Level set framework thesis

We have a new office!

2007-01-24T16:58:00.000+01:00

After a nice Christmas break we are now back at our new office (see picture below).
We have now completed the design and will start on the thesis report next Monday after a quick weekend trip. Hannes and his girlfriend Ingun are going to Stockholm and Anders is going to Odense, Denmark.
It is now looking much like a real framework, we have documented every little thing in it and created a really nice help documentation file where users can find more information on every class, method, property and also find some code examples of how to work with the framework.
We have designed it so that it will be as general as possible with some restrictions. The user can easily add extra functionality to the framework and exchange the base data structure which is the frameworks base pillar.

We have been talking to our examiner about the defense (swe: framläggning) and it will be some time around 1 march if any one wants to know.

Picture of the day is on our new office, we really like it! =)

Thats all for now!
Next week is the week for Word to be our best friend. =)

Merry New Year and Happy Christmas!

2006-12-24T07:42:00.000+01:00

Merry Christmas to you all that reads this blog and hope you have a nice holiday, I know we will even though we both have an exam in the beginning of January.

We made e redesign work of the framework; right now we are just having some performance problem. We will try to fix this in January when we return to the thesis. The framework is now build on modules that can be replaced with new version if the just follow the contract for the different parts.

That’s all for now, have a great holiday!

.NET version is forming and santa is at the door

2006-12-18T14:06:00.000+01:00

So, our status right now is that a .NET (redesign) is forming and we have a little bit more user-friendly framework. Much left to do but at least we can say we have something. Working with this is great, actually I think both Hannes and I would like to continue working with our thesis work after graduation, if we could (anyone wanna pay us to do it? ;).

Well, we have also found out that there are no end to fixing things in a framwork, you can always make things better or just some adjustments. We'll se how much "cool" stuff we have for the presentation in the end.

If someone is confused about the erosion things we talked about earlier, we can only say that erosion earlier was dilation and that we now have both a dilation and erosion in the framework. If you din not understand that, don't bother :).

Well, we probably post one more time before santa, see ya!

Rendering examples

2006-12-08T22:45:00.000+01:00

Press the image to view some demo movies.

Note that the middle example is called Erosion but is in fact a dilation.
Well, end of the week and we are really pleasant with what we accomplished so far. Now we have a fully functional version on the C++ framework in C# .NET. We are currently measuring performance between the native C++ and C# .NET versions that we have of the framework and as far as our numbers tells us, there is only a smaller delay in the .NET version. That is really cool, I mean, REALLY cool, hehe. Adding to it is that the C++ version is fully based on templates and the C# version is currently a naive implementation using non-template objects. Meaning the .NET version might be slower, but that is to find out I the near future =). Next step is to optimize and re-structure the design to use the full power of .NET. We’re really looking forward to this, to show the magic of .NET and the simplicity that is offers.

We are rendering out images of the two versions to create movies that will show all the super cool operations. We started using Windows Movie Maker to create the movies out of the rendered images; it is so easy to do fun videos in that program. Try it out!

Picture of the day is from the two renderings. The one on the left is from the native C++ framework and the one on the right is from the C# .NET version. The two versions are identical on shade and vertices count after mesh extraction via a marching cube algorithm.
Dont let the bad compressed images fool you. We put this image together really fast.

Peace!

Success!!!

2006-12-05T10:55:00.000+01:00

We did it! We finally dit it! Today at 10:51 we have a working small version of a dense grid implementation in c# and .NET. It is far from perfect and optimized but it works!

yo!

NDoc rules!!!

2006-11-27T17:05:00.000+01:00

After some struggle with a lot of compiler errors we manage to close some big ones in the translation. We are aiming at having the framework in C# by the end of this week; we’ll get back on that. Today we also manage to generate a standard MSDN-look-a-like documentation out of out framework. This is really god and a large step in our designing. Now we can create a user-friendly documentation to our framework that users can turn to when they need help with the functions that we have created. We use a free tool called NDoc, version 1.3.1, with some modifications in the configuration allowing it to handle .NET 2.0. It is really cool to see how nice the documentation turned out and how easy it was to create it.

That’s all for today, we are really bissy bissy bissy, and now we have started working against the clock to have most of the design ready before the Christmas break. So bee god and Santa might give you a present.

Ciao!

C++ to C#

2006-11-21T17:07:00.000+01:00

Long time... no blogging. Well, now we are posting a new blog, sorry for the delay.

We have our hands full right now; we are trying to translate quite complicated, undocumented and optimized C++ code into C# and .NET. First we have to understand the structure of the code we are translating and then we have to understand some C++ specific syntax like triple nestled templates with some typenames on the top (?!). This is a though job, but at the same time fun and we learn a lot.
Honestly it felt very, very difficult and almost impossible a few days ago, but such things just makes us running even harder.
Our plan is to have translated the code in the beginning of Dec. and then start to redesign it so that we can use some .NET specific stuff to optimize it. Maybe we need to optimize some parts a lot, and then use C++ (unmanaged block in C#), but we’ll see about that.

Our office is comfortable, Hanne’s desk is in the sofa and Anders is in a very nice chair.

Tomorrow we are going to Skövde for Sigrad which is a Scandinavian Computer Graphics event. It will be a nice and hopefully an inspiring event.

Hello!

2006-11-16T10:00:00.000+01:00

We are not dead, we are just full of stuff to do right now so the blogging comes in second place. sorry. A small glimps of what we are doing. Currently we are implemeting a lot of functionallity thanks to the help Ken gave us, but doing it in a .NET way. We also set up a schedule and a time plan for the up coming, and last part of our thesis. So I guess we wont post to many blogs right now, just so you know.

Implementation and performance testing.

2006-11-02T16:51:00.000+01:00

We are terribly sorry for not blogging in what looks like a whole week. We have been so busy implementing and discussing the current design we have. We also have been building some performance project to decide what and not to use in our design. Trying out different types of temporary data storage in the IO interface and tearing our brains out with the interoperability between managed and unmanaged code. The solution were that we decided more or less to let this one go for now and just concentrate on the managed .NET framework. If there is time over we will get back on to this, or if there is interest in further development after examination. We now have one mesh interface that can speak directly to both DirectX and OpenGL.

Next up is to put together a beta version of our design so far and do some performance testing to see where and what to change or what to keep.

Over and out!

Exjobbing over.... ???

2006-10-26T17:54:00.000+02:00

Today we went to Meadia Markt here in Gothenburg. It's a massive electronic store where they got pretty much everyting. They had a 800'000 sek, 82" flat tv. It was so massive. They had everything from proffensional photo equip to advanced kitchen equip. It was a really cool place and becouse of the time it took to get there we are now working over.

Right now we are looking at different ways of implementing the design. We have created, and is still doing it, a couple of different designs using interfaces, abstract and base classes and the new cool thing in .NET 2.0 called generics. This to test performance of the designs to get the best. We will present the result when we have them.

Untill then, so song and godnight. Have to get back to work now.

So what is a level set anyway???

2006-10-23T16:47:00.000+02:00

Today we wrote an explanation to what a level set is and the difference in modeling with it to modeling with a normal 3d-mesh. It was really good to get it on paper so that we cloud define it and have something to describe to our friends and families when they ask of what we are doing. It will be a part of our final thesis report, but for now will use it to describe level set for people that we know and meet.

Here is a short description:
If we take a level set in 3-dimensions as an example. Think of an onion where each layer in the onion is given a value that represents the distance to a chosen “zero-onion-layer”. The level set function is the whole onion, each layer is a level surface and the layer that we are interested in is the “zero-onion-layer” which is the zero-surface. All the other layers in our onion are then representing as the distance to our zero-surface.

Last Friday we were on a speed trip to Norrköping on a meeting with our examiner to define our thesis plan a little more and to make sure we are on the same track. The meeting went really well, so now we are in gear two and accelerating.

To see wikipedias explanation follow the link.

That’s all for now!

Över och ut!

Defining our goals

2006-10-17T16:53:00.000+02:00

Yesterday and today we have gotten back to defining our goals to se if what we are doing and planning to do is doable. We sat down and started to discuss what we really want with all of this thesis work and framework. We made a PowerPoint presentation to show to our examiner and present what we think see as the plan for this thesis.

One question was: Is it possible to create a general framework design at all for level sets? Well, that’s what we are to find out. Many questions came up before we even came to the actual design questions.

A very relevant question is what data structure to use, if only one. There are a couple out there but as we what to design a top of the line framework; off course we want the best. But is this doable? On Friday this week we’re having a meting with our examiner to clear out some, and hopefully all question marks.

Microsoft Live 2006

2006-10-13T09:26:00.000+02:00

We spent a really interesting Thursday at the Microsoft Live event here in Göteborg. We got introduction to Vista, .NET 3.0, Windows Presentation Foundation (WPF), Windows Communication Foundation (WCF), Windows CardSpace (WCS), OpenXML and more.

We also spoke to some people at Microsoft about our master thesis and they thought it was really interesting and we are looking forward to further contacts there.

One other thing that made us really, really happy was that our education at The University of Linköping is a good piece of investment. From all the cool (or should we say coolest) stuff they showed during the day we hade done most of it.

When they demoed WPF they presented an example from The North Face (you find it here: North Face demo) and they had a jacket that could be rotated and stuff in 3D. This is Image Based Rendering and we have done that. Actually, I think it is a really good idea to create a “3D-scanner” where you can take some photos of an object and create a 3D-model for it, or create Light Fields of it. We could do that :).

WPF is also a technique to separate the logic from the presentation (GUI) and I think we are a competent bridge in-between these two parts. We both have the understanding for the programming part and also a good knowledge about usability and understanding for the design. But maybe we are not the people to do the real design.

In another demo for WPF they showed some really cool stuff about how to present information in different and dynamic ways. What they showed was a newspaper who presented articles in a tree map and in another way a social diagram in a hierarchical way like a hyperbolic tree. They also showed some other diagrams. We learn about this in the Information Visualization course.

They presented XNA which is the old Managed DirectX 2.0. This is, among other things, a game development environment for both PC and Xbox 360. This means that you can write a game that will work on both platforms. We have the game development course in Norrköping and we also have a lot of 3D-graphics during the whole education. A lot of project in other courses also end up as games in some way. Cool stuff and we can do it.

A short presentation of some research projects that Microsoft are working on was presented at the end of the event. They showed some Augmented Reality stuff and a way of interacting with a computer using your hands or projected objects. This is Augmented Reality, Virtual Reality and tracking in a nutshell. We have studied all of these areas.

So, what’s our point with all this bragging, do we think that we are better than Microsoft? No, of course not, but we are studying a lot of cool and upcoming techniques’ and we wonder why we don’t use more .NET and Microsoft techniques in our education? Maybe something to work on?! What do you think?

We like C#!

2006-10-11T17:35:00.000+02:00

We have been looking deeper and deeper into the input, and also out put, to the framework. After receiving yet another mail with answers of our level set questions we started to understand how it was used in the special effects industry. The answers gave us lots of information about the daily use of level set and what was missing in the existing usage of level set frameworks.

Tomorrow we’re going to a Microsoft all-day event here in Gothenburg. It’s about their upcoming products and packages. Really looking forward to this!

We will give a report tomorrow, or the day after.

That’s all for today!

// Anders & Hannes

Coding

2006-10-10T17:21:00.000+02:00

Today we have started coding. We are working on implementing some input formats, so it is all aboud creating parsers for different file formats. Now we are talking about robust parsers that should be general and not an ugly hack. Thats about it... the sun is shining and the sky is blue and we are sitting inside, programming...

Tjo!

Input to the framework

2006-10-09T16:42:00.000+02:00

Now we have received answer on our question forms that we earlier sent out to some people. Today we have analysed some of these and we have also started to work on the input-part of the framework. This includes opening and parsing different file formats. We will try to open both volume and mesh-files... this is what we will be doing this week...

Tjo!
//Anders o Hannes

End of this week

2006-10-06T16:28:00.000+02:00

Ok, I don't know if we have that much to say today. We have reviewed some design, added some classes and removed some other. We feel that we want to start the usability testing and also do some coding. Next week will be focused on input and read from file functionality, but first a nice weekend!

Tjing!

An Agile Development Process

2006-10-05T17:23:00.000+02:00

After a review of our progress so far we have changed some of our design. We have had some discussion about which parts we will include in the framework and then thought about using an agile development process in the development cycle.
What we have done so far is the overview of the whole structure, then in step 1 we will look at the input/read part of the framework. Step 2 will be the storage and data structure, step 3 the display, step 4 the output/write, step 5 the processing/operations and inbetween all of these steps we will stop and review the overall structure of the framework. Step 1 is planned to start next week with a first design, some spike implementation, testing and evaluation.
Then we will also iterate the whole process of all these steps, wow, I thing we will be busy for a while :).

You learn something new every day

2006-10-04T16:56:00.000+02:00

Our first discovery this day was the answer to yesterdays question about performance. This was a really nice piece of knowledge. The day continued with more designing and discussion most about the rendering part of our framework. We made a stop here to search for some ideas how to do this and we are currently trying to figure out some alternatives.

We did a deep dive into Ken Museth level set framework to see how they solve different parts of the solution. There are a lot of really cool algorithms in it that I really recommend people to take a closer look at and realize that some of it apart of the things we learned, and probably forgot in school. If you haven’t taken the course TNM079 Modeling and Animation at Campus Norrköping, we really recommend it. It’s all cool stuff.

Picture of the day is from Linköping’s local castle. We couldn’t keep away from this one.

//Lennart und Mats

Blog from Firefox

2006-10-03T17:34:00.000+02:00

Today we started to look more on the design to get a first version to validate. I guess we will make a lot of versions of this framework until we get to v1.0, the first release candidate. We also looked at performance a lot when choosing between abstract and virtual classes and functions. Anyone got a good answer, please tell us. Also we decided more our goal and restrictions with this thesis. It's good to look over them so we don't run of to some where else.

Edited:
Today we found the answer to our preformance question. If you are creating a visualization framework you need to consern about the structure and how your chosen structure will perform. We found a thread on this subject that discusses the performance of non-virtual, abstract and interface calls. After running a test program this is the result:
Non-virtual: 717,62 million calls per second
ABC: 309,01 million calls per second
Interface: 216,57 million calls per second

Quite a difference in cost, isn't it?!
End of edit

We found a new Microsoft event here in Gothenburg this time that we will attend to 12th of October. It's an all day event with lots of cool presentations about new releases from Microsoft including Vista, Office 2007, .NET 3.0 and so on, really looking forward to that.

At lunch we found a cool add-in to Firefox as well. It’s an add-in that let’s you create new posts in your blog without logging in to the blog. Really nice, Performance is it called and can be found here, (https://addons.mozilla.org/firefox/1730/).

New week, new stuff...

2006-10-02T16:24:00.000+02:00

A new week and we were both pretty tired after a long weekend. This day we have finally finished the video design guidelines series, sent some e-mail to important people and last also discussed a lot of designing. The design we are discussing is who our users are and how every type of user want to interact with the framework. It is very important for us to get this input from the users to be able to create a user friendly framework. We hope to get some input from every group and then the next step will be to create a first blue-print of the framework that we then can test.

Visual Studio Team System

2006-09-29T00:49:00.000+02:00

Today we have been on a Microsoft event presented by SweNUG and Dag König. The presentation was "Visual Studio Team System from A to Ö". Since we are searching all possible information about .NET this event seamed like a god one to attend.

VSTS is not only a develop environment for developers, but also a tool for the whole development process, from planning to deployment. The lecture had some interesting moment but as for our thesis work it might not give us that much. The good info was more for us to have for later use at a company.

//Anders o Hannes

Designing in the rain

2006-09-27T16:58:00.000+02:00

Back in Sweden from our trip to Denmark we are now preparing for the trip to Stockholm, to the Microsoft event. But first, the trip to Denmark.

It started as every journey does, whit an early morning at a train station. Well first there was the even earlier stop at the cable car station out side of my house. At 06:02 it all started. First the cable car down to the train station, then the train to Malmö. Here we switched train to go over Öresund to Copenhagen. We slept most of the way, but discussed a little design as things popped up in our heads.

Once in Denmark we went to have our second breakfast. The time was now around ten a clock and we both where longing for something to put in our empty stomachs. After about one hour of shopping, none of us found anything actually, we head of to wards Odense. Odense is in the middle of the island Fyn, Denmark’s third largest island if you don’t include Greenland. Here we met up whit Anders sister Carina. She’s a medical student in Odense. After a quick pasta meal we then went for our goal, the beer.

We found a grocery store called Netto that had really cheap beer, only 2.75 DK/bottle. That’s something like 3.43 SEK/bottle and that’s cheap. So around 150 bottles later and a quick visit to Carinas apartment, that which by the way was amazing, we were on our way home, Home to Swedish radio and Max burgers.

Today we’ve studied more .NET lectures held by the founders of .NET. It’s really paying of and as more pieces falls in place more are also added. We learned how FxCop works and how to use it. It is kind of having someone looking over you shoulder when you developing and telling you how you should code and how you should name members, classes and other parts. Really cool feature actually, look it up if you’re interested in development in Visual Studio.

To morrow we will start our journey to Stockholm, la capital! We will pick up a mister Arrelid in Linköping and the head of to Stockholm to attend on a Microsoft event there. We will also meet up with Mikael Uddholm to try out his gymnastic court. See if that back flip is still there =).

Picture of the day is of our office at my place.

Live strong and prosper!

// Hannes å Anders

Back

2006-09-26T22:54:00.000+02:00

...now we are back from our trip to Denmark. We just had to post here to make Mattias happy (he asked for posts EVERY day).

//Anders o Hannes

A new week

2006-09-25T17:10:00.000+02:00

Today a new week started and it started with some planning and mailing. We are trying to get our usage question forms in and also to prepare for some interview. Our days are pretty full already, tomorrow we will go on a kick-off journey to Denmark and on Thursday we will go to Stockholm for a Microsoft event. We are really looking forward to this event because they usually give a lot of nice and tasty information. On the way to Stockholm we will drop by Norrköping for some meetings.

And as usual we studied more .NET today.

//Anders o Hannes

The designing begins

2006-09-22T15:58:00.000+02:00

Finally, today we took one small step for man, one giant leap for mankind, i.e. we started the designing phase. You know, it is not easy to design a large system. Where do you begin? How much should you do? What is important in a first release? How much time do you have?

Our designing kit!

We have consulted the Visualization Toolkit handbook about this and read some chapters about their process of creating VTK. It took four professional developers 10 months to design 25 classes, without even producing one line of code. Our great team are two man strong, that gives us 10 months to create 12,5 classes or 7,5 classes in 6 months.

However, today we went for some shopping and bought a big white paper (72x104 cm) and a lot of colorful post-its. Then we started the design. So where do you start the designing of an API? Well, as we have told you in a previous post, usability is the most important thing! We started to look at how we wanted to use the API, that is, how would we like to create a level set object and which methods should a basic version of a level set object have? It should be the users that should tell us these things, but we start with something and then ask the user for comments and input.
That’s the start of the designing, and you can read about the result in the report that will be finished some day in the spring of 2007 :).
From now on we will read in the mornings and then design in the afternoons. Except for a few days next week, when we are going to buy beer in Denmark and then go to Stockholm for a Microsoft event.

Have a nice weekend!
//Anders & Hannes

Like packing your sleeping bag

2006-09-21T17:21:00.000+02:00

Today we agreed that our days are like packing your sleeping bag after one night of good sleep. In the morning it is easy, we learn a lot and everything is registered in our small but effective brains. This is just like starting to put the sleeping bag in the container, everything goes smooth. The later in the day, the harder it gets to learn and the brain starts to get full of important .NET design guidelines together with some hard core level set algorithms. At the same time you struggle to stay awake. After a few cups of coffee you are back on track and can push the last piece of sleeping bag in the container before the day is over, and you just have to tie the knot of the bag.

Picture of the day - Our brains...

Today we have learned the difference between reference and value types in .NET, how and when to use enums, the difference between enums and flags, the 80-20 case, generics, boxing, exception handling and other interesting stuff about .NET. Actually we start to get some meet on the bones here (directly translated from Swedish). Our self-confidents in the .NET framework are starting to grow and we now know how the framework is working under the cover and this gives us the opportunity to use the different framework components in an efficient way.

What is .NET 3.0 and more design guidelines

2006-09-20T16:51:00.000+02:00

First of all, what is .NET ?

.NET is the Microsoft strategy to connect information, people, systems, and devices through software. Integrated across the Microsoft platform, .NET technology provides the ability to quickly build, deploy, manage, and use connected, security-enhanced solutions with Web services. .NET-connected solutions enable businesses to integrate their systems more rapidly and in a more agile manner and help them realize the promise of information anytime, anywhere, on any device.

The .NET structure contains class libraries that developers can use to build their applications without programming everything from scratch. Up today there have been three real releases, 1.0, 1.1 and 2.0. A new version, 3.0, is coming soon and this version is built on top of .NET 2.0 with the extensions Windows Workflow Foundation, Windows Communication Foundation, Windows CardSpace, and Windows Presentation Foundation.

We have today read an introduction to .NET 3.0 (Introduction to .NET 3.0) most because we are thinking about doing some of our demos using version 3.0.

More information about design gudielines has also been collected and we are now watching a video course about “Designing .NET class libraries” held by some of the authors of our black book.

Yesterday we missed the top event of the day, our lunch at Mongolian Barbeque.

Todays top event has been that Anders has got a new mobile phone! Yeh!

//Anders & Hannes

A summary of our progress so far

2006-09-19T17:11:00.000+02:00

This (our super duper blog) is our way of telling the world what we are doing in our master thesis and also a way of telling everyone if we are doing any progress or not. To summarize what we have done so far we can tell that most of the past two weeks have been reading, reading and more reading. We have also sent a few question forms to people that are using Level Sets today or that might use our framework tomorrow. One of the most important things, or actually the most important thing, when designing a framework is usability. We want to know how our prospective users want this framework to work, so that’s one task on our schema right now. This is an iterative process so if you want to help us designing a nice and usable framework, please send us an e-mail!

One other thing in the pipe is setting up all tools for the development of this project. This includes UML-tools and Visual Studio with proper plug-ins for documentation and code consistency among others.

Sources for .NET are infinite and the challenge is to sort the good ones from the bad. This is what we are doing and one source that we can strongly recommend for a framework designer is our “Black Book”. This is a book full of Do and Don’t and a lot of comments from experienced developers and designers of .NET.

Black Book
Title: Framework Design Guidelines – Conventions, Idioms and Patterns for Reusable .NET Libraries
Author: K. Cwalina, B. Abrams
ISBN: 0-321-24675-6

//Anders

First blog post

2006-09-19T14:29:00.000+02:00

Well, here we have it! Our own thesis blog. We will post a new post everyday at around 5 PM just to have people know what we are doing and to keep track of what we have done.

//Hannes