I attended a talk on the Google App Engine in Seattle on Monday March 30, 2009 hosted by the Washington Technology Industry Association.  It was one in a series of talks on Cloud by industry leaders.  Mike Repass, a product manager for Google, was up from the Bay to present.  What follows are my notes from the presentation with a bit of commentary.  [Note, I have not used Google App Engine myself so I'm no expert, but thought the talk was well done.]

Google App Engine Primary Value Propositions

The primary value propositions Mike notes for Google App Engine:

No Assembly Required - To get started you need a Google provided SDK and Python installed.  They support Mac, Windows, and Linux platforms for developing your application.  The run-time environment is a Google version of Python which implements 90-95% of the Python API.

Easy to Scale - For security and scaling purposes some portions of the Python API have not been implemented in the app engine run-time.  Applications must be stateless to support the auto-scaling capabilities.

Free to Use - Mike talked about the free usage of up to 5 million page views.  Five million page views is just marketing speak for usage based quotas that include both volume and rate.  For example # of requests total and # of requests per minute.  Details can be found here, but I did note on the blog that reduced free quota levels will take effect on May 25th with the following changes:

• CPU Time: 6.5 hours of CPU time per day
• Bandwidth: 1 gigabyte of data transferred in and out of the application per day
• Stored Data & Email Recipients: unchanged

Five Core Components

The App Engine is comprised of the following five components:

1. SDK + Python - A 2-3MB download includes Python libraries, command line interface, and docs.  Basically, the libraries are Google Python extensions, and CLI allows for easy app uploading.
2. Sandboxed Run-time environment + Your Code - A run time environment created with a custom version of Python that implements most of the python environment.  Things like direct access to the file system and other system features are disable to enforce the sandbox rules.
3. Datastorage via Python APIs - An API and library that gives authors access via standard Python APIs to Google’s Big Table data storage system (non-relational, no fixed schema).  Does not handle blobs/large files today.
4. Servicing Architecture - An architectural framework which allows properly architected Python applications to be scaled up/down based on demand including handling all traffic/network routing.  To do this, python applications must be stateless.
5. Administration Console - A web based GUI to monitor your apps, make configuration changes, and handle billing

App Engine History & Examples

• Launched in April 2008
• 150k developers
• 50k applications
• 100 million page views
• Example: Buddy poke, 1 developer with 36 million users

The main web site for information is appengine.google.com.  They also expose all of their operational data including latency and service quality.

App Engine Fit

At this point, Mike discussed a framework in which to consider the App Engine’s fit. He was clear this was not intended to be a better/worse competitive framework, but only a way to look at where App Engine fit vs. other services (i.e. Amazon). These were the categories he defined:

• Owned Machines - Capital expenditure required.  Owned and managed by the customer.
• Leased Machine Physical - Not capital expenditure.  Managed by a 3rd party.  Full access and full flexibility to do whatever you want with the machine.
• Leased Machine Virtual - Not capital expenditure.  Managed by a 3rd party.  Leasing the compute capacity more than physical hardware.  Full access and very flexible.  This he said was where EC2 fit.
• Managed Machines - I didn’t get this one)
• Managed Platform (non-machine) - platform as a service.  This is where Google App Engine fits.
• Mash-ups - no programming

I find this breakdown of categories to be off.  There are several axis being discussed here that really are separate and independent. For example, there will be owned solutions that will deliver compute and platform as a service within the Enterprise.

He noted that legacy applications and applications that require integrations with other software and hardware systems are not good fits for the Google App Engine.

In summary, the pros are you don’t need to worry about hardware and its configuration.  The cons are you give up flexibility and control.

Six Month Roadmap (from March 30, 2009)

• New language and new run-time environment (no leakage of the actual language.  April fools blog post says Fortran, I think not.
• XMPP support (for jabber client apps), Mike also seemed to position this as potentially useful for inter-app messages, although seems like a poor mans implementation as opposed to Amazon SQS
• Cron - support for scheduled hits to an URL to drive work processes
• Large file/blog storage for images, mp3s and the like (must be a lot of demand for this)
• Background workflow / task queues - the limitation cited was App Engine was good at traditional web applications, but not easily used for background/worker processes.  The idea behind this is to support some architectural work to make this kind of application or subset of an application easier to create
• Incoming e-mail support - apps can listen to incoming e-mail
• Datastore export - many customer fear vendor lock in with these cloud platforms.  This allows you to extract your data so you can use it elsewhere.  Lively discussion on the usefulness of the data without the business logic behind it.

More on Billing

• Rates and daily free usage based on 5 primary areas bandwidth in/out, storage, cpu, and e-mail users
• You can down grow beyond the free quota by enabling billing (I noted earlier free quotas are to be reduced in May)
• You set a billing budget, your costs will never grow beyond your budget (Customers will experience a Quota Denied error once your quota has run out, is this really a good experience???)  There is also no proactive alerting mechanism to let you know you are about the run out of quota.  Although Mike noted several technologies were being investigated including being able to code to budget (dial back application usage based on current quota thresholds)
• Maximum billing budget is 500 hits/second and up to 1B hits/month.  If you need more call Google directly.  Cited example of whitehouse.gov.
• If you register your domain with Google, you can use your domain for your app vs. the app engine provided domain

What followed was a demonstration of building and deploying an app and some final Q&A

• Q: Can users tie into Google Earth/Google Maps A: Not today, would like to get there
• Q: Can I port existing applications to the App Engine  A: The easiest existing apps to port are those based on Phython web frameworks such as Django.  The main challenge is the datastore.  There was an open source project built by a Google intern to extend the Django framework with APIs to Googles BigTable data store.
• Q: How will you handle Phython version upgrades?  A: You will select Phython runtime you want to use in the app.yaml file.

The main question posed here is how to setup Opensim on AWS EC2 and what might be the costs of such a deployment.  If you know of any prior work in this area please let me know.  As time permits I may do a series of tutorials if there is interest.

I am assuming for the sake of this post you already know about Amazon EC2 and S3web services, but perhaps not how they work.  I started thinking (and I’m sure I’m not the first) could you run a virtual world grid business using Opensim on EC2 and would a tutorial on how to set this up be of interest?

I had three main questions going in (I’m sure there are more):  

• Question #1 - Setup
  Each Opensim region is practically like the other, except with some tweaks to config files.  Is it possible to have one standard region Amazon Machine Image (AMI) that you clone then tweak once it is running or would you have to automate the construction/upload of a custom image for each region and pay for storage, etc.?
• Question #2 - Bandwidth/Cost
  With EC2, you also pay for bandwidth.  How do you go about quantifing the bandwidth required on average to support an Opensim Simulator?
• Question #3 - Database
  There would be one or two servers handling avatar and inventory services and then multiple region servers.  Opensim supports a variety of databases.  Is the Amazon storage service (S3) an option for Opensim storage?

I asked a Jeff Barr an AWS evangelist about these questions.  His response ”Wow, these are great questions.”  He went on to provide some thoughts on how I might run these questions to ground.

For #1, You should be able to create a generic OpenSim image. The image
would use a runtime parameter to determine its identity. There is a
standard EC2 mechanism for passing and receiving such parameters.

For #2 - You could take an otherwise idle Windows machine, start up the
SL client, then open up the Windows Network Connections dialog. From
there you would find the icon for your network interface,
double-click on it, and note the initial byte count. Then you would
log in, do some TP’ing, and so forth and watch the byte count rise.
You would then be able to measure bytes/hour and use this to
estimate your bandwidth costs (multiplied, of course, by the
expected number of users).

For # 3 - Using our soon-to-be-released persistent storage facility for EC2,
you will be able to run a MySQL database on EC2 without worrying
about instance persistence. I blogged about this a while ago.

Adam Frisby of DeepThink and a few others seem to have made this work, but found it costly based on this thread.  Adam blogs about this here.

As time permits and if there is interest, I will proceed with a tutorial.

I’ve gone dark for quite a bit due to other projects, but wanted to put a post up on several issues which temporarily stalled my OpenSim efforts.  My main goal here is to ensure others don’t run into these AND for someone to post links on better approachs/solutions.

I got OpenSim stood-up and running on Debian Linux.  I then sorted out how to configure the terrain to my purposes and setup physics.  Using the client estate editing tools I also sorted out how to build/apply custom terrain textures (haven’t posted on this yet, there are some good resources out there though). 

Then I invited a few friends from SL to come on over and start building.  This is when I started to run into very practical problems.  The main issues were around avatar and object persistance across OpenSim upgrades and restarts.  The first issue I ran into is somehow I had not setup persistance of objects in the world.  Things were built, but when I restarted the server one day, it was all lost. 

Test #1: Test object persistance with a friend.  Get another avatar in your world.  Build a few things together.  Leave.  Restart the server.  Login and see if things are still there.  Do this early before you create a master piece.

Secondly, when upgrading, I had used the default MSSQL for persistance.  I don’t really know how to manage MSSQL or have any confidence that if I backed up the data I could script my way through a schema change if needed.  So, I tried to move from MSSQL to MySQL which is more familiar territory for me.  This looked like it was going to be a major scripting effort and I could not find any tools that existed to do this.  So, once again I scrubbed my sim as part of the upgrade to get to MySQL.

Test #2 Test database backup/restore.  Once you have confirmed suggestion #1, create a backup of the stores.  Try to spin up a new instance of OpenSim using these stores.  See if you can do it.

I haven’t had time to do this yet, but it will be a real issue if I start inviting several SecondLifer friends into my SIM only to have them loose everything during some upgrade.  I expect there are some solutions to this problem which have been built mainly by other grid providers like OSGrid and the like, but nothing quickly found in a web search.  Send me info if you have it :-).

Finally, I ran into a physics problem which required me to rebuild ODE on Debian Linux in order to get up to date with the latest OpenSim trunk.  This exercise and subsequent debugging was where I had to leave it.  I finally decided to wait a few months and then do a clean install/upgrade.  It was a far cry from the fun of building out your own sim.

No surprise, but OpenSim is still early days.

I named my first OpenSim region Orcas after a real world island.  I decided to try and see if I could get the actual Orcas Island terrain into OpenSim to see how it looked and uncover any challanges one might face in bringing RL terrain into OpenSim.  This tutorial will walk you through the process I followed and and various software packages required to make this happen (done on Windows).  The net, net is I succeeded, but the results were uninspiring.

My thanks to Petrolia and this tutorial which got me started.

[UPDATE: See Darb Dabney's excellent follow-up post here which shows how to do this more accurately.  He completes a usable Orcas island with 54 OpenSim regions at proper scale!  Great work Darb love it!]

Step 1 - Find Some Terrain 

The US government maintains a great site which while a bit cryptic contains tons of data.  Point your web browser here: http://seamless.usgs.gov.  Navigate to “View & Download United States Data”.  Make sure to enable pop-ups you’ll need them.  Google maps it is not, but use the zoom and navigation controls to find a piece of terrain you want to put into OpenSim.

USGS Seamless Server

Step 2 - Select/Deselect Display Options

On the right hand side, there is a display panel with several drop downs.  Deselect everything (a bit painful) and select “NED Shaded Relief (1/3 arc second)”  This relates to the level of detail in the data.  1/3 is the more detailed than the 1 arc sec, but less than 1/9.  1/3 creates fairly large download and you may be able to get by with 1 arc second since this all goes down to 256×256 pixels.

1/3 arc second Elevation Selection

 Step 3 - Select Download Area

On the right hand side of the display select the “Define Rectangular Download Area” and then drag a selection around the terrain you want to use.  The process I describe in this tutorial fits what you select into a square image so the closer to square you select the less distortion you will encounter.  Once you finish your selection a pop-up will come up summarizing your request.

Step 4 - Modify the Data Request

You want to get a .BIL formatted file, so click at the top of the request summary page on “Modify Data Request”.  It looks like a column header but it is a link.

Modify Data RequestScroll down to the 1 selected download size and change from ArcGRID to BIL Change Download Type to .BIL

Now scroll all the way to the bottom and select Save Changes.

Save Changes

Your download should start, although you may have issues as I did with pop-up blocking settings.

Step 5 - Download and Install MicroDEM (free)

You now have raw terrain data in a .BIL file.  To work with this, you need to convert it from BIL to a height map which is basically a greyscale image where 0 (black) represents the lowest point in the terrain and 255 (white) represents the highest point.  The MicroDEM software lets you do this.

Download and install from here:

http://www.usna.edu/Users/oceano/pguth/website/microdem.htm

 

Step 6 - Open .BIL and Change to Grayscale

• File -> Open DEM (select the .BIL file from your download)

• I Needed to create “images” directory under c:\mapdata to get the save BMP to work.

• Modify->Elevation->Grayscale

• Modify->Grid->Neither
• Remove Legend Overlays by right clicking on the image and selecting “Legends/marginalia”

Legends Context Menu

You’ll see a dialog like this, deselect all the check boxes and click on “redraw map”.

Remove Legends

Save it to a .BMP with

• File->Save

Step 7 - Crop/Scale in Your Favorite Image Editor

You now need to scale/crop the image down to 256×256 pixels.  For this one I used the freely available GIMP, but Photoshop works well too.

The GIMP

 

 

 

 

 

 

Orcas Sized and Scaled

Step 12 - Load into OpenSim

You load the terrain into OpenSim from the OpenSim console with ”terrain load /path/to/file.png”.  For SecondLife you can upload a .raw file from the estate tools, but this does not seem to be implemented yet in OpenSim.  Converting from a heightmap to an SL raw file is a whole ‘nother topic.

My first pass ended up with a very craggy landscape.  The black turned into the floor of the sea with steep rises.  In short, the terrain scale I picked was far to large for a single region.  [I also learned later that you really cannot use the full range of black to white or you will get pretty unusable terrains.]

Craggy from Ground Level

I lowered the water level to zero to get a better perspective and took the following shot from above.

Aerial View With 0 Water Level

Step 13 - Make Adjustments, Rinse and Repeat

To get a smoother look, I used Gaussian blur filter in Photoshop and reloaded for the final results. [Reducing the contrast would have helped as well].

Terrain Smoothed with Gaussian Blur

This tutorial takes you through the process of hosting a public, standalone OpenSim server running Debian 4 Linux at the cost of $15/month.  Major credit to Adam Zaius and his [HOWTO] Get OpenSim running on OSgrid.org for cheep cheep without which this tutorial never would have existed.

Unlike my OpenSim server on Windows this is not really guide for non-developers.  To succeed you need to be familiar with things like Unix/Linux comand line, jed, ssh, putty, screen, and apt-get.  However, I have tried to give you enough detail that you have some hope of success, but no gaurantees offered.  So, let’s get started on the journey.

After spending some time in my new Windows Vista hosted region, I started to get lonely and well bored.  Perhaps there was some Second Life builder who could stop by and whip up a scripted sailboat I could use.  Trying to give someoneelse access to Sim from my home PC-based had a few challanges and the reality of hosting your own virtual world started to emerge.  The challanges mainly took the form of Comcast and their “high-speed” broadband internet access (quotes intended).

It may be high-speed for downloading, but it is more like low-speed narrowband internet access going the other way.  Coupled with the additional costs of a static IP address and punching a hole in my wireless router led me to the conclusion I needed to host the region in the cloud.  So, I went in search of cheap but reliable and fast, hosted server which would fit into a hobbiest budget.

Step 1 - Find a Place to Host Your Region

A dedicated Blade for $60/month — Almost all cheap web hosting options do not allow you to upload and manage a customish application like OpenSim so I started to look for a basic server hosting solution.   A good friend of mine pointed me at http://www.cari.net where you can get your own dedicated blade for the extreme deal of $60/month.  This gets you dedicated hardware, a single core Celeron D processor, 1 GB of RAM, Fedora Core 7 OS, and 160GB Disk and 4Mbps burstable to 100Mbps up to 1300GB a month.  Now thats pretty great for $60 a month.  However Windows was +$20 a month another gig was +$20 a month and so on.

Other options considered:

• Co-location (some services allow you to buy a machine and rent space in their datacenter)
• Amazon EC2 (more on this in a future post)
• Get IT friend to sneak your machine into their companies datacenter

I finally decided to start with Cari, but couldn’t stomach an extra $20 a month to run Windows.  However, before I committed to Linux I need to convince myself that my rusty *nix skills were up to the task.  This is when I found Adam’s excellent HOWTO referenced above.

Step one of Adam’s tutorial pointed me at TexTonic and a Virtual Private Server virtualization at work.  The barebones, unmanaged VPS is $15/month or 4x less than Cari’s bottom shelf.  By comparision this gets you 294MB dedicated RAM, 30GB disk space, CPU burstable to 25% of available CPU, and 500GB of monthly transfer.  Note, no meantion of what type of CPU is behind the 25%.  But at $15 a month I figured I could risk it.

In the time it has taken me to write what I did in Step 1, I had an e-mail message with my new IP address and root password.  A very easy and quick setup.  I picked Debian linux and based on my experience following Adam’s tutorial I think the Linux flavor you pick does impact the exact steps required for success.  I suggest you pick the one you are most familiar with as Ubuntu and CentOS were both options.  Otherwise, I suggest you pick Debian.

NOTE: I also picked a 32-bit operating system.  OpenSim supports 64-bit the process is much more complicated.

Step 2 - Login to Your New Server

To login to your sever you need a secure shell (or SSH) client.  The most widely used, reliable, and free is Putty download here.  One of the things I love about Putty is when you download it, there is just one file and it is NOT a zip, it is an executable.  Put it somewhere and double click it and it runs.  None of this run an installer, agree to terms and conditions, unpack and copy a bunch of crap to my harddrive, then twiddle registry keys so your presence on my machine can never be fully purged.  Nope, just one simple file. 

When putty comes up, simply enter your new IP address in to the field as shown below, leave the port to 22 and hit Open.

Putty Login Screen

Step 3 - Login and Load an Editor

When the login screen comes up enter ‘root’ for the user and the password TexTonic provide you.  You should arrive at the command line of your newly dehydrated virtual server.  Before you get all the packages you need to build and run OpenSim it is necessary at least on a TexTonic clean Debian 4 linux to update where you look for packages in order to get the right version of a little program called mono. 

WARNING: The wrong version of mono caused me extreme problems trying to build OpenSim and several hours of debugging so follow carefully.

Type the following to ensure your catalog of software is up to date.

apt-get upgrade && apt-get dist-upgrade

Then load the jed editor (or another editor of your choice) with:

apt-get install jed

You can now add one location to where your server looks for software to ensure you get the latest version of mono.

Step 4 - Add to and Update Your Package Distribution List 

I won’t get into how to use JED, if it is not obvious try some docs here.  Type the following:

cd /etc/apt

jed sources.list

Add the following line at the top of the sources.list file

deb http://debian.meebey.net/etch-backports /

There is a space between backports and the last backslash.  I discovered this little gem obscurely on the opensim site under http://opensimulator.org/wiki/Monodevelop.  Under troubleshooting for Debian 4 there is a section which suggestions you use all “unstable sources”  which can’t be a good thing.

Hit ctrl-x ctrl-s to save and ctrl-x ctrl-c to exit.  Now reload the sources for your packages with:

apt-get upgrade && apt-get dist-upgrade

Step 5 - Load the Necessary Packages

Using “apt-get install packagea packageb packagec” load all the following packages. 

apt-get install subversion nant, monoSubversion is only required if you want to pull directly from the source code repository.

OpenSim Documentation lists the following required packages for Debian 4

• mono
• libmono-corlib2.0-cil
• libmono-sqlite2.0-cil
• libmono-system-web2.0-cil
• libmono-microsoft8.0-cil
• libmono-system-runtime2.0-cil

Adam recommends a bit larger set including:

• subversion
• nant
• mono-gmcs
• libgdiplus
• libmono-i18n2.0-cil

I ended up with many more once it was all said done, but never got the libmono-i18n2.0-cil which may be one of those “unstable sources”.

Once everything is installed which may take a little bit, check some versions with the following commands:

svn --version (mine was 1.4.2 (r22196))

nant --version (mine was 0.85)

mono --version  (mine was 1.9.1)

WARNING: The version of mono should be 1.9.x, I could not build OpenSim 0.5.8 or the latest turn using mono 1.2.2.1

Step 6 - Download and Build the OpenSim Source

To download 0.5.8 (latest stable):

cd /var
svn co http://opensimulator.org/svn/opensim/tags/0.5.8-release opensim
cd opensim
./runbuild.sh
nant

To download the latest and greatest trunk version substitute:

http://opensimulator.org/svn/opensim/trunk opensim

With lots of luck, you will get a clean build.  If you have issues many common build problems are documented in Opensimulator Troubleshooting (your mileage may vary).

Step 7 - Edit the OpenSim.ini file Spin It Up

First you need to create a new ini file and edit it to use the OpenDynamicsEngine or else you will be able to just walk through prims.

cd bin
cp OpenSim.ini.example OpenSim.ini

jed OpenSim.ini

Comment out the line: physics = basicphysics and uncomment out the line physics = OpenDynamicsEngine. Ctrl-x Ctrl-s saves, Ctrl-x Ctrl-c exits.  Now spin it up.

mono OpenSim.exe

Then answer the questions thusly:

• Region Name [Opensim Test]:
  • The name of the your new Sim, you can safely change this but keep it short and one or two words, I no evidence to suggest it would not work with a long sentence, but based on SL sim naming conventions I’m suggesting you adhere to this if you are not a developer.
• Grid Location (X Axis) [1000]:
• Grid Location (Y Axis) [1000]:
  • Starting location in the world for your avatar.  Keep defaults.
• Internal IP Address for incoming UDP client connections [0.0.0.0]:
  • Should remain 0.0.0.0
• Internal IP Port for incoming UDP client connections [9000]:
  • Modify only if you are planning to use a different port or expose your sim to the internet
• External Host Name [127.0.0.1]:
  • WAKE UP HERE!  Put in the external IP address to your server 
• First Name of Master Avatar [Test]:
  • First name of your main avatar
• Last Name of Master Avatar [User]:
  • Last name of your main avatar.  Yes, you can pick this in Opensim :-).
• (Sandbox Mode Only)Password for Master Avatar Account [test]:
  • Password for your avatar

At this point the Opensim server should be up and running

Step 8 - Login From Somewherelse and Test with your SL Client

You can’t just login with your Second Life client because it has no way of knowing about your Opensim server.  You need to launch the SL client with some command line parameters to tell it to use Opensim vs. the SL grid.  Open a command window and navigate to the directory where you installed Second Life.  You should find SecondLife.exe  You can then type in:

SecondLife.exe -loginuri http://<YOUR IP HERE>:9000/

Start Building.  Or if you are so inclined learn how to create your own custom terrain here.

Step 9 - Keep OpenSim Running After You Log Out

The final step is getting OpenSim up and running even after you log out of your Putty/ssh session.  To do this first shutdown OpenSim by typing “shutdown”.  Then enter the following sequence:

screen
mono OpenSim.exe
ctrl-a ctrl-d

You should now be able to logout and your region will still be up and running. Try it!

Continuing to drive an interop standard, Linden Lab starts to build code to bridge into OpenSim.  Biggest open question is how will content movement happen and how will creator rights be protected or put at risk?

My second experiment in writing for the press. 

MIT Technology Review Article

Today is a historic day indeed the opening of a public bridge between the Second Life Preview Grid and the Opensim Grids who choose to participate.  Now for the uninitiated the bridge looks something like the following (image thanks to Crooked Brains):

It is narrow a bit wobbly, but it is really a masterful start and a further signal of Linden Lab’s strategy to open the architecture.

There are limits of course, no inventory is transferred so your fancied up SL avatar will appear in Ruth format with an empty inventory.  But if you only focus on the limitations, you might miss the big picture.  Opensim is have fantastic success, with 26 grids registered hosting approximately 2300 regions.  The SL grid has 28070 regions.  By these numbers Opensim grids combined are about 8% of the size of the entire SL grid, and I for one know there are many singleton Opensim regions waiting to go public.

SL’s Columbus has arrived in the new new virtual world and Opensim residents represent the native population.  So far, it is all turkeys and cranberry sauce.  I can’t wait for the next 12 months.

Here are some key links:

• Linden Lab Announcement
• UgoTrade Blog
• Original Linden Lab / IBM Annoucement
• Relevant Reuters Article

Ok, so you’ve spun up a new Opensim and logged in a couple times.  You are the proud owner of a big mound of grass in the middle of an infinite ocean.  Is that all there is?  Of course not.  In this tutorial, we will explore how to create a custom terrain using Bryce. 

Jane on Default Opensim Terrain

You can of course create a terrain with MSPaint or other more sophisticated 3D landscape generating packages like L3DT, but I have always had a love for Bryce’s fantastic landscapes long before the availability of shared 3D spaces.  Embedded in it is a nice little Terrain editor and I was curious whether I could walk upon a Bryce terrain.

A couple things.  1) This is not a Bryce tutorial, although I’ll try to take you through the steps in case your Bryce usage is rusty.  If you don’t know anything about Bryce, just skip ahead and grab the texture image to use as your new terrain.  2) This tutorial was built/tested with Opensim 0.5.8.

Step 1 - Launch Bryce and Create a Terrain

Along the top, look for the mini-terrain icon and click it.

Creating a Terrain in Bryce

You should get something like the following.

Click on the “E” next to the terrain to enter the terrain editor.

Step 2 - Edit the Terrain

I won’t go into details here on the Bryce Terrain editor but it is basically a mash-up of painting tools and pattern/noise generators. In terms of using it for Opensim, I’ve found it is a bit challanging for fine grained controls.  The terrain editor looks something like this:

Bryce Terrain Editor

Step 3 - Avoid Gottchas

A couple lessons I learned.

• A dramatic landscape in Bryce becomes a REALLY dramatic landscape in Opensim.  Less is more, use Lower and Dampen.
• Avoid spikey, drastic changes in elevation, veritical cliffs don’t show all that well
• Experiment with New and the drop down under Fractal to get some unusual terrains
• Avoid a terrain that runs up to the edge

If you have a terrain that is not black at the edge, it will also not have zero elevation in Opensim.  This leads to some interesting visual artifacts.  Below is a terrain without black edges and one screen capture of how it looks in Opensim.

Bryce Terrain

The above terrain loaded into Opensim.  Notice the artifacts at the non-black edges.

Terrain Artifacts

To avoid this, use the Round Edges control.

Step 4 - Export Your Terrain

Once you get the terrain looking like you want, the next step is to export it to an image file.  At the top of the main editing window inside the terrain editor, there is a drop down.  Select Export Image.  You have the option of PNG and TIFF.  Choose PNG and save it either alongside your Opensim.exe file or in a path that is easy to type (e.g. c:\terrains)

Export Image from Bryce

Here is the terrain image I ended up with (click on this and download it to use if you like).  You’ll notice I tried to edit in an inlet for water, and a few flat spots to build on along with a “secret” narrow passage through the ridge line to another over look.  This was done using the Bryce drawing tools and where the whole Bryce terrain editor thing broke down for me.  It is likely the best workflow is to generate the basic terrain and do hand editing in the Opensim client.

Bryce Terrain Final

Step 5 - Load the Terrain in Opensim

The file you want to load needs to be .r32 (raw 32-bit), .png, or .bmp.  I did not have success with .jpg at least in the version of Opensim I was using.

Spin up Opensim (remember to run it as Administrator on Vista).  Once it spins up, issue the terrain load command.

terrain load <file path/filename>

Example:

terrain load c:\terrain\myterrain.png

Loading A Terrain Opensim CLI

You will see either a success message or some complaint.

Step 6 - Login to Opensim and see your results

Here I am flying around in the new terrain. 

Bryce Terrain Loaded in OpensimHere is the terrain from aerial view rendered in Bryce.Bryce Render - Aerial View

 And finally, an aerial view of the same terrain in Opensim.  Love the waves, love Windlight.

Aerial View in Opensim

Would you like a tutorial on how you tune the textures on your new terrain?  If you’re an estate owner in Second Life, this may not be required.

Well I thought I would try publishing my first lively space to the blog. There are at the time of this writing 71 space “templates” available. I picked “Superhero Underwater Headquarters”. No, I never was a big fan of aqua-man, but this seemed the most interesting space after a quick browse.

I splattered in some seating arrangements, a clock, TV, found some cool purple lanterns that give off butterflies which I thought was a nice touch for an underwater lair. Drop in a check it out here:

My main gripes so far are as follows:

• Lack of ability to create my own content, I found this one of the most fun aspects of SL, in Lively I feel more like a decorator
• No basic keyboard movements keys WSAD please!
• No ability to re-size objects at least that I could find, move and rotate only. Object orientation seems to also be automatically attempted. So placement and rotation and relative scale was tricky
• The photo gadgets which would allow you to place any Internet image onto a surface are “off-line” per support. This could be a legal or porn issue, not sure but I have no real way of customizing my space with my own content other than You Tube video streaming which lacks even basic controls.

My first press article… ever

http://www.technologyreview.com/article/21100/