directed me how to do it.

Some of the tests have been less than perfect and more than a few times have I just wanted to give up. However, more often than not, the test have come out much better than I thought they would after a lot mor persevering.

At one point I did completely run out of ideas about what to do because I thought I had exhausted all my options. It just shows how important reference and research is because I probably would not have touched the ocean shader again after the first couple of tests but it actually turned out to be the best option.

So a fews things I have learnt:
1) Never underestimate Maya
2) Fluids will never render or even simulate on my computer
3) Realflow is definitely a good second place option
4) If in doubt, look for reference
5) Research, research and more research

The first and last thing I tested was using the ocean shader. From early tests it did not look like it was going to work but I should have researched and looked at reference sooner because I managed to produce a good river and find a solution for objects that need to pass through it i.e wakes. Not only that but I also managed to make the log float on the river. This means that is does everything we need it to do to make our film. It does not take too long to render either so in the long run, will save us a lot of time, energy and frustration. I think I could still explore this more. As I have said before, I think I would not be completely satisfied until I have tested it using the actual models but time and the fact that they havent finished being modelled yet, has not allowed me to do this.

Maya fluids were good to try but did not work at all in the end. My computer just kept crashing before I could do any real tests with it. It would not even simulate most of the time. They could have worked but it would have been a lot of time spent trying to set it up and test and we would have certainly needed to use the render farm. I think because of the time constraints of this assignment and the film, I dont think it would have been a good idea to use them.

Realflow was something I really enjoyed using. It was something new and different to learn. I was surprised at how easy it was once you knew how. The effect it created was really good but I would of liked to have the opportunity to render it using mental ray in maya just to see how the finished piece looked. It was also a lot of money to spend on an educational license if I only used it for our film. I think if the ocean shader had not of worked out then I would definitely use realflow and spend the money on the license because it is a million times better than anything else you could use. It did not take too long to simulate but I would have liked to have checked rendering time. Unfortunately this was an option that I did not have the resources or finances to test properly but was definitely great to use.

Another method that was spoken about but that I did not test was layering 2D over 3D. This was something I did not really have the time to test as all my focus was on rendering the other tests. Rendering took up at least half of the project time and I spent far too much time on working issues out with my maya crashing and with my computer. I think if it had all gone perfectly and nothing had crashed or frozen or I had not run into any problems, I would of had time to explore 2D over 3D animation. To be honest, I dont think I would have been the best person to try this anyway as I'm not very good at drawing.

I think if I did this again, I would research a lot more and get all my reference before testing. I would also make sure I had a computer that would be able to cope with the rendering. I could have used the render farm to get at least one fluid test done but time constraints and the fact I had other projects to do meant thatI completely forgot and then when I remembered it was too late. This taught me to have a plan of action before I start. I would have tried to contact more companies as well. I did think about doing this at the start but again with the other projects, I completely forgot and then Dan mentioned it and tried to contact Philip Hunt (Director of "Lost and Found") but it was too late and I had hardly any time left. This was another lesson to keep organised for any other project I am doing.

It was a good learning experience and one that I can use to help my film when it does come to the river scenes. I cannot say it went smoothly because I ran into more problems with maya than I ever have done in the past but if things had not gone wrong then I dont think I would have got as much out of the experience.

Making things float

I was trying to think about what else we needed the river to do and remembered that there is a log floating in the river that the cat grabs hold of. When I was reading the book Mastering Autodesk Maya 2011 by Eric Keller I came across how to make things float in the ocean shader.

I created a basic log shape using a cylinder and then moved it so it was half in and half out of the water. I then put it at a slight angle and, once selected, went to "Ocean - Float selected Object". Voila! My log now floated in the river. I rendered out a scene and this was the outcome.



There is only very subtle movements as the river is very slow moving but the log is definitely floating. At the very beginning though, it does this unnecessary bounce which I had not told it to do but it does anyway.  It is very small and hardly noticeable but I noticed it and now cannot stop noticing it annoyed me. I decided to go delving into the settings to see if it was something in there. These settings are agin very sensitive because if I changed something by the smallest amount, it made the log come out of the water of go under or bounce around even more. I reverted back to the original settings and saw a setting called "Start Y" which was set at -0.064. I thought this referred to a position and my logic told me it was probably the tart position. I started to change it and saw the log jump up the the air again and back under water. In the end I set it to 0 and it actually worked. I did a render with it at zero and the bounce had pretty much disappeared. I tried to get it to disappear completely by changing the setting by 0.001 each time but it kept making the bounce worse again. I left it at zero and even though the bounce has not completely gone, it is barely noticeable.



As with the other test the movement is very minimal but that is because the river is a slow moving river. It still looks slightly static but I think that is down to the model. It is just a basic shape so looks solid and static but if it was a proper model of a log it would would blend in a lot more and be more believable. I did not use a model of the log as it has not been made yet but I will do this test again later on in the year when the model has been made and see if it works just as well.

Adding Wakes

I have been researching and trying to make objects move in water. I knew that this could be done with fluids and particles but I have already tried to just create still water and it was just not feasible because of rendering times.

I was poking around Maya again and found the "create ripple" button. I tried to deconstruct it and work backwards to see if I could use it when trying to animate the animals through water. After further inspection I deemed this not possible as it would not look very good and not be the effect I was after.

I started to mess around with Maya again and tried creating some wakes. The wakes Maya creates automatically were far too extreme for what I was after but again I messed around with the settings. I did a really simple test with the "create ocean" button and then "create wake". I attached the wake to a small cube and then animated it to see what it would do. I used the "Preview ocean plane" to get an idea. The first result was quite shocking because of the size of the wake but then the small test that I did made me realise that this could be something that can be used for the film.



I then had to produce this on a larger scale because it was ok doing it with a small cube in a large ocean but I needed to do it in a river with a much larger object.

I went back to my river bed scene and tried to create an object and then add the wake to it. It was only a cuboid but it was about the size the ox was going to be. This was to give me an idea about how to adjust the wake to make it look like and ox was passing through the water, which are very slow and large, instead of a light small cube.

Once I had added the wake to my object I was trying to preview how the wake would react. It was not doing anything, The cuboid moved but the wake did nothing. It made no effect on the river at all. I tried turning the "Density/Voxel/Sec" setting to maximum (which is the setting I had used in the previous test to change the size of the wake) but it still did not work. It wasn't until I zoomed out a little bit that I saw a box similar to a container for fluids. I selected it and the notes explained that it basically was a container. Because it wasn't over the wake I presumed this was why the wake was not working. I tried to move the box over the wake and this did not work. I tried to scale the box so that it covered the whole scene and this did not work either. My wake would not work at all.

I was getting pretty frustrated so I thought rather than scaling up the box, why not scale down my scene. At this point the scene was about 8 times larger than the grid. I scaled down my scene so it fit onto the grid and therefore fit into the container without having to touch it and tried to add the wake again. This time it started to work. I honestly have no idea how or why this would not work before. I don't know why it would make a difference if I scaled up the container or scaled down the scene but it obviously did because it worked.

Now I could concentrate on adjusting the wake the settings. I thought about an ox and how the ox would effect the wake. It would not be a very large wake as they walk slowly through water so would not make a large impact on the water.

The settings were extremely sensitive for the wake. Even a 0.050 adjustment meant the whole scene changed when it was rendered. It would go for a massive wake to no wake at all. I was trying to think about and ox's wake but mostly, all I kept achieving was this:



And as you can see, there is no wake at all. I persevered and kept changing the settings and eventually I managed to achieve something that could be used for the ox.



The problem with this test was that the cube was going quite fast across the river. Again there was a very small wake behind it which you did not see until the cube was at least half way across. I decided to try one more time. After adjusting the settings again and again, I found it hard to get an exact wake that would work with the ox.



I think this test has been the most successful because it creates a small enough wake but it is still obvious there is one. If the cube was going slower it, the wake effect is exactly what it would look like when the ox crosses the river.

I then had to adjust the settings to create a wake for when the cat swims across the river. The cat does not start off in the river, but after getting pushed in by the rat, she has to swim to the log and then to the edge of the river.

I adjusted the settings in the same way that I did for the ox. After a while, I managed to achieve the exact wake I was after for the cat.



It has got that effect of swimming and has small and quick strokes because the cat is a lot smaller but faster so would swim faster and kick quickly giving that short flicking motion. In the tests it is the same size as the ox but in the film the ox would be much larger and the cat is much smaller but this will not be a problem as the settings will still effect the smaller object in exactly the same way as in this test.

I think if I had more time, I would do these tests again but with the actual models. I would use the ox model and the cat model and see how the wake effects work with it all animated. The problem is that none of the models are completed enough to be able to do that. Now that I have this understanding of the wakes and how to use the ocean shader, I dont think the river scenes will be such a problem anymore and I can easily transfer what I have learnt to the film. I still think some things need tweaking but I dont think there is any point trying to get this exactly perfect using the cubes when the geometry of the models will change how the water effects will look. Something that could look perfect on a cube will not look good at all when using the actual models.

I think these tests have been successful and have helped me learn. I think I underestimated how sensitive the settings can actually be. It has taken me a long time to work on these tests as my computer is not especially fast so rendering time has ben a killer. I think all the tests have helped me to understand and to improve. It has definitely given me a better understanding of the ocean shader.

Back to the Ocean Shader

So I thought about the ocean shader again and think there is a lot more that I can try and that I can do. It seems to be the most logical method at the moment because it is part of Maya so not expensive and would cut rendering time down by so much. I think it is definitely worth another try.

Firstly, I had to create a river bed for the realflow test so to save time I decided to use that as my base. I added a plane and then assigned the ocean shader to it. I tried what I did last time and that was to just mess around with the settings and see what I could achieve.

This was the outcome.



Again not at all what I was looking for, Firstly it does not look like a river t all and just looks like my previous attempts which did not work out the way I wanted at all.

So at this point I am wondering what I could do to make this time around better and different. I decide to look at some reference images. I have no idea why I didn't do this before but I managed to find some really good images showing me the ripples and tides of the river.

This immediately showed me where I was going wrong. For a start the waves I had created were far too large and secondly, The waves I have created were only going in one direction making it look more like a tide that you would see on a beach than on a river. 

I really needed to make this look like a good river before I could try to do anything else so I constantly kept changing the different settings. This was a really long process because I had to keep rendering to see if my settings were successful. They were still coming out the same so I decided to research the settings a bit more and to look into what they mean. Whilst doing this I found the "create ocean" option in Maya. Just to try it, I clicked on it and it created a realistic ocean straight away. I started playing around with the settings and saw that there was a ocean preview plane on the ocean shader meaning that I didn't have to keep rendering and saved me a lot of time. I then began to experiment with how to add this ocean preview plane to my own scene. When I did this successfully, it made a huge difference. I could now see what the settings were effecting and how the river would look without having to render. This helped me to figure out which settings I was moving incorrectly and how to make the river look a lot better. Just with the help of the ocean preview plane, I managed to create this river instead.

The river now looks a lot better and more realistic to how a river would really flow. It look slow but it also looks really strong which is the look I  am trying to achieve. This is only a test but i think it really works in the environment and I am confident that we can make it look good using the ocean shader.

I now need to figure out how to make things move through it.

Realflow 2012

This week I have been learning and testing using Realflow. I managed to download the free trial which is only valid for thirty days. This means I have had to learn very quickly and find my way around the program so I can start testing with it.

As it was only a free trial, there are a lot of limitations with it. For example, I can import objects into it from Maya to set my scene but I cannot export the simulation data to add the my maya scene to render it.

I began by using the container of the particles to make a basic seabed shape to see how the particles reacted. When using real flow, the particles are have to be emitted from an object, I could not find a way of just creating a body of water in a position. When the particles were emitted from the object, they fell to the bottom of the container but then bounced and did not react how I expected.



This did not make it look like a river at all and meant that the particles would just bounce around my scene.

I started again but this time, instead of using a container to b the river bed, I created a simple one in Maya. I then had to import this into realflow. This changed my method as instead of creating a small container for just the water, I created a large container to fit my mesh as well. I then had to do the same as before and create an object to emit the particles from. The results I achieved this time were a lot better than the first attempt.



This test was a lot more successful and made me think that using realflow could be a big possibility. I tried to then export the particles to Maya so I could render it properly to see how it looks but this was where the issues lay.

The trial does not let you export simulation data so I could not render it properly. This would be a bit issue for our film as we would definitely not be able to get away with using the free trials. This would mean having to purchase it which was over £100 for a student license.

It could be a good investment if it does everything that we want it to do but there was no way of going any further with the testing as you cannot export to Maya and therefore not animate anything going through the water to test to see if it worked.

It would work really well as a last resort as it is clearly going to work for the still water scenes and the particles react really well and naturally. I would have loved to have tested this further but I could not afford over £100 especially if we had decided not to use it. It also meant that no one else would have been able to do the animation during the river scenes as I would have been the only one to purchase the license.

Further Research

I have been doing some more research into different methods of creating water for 3D.

I have recently been looking at different companies that have created water visual effects and seeing what they use. Most of the research has come up with in-house software that they have made and created but a few other names of software did come up.

The first was Maya fluids which I have already had to rule out. The second was some software called Flowline. This has been created by a company called Scanline VFX which again have not released it for public use but does very much the same as Maya fluids as you can use it for smoke, clouds, fire and water.

Lastly, Realflow has come up time and time again. This is definitely something I would like to look into as method as it could save a lot of rendering time compared to the use of Maya fluids.

I have also been thinking about 2D methods. Mainly overlaying the 3D animation with some 2D to make it look like its splashing and rippling around the animals.

A good source of research was "the making of Lost and Found" by Studio AKA. They use water as a large part of the story as the boy rows the whole ocean to take this penguin back to his home. It was interesting to see the early tests they did with the 2D waves and how they had decided against against them because they needed "a space you could move around in". This is very much what our film needs too. Our film is largely set in or around the river but it needs to have that 3D feel because the audience need to feel the strength of the river and feel that depth, which could take much longer to achieve and blend with, in 2D

Whilst looking into different methods, I remembered that one look that they were trying to achieve with the film was toon shading. It needs to be thought about more in depth but, if toon shading was to be used, most of the methods I have been talking about would be useless. Most of the methods use there own shaders to give the realistic water effect but if it was to be toon shaded, these shader could no longer be used and it would not look like convincing water.

If this was to happen, the only method I would be left with is to draw 2D over the 3D animation. This would allow me to use a plane or something similar that could be toon shaded with the ripples and splashes drawn onto of the rendered clip.