:: dreams illustrated ::

CSS nerdery

2010-02-09T18:28:00.000+02:00

Remember to check out:

{LESS}.app @ http://incident57.com/less/
{LESS}

Ruby based CSS compiling. Advanced CSS rules and such.

Tools for JavaScript optimization

2010-02-06T11:59:00.001+02:00

This is an oldie, but I was cleaning my link collection and thus I better store these here:

http://code.google.com/closure/

L-Systems in Python

2010-02-06T00:04:00.004+02:00

I thought this would have been much harder. Here ya go, simple and probably a bit sucky L-System in Python. Expand in your favourite 3D package to create what you need based on output:

#
#  L-System Generator
#  Janne Kaasalainen
#   7.2. 2010
#

class LSystem:
  ruleDictionary = {}
 
  def __init__(self):
      self.ruleDictionary = {}

  def setBase( self, aBase ):
    self.systemBase = aBase

  def addRule( self, aReplace, aWith ):
    # Add the rule to tule array
    self.ruleDictionary[ aReplace ] = aWith

  def iterate( self, aBase, aIterations ):

    if aIterations > 0:
      replaced = ""
      for i in aBase:
        replaced = replaced + self.ruleDictionary.get(i,i)
      aBase = replaced
      return self.iterate( aBase, aIterations-1 )
    else:
      return aBase


### Example Usage ###
system = LSystem()
system.addRule( "A", "AB" )  # A is replaced with AB
system.addRule( "B", "A" )  # B is replaced with A

iterations = 4
axiom = "A"
print "Starts with base A"
print "L-system with " + str(iterations) + " iterations: " + system.iterate( axiom, iterations )

It should be obvious you need to go through the returning string and visualize it as appropriate. But that's for another day.

Update: More comprehensive examples are now on StudioE18 blog:
Simple L-system generator
Visualizing L-Systems

It's in the eyes? Or is it?

2010-02-05T19:47:00.002+02:00

On Magic

2010-01-08T11:01:00.003+02:00

“Make no little plans. They have no magic to stir men’s blood and probably themselves will not be realized. Make big plans; aim high in hope and work, remembering that a noble, logical diagram once recorded will never die, but long after we are gone will be a living thing, asserting itself with ever-growing insistency. Remember that our sons and grandsons are going to do things that would stagger us. Let your watchword be order and your beacon beauty. Think big.”

—DANIEL BURNHAM, CHICAGO ARCHITECT. (1846-1912)

Reminder : Try this someday

2009-12-21T21:55:00.000+02:00

Enrique Gallar tweeted a link about VirtualAwesome:

VirtualAwesome is a cross-platform applications framework for advanced computer graphics and new user interfaces. It is built on top of a powerful graphics core (OSG) and has a strong emphasis on multi-focus interaction. This means, next to regular mouse, keyboard and tablet input, it supports multiple people interacting with multiple pointers/fingers at the same time.

I'd better play with it some when I find time from my current projects. Sounds interesting.

Automatically download and convert MythTV recordings for Nokia N900

2009-10-18T20:25:00.005+03:00

The previous post dealt with converting video on a Mac to a format that Nokia N900 can handle. It turned out to be a brief about the software available for the task as the device itself seemed to be versatile enough to handle all the tested videos. To push things little further, I'll now try to:

Have MythTV publish an RSS feed of its new recordings
Have the recordings be in format that Nokia N900 can handle
Download the recordings automatically to the device via a Mac

Sounds simple? Let's see. But at first, I'll assume there is a working MythTV install running on a server of some sort. Mine is a Mythbuntu 9.04 system, but I trust there are no big differences between different Linux based servers.

MythExport: A script that converts recordings and provides and RSS feed
Automatic: A Mac RSS client that downloads enclosures to directories

Installing MythExport is easy, just make sure you have the needed repositories available and install it via apt-get. It does ask ask a few additional question upon install such as directory into which it puts the exports. Perhaps the trickies part is where the installation process asks for a password but does not tell what the password is for. It is for MySQL. Settings can be done via web browsers, but one should remember to start the daemon first:

sudo /etc/init.d/mythexport start

If you have a custom Wwwroot, you need to create a symbolic link to mythexport, but note that the exact path may depend on a system so check it before copy-pasting:

ln -s /usr/share/mythtv/mythexport mythexport

Now, directing your browser to http://server/mythexport should return a list of files. This is not what we want, but hey, this is Linux, what did you expect? This sucker probably needs libapache-mod-perl2 but it's not installed as a dependency. I installed that, but the next thing needed was to enable *.cgi support in mythexport directory. So, modifying /etc/apache2/sites-enabled/mysite I added:

AddHandler cgi-script .cgi .pl
directory tag + directory
Options +ExecCGI
close directory tag

After this the web interface started to work. Usability wise the page is a disaster, but at least it works. I recommend browsing the net to get some grasp on how to fill in the fields and then testing the system to see how it behaves. In any case, you want create a user job that can then be triggered via MythTV front-end. For me, this didn't work out straight out of the box, as there were errors in AtomicParsley (via /var/log/mythtv/mythexport.log). Of course, there is yet another dependency that seemingly needs to be installed; ffmpeg. So, sudo apt-get install ffmpeg off you go. Heck, why not install lame at the same time too. And all the other codecs that come to mind.

However, after all these there seemed to be an error with the ffmpeg command line option as well. It might be worthwhile to see the error log issue and copy paste it to command line to better understand what is going wrong. In this case, there was a command line switch missing (-ar 44100 to set audio sample rate). I ended up using this custom configuration (with file extension .mp4):

-y -acodec libmp3lame -ab 128kb -ac 2 -ar 44100 -vcodec mpeg4 -b 600kb -mbd 2 -flags +4mv+aic -trellis 2 -cmp 2 -subcmp 2 -s 400x240

This is quarter of a file size of N900, so those who are into hifi may want to adjust settings here and there.

By the way, using command line tool "top" should show you nicely what's going on in the system. Video encoding is processor intensive operation, so it's not done too quickly. In any case, none of this is really too helpful yet. So far we've configured a user job that needs to be launched manually. I could have just as well dragged the recording to a converter that would have uploaded it to a mobile device. This is a no-no, we need something to automatically launch the user job once a program has been recorded.

MythTV front-end can be used to run the user job automatically. Go to Utilities/Setup -> Setup -> TV Settings and find a screen where you can assign jobs to be ran after the recordings. It may be possible to do this via MythWeb as well, though I didn't yet come to check if the advanced options get saved as defaults.

After all this, we need to set up the RSS feed reading on a client machine. Any feed reader you wish to use should work OK, of course. I'm a fan of NetNewsWire, but for this I chose to try Automatic. I simply made it to check the RSS and download straight to N900 video directory. It should be noted that I'm not fully aware what happens if the N900 is not mounted and Automatic tries to download items to it, but that's a subject for another post and another day.

I should point out that MythWeb UI, while functional, is a horrible mess and should be re-designed. Also, wouldn't it be logical that MythWeb UI would provide the RSS links itself? If anybody is willing to work on usability matters, contact me and perhaps I could lend a hand as well.

Anyhow, enough about ranting and let's enjoy this system what we can.

Links:

MythExport

Coding Curious Automatic

Addendum: One could always adapt this to iPhone/iPod touch by using iTunes as an RSS client. Or, with N900, use a client on it to directly subscribe to the RSS feed and thus negating the need of having a computer in between the device and the MythTV server.

Converting Video for Nokia N900

2009-10-18T17:29:00.000+03:00

This post is about converting video for Nokia N900 on Mac and I hope that it can be used as a partial solution for more automated home media networking alongside MythtV. While this stuff ought not to be hard, it may easily take a moment to dig up the right software to do the work and find the right settings for the conversion to play on the device. Thus, I'll use this as a quick checklist for later conversion needs. As a source I'll be using an old aikido clip that was made for a university course on multimedia production and a CG animated test clip.

Multimedia Support:

Video: 3GPP, AVI, Flash Video, H.264/AVC, MPEG-4, WMV
Audio: AAC, AAC+, eAAC+, M4A, MP3, WAV, WMA

This information is available via Forum Nokia. The screen size is 800x480 pixels and at that resolution the video playback is said to go up to 25 fps. Now, let's take a look at an example video that was proved to run on the device:

Resolution: 800x480, MPEG4 (.mp4)
h.264, 25 fps
AAC-Stereo, 44,100 Hz

Now, I had to go through several pieces of software to find a simple way to do conversions. The task could be accomplished with Quicktime Pro or actual video editing suites, but I deemed those overkill for the task at hands. However, it was rather surprising that other issues popped up with most common applications:

Evom: Successor to iSquint. While this application seemingly worked, I didn't find an easy place where to crop and resize the videos.
Video Monkey: Successor to VisualHub. Same issues as with Evom, though it does seem that the missing features will be addressed later on. There are already interface elements that would seemingly lead to required features.
MPEG Streamclip: Streamclip has saved me a few times in the past, and yet again it was found to be a working solution. It does appear complex, however, but revealed options can help to fine tune the encodings further.
Handbrake: Originally meant for DVD back-upping, Handbrake has recently added features to convert movie clips from computer as well. It

All of the utilities mentioned above produced working clips. None of the apps seemed fully optimal, though. It might be a nice idea to be really able to set up device profiles to which videos could simply be assigned to. In any case, both Handbrake and MPEG Streamclip do support presets. Video Monkey something to keep eyes on, if it gains easy to create profiles for other devices it could well become one of my favourites.

Personally, I'm somewhat divided between MPEG Streamclip and Handbrake. Handbrake GUI is much more polished, but it does seem its preset system, cropping and scaling are still more mysterious than those on MPEG Streamclip. Handbrake also has a command line client, that could be used for more complicated or automated tasks, which is a definite plus as it allows, for example, watch folders.

Links:
Handbrake
Forum Nokia
Video Monkey
Evom
MPEG Streamclip

Playing with Arduino

2009-09-28T10:31:00.000+03:00

I've been interested in electronics for quite a while. In fact, I chose my university back in '98 based on the thought that it would be fun to be able to make little things for oneself. As it turned out it wasn't really my thing to work with the circuit designs and design field got my attention. However, with little kits like Arduino one can play with very little knowledge about deep math involved and easily create simple devices.

I ran into Arduino the first time in Ars Electronica 2006 and played some with Basic Stamp and MAX/MSP while staying at Media Lab. With little knowledge one can get these things running fast. The first experiments with Arduino took only a few hours after which I had my first working prototype.

Here is a little board of mine, something really, really simple. It just turns on a led to a given brightness based on sensor input (ambient light and proximity). While this is both mechanically and software wise rather trivial, it gives quite a few possibilities to bring information technology past the computer screens.

Some notes and links

2009-09-18T08:29:00.001+03:00

I've been playing with a few personal projects of mine (Arduino and some mobile media stuff for home) and thus ran into following links for those designing and implementing projects in mobile space:

jQTouch - A jQuery plugin for mobile web development on the iPhone, Android, Palm Pre, and other forward-thinking devices.
PhoneGap - PhoneGap is an open source development tool for building fast, easy mobile apps with JavaScript.

Then some more visual links to serve as notes:

Tungsten - That rarest of species, Tungsten is a compact and sporty sans serif that’s disarming instead of pushy — not just loud, but persuasive.
Logo Jr Black
Heroine

Google Reader and Calendar Helveticafied

2009-09-06T13:10:00.002+03:00

As I am a big fan of the document Helvetica and it recently came to my attention that there has been Greasemonkey projects to make Google Reader and Google Calendar look better. While the looks of these seem very fascinating, I'm afraid I'm going to pass as I don't really use Google Reader that often (NetNewsWire syncs via it, though, which I do use) and Greasemonkey is not really what I'm too keen to use myself. Anyway, check these out:

Helvetical
Helvetireader

Speaking of which, Objectified is soon the be released on DVD and Blueray.

Holiday

2009-06-29T11:01:00.003+03:00

I've been in need for a vacation for a while now. My initial idea was to go to sailing but I was forced to change that plan due to timing issues. Thus the next logical alternative was to go to China by train.

Trans-Siberian railways start from Moscow and go all the way to Beijing and Vladivostok. We came to extend this trip a little bit from both ends. Today we head out from Helsinki to Moscow and once arriving o Beijing we continue onwards to visit Terracotta soldiers and go down to Hong Kong before flying back. Via Istanbul.

Thus, no news for a while, professional or not. I try to post updates every now and then to Nokia Image Exchange and Nokia Friend View.

Studio Lighting in Maya

2009-06-10T11:00:00.003+03:00

This is rather interesting, even if simple, video tutorial to try out. I guess it should also work for other 3d packages than Maya.

http://greensoda.squarespace.com/

So what is your duty?

2009-05-13T14:29:00.002+03:00

I've read quite a few times arguments that some problems that people face with computers do not belong to a specific party in question and ought to be solved by someone else. While this is technically true in some cases, it often does not really help the end user at all with the problem that they have at their hands. Thus it was refreshing to see these bullets from Google:

If users can't spell, it's our problem.

If they don't know the syntax of search, it's our problem.

If there is not enough content, it's our problem.

If they can't speak the language, it's our problem.

If the web is too slow, it's our problem.

Now, obviously you could always argue that people should learn to spell, but bullets like above show what you can at least want to try yourself while waiting the world to become a better place. This is the spirit to make things that are delightful to use.

Link:
Google Searchology 2009

In Defense of Eye Candy

2009-04-24T15:42:00.001+03:00

Nice to see something with examples and that goes deeper than the study that Norman has famously quoted.

We’ve all seen arguments in the design community that dismiss the role of beauty in visual interfaces, insisting that good designers base their choices strictly on matters of branding or basic design principles. Lost in these discussions is an understanding of the powerful role aesthetics play in shaping how we come to know, feel, and respond.

Consider how designers “skin” an information architect’s wireframes. Or how the term “eye candy” suggests that visual design is inessential. Our language constrains visual design to mere styling and separates aesthetics and usability, as if they are distinct considerations. Yet, if we shift the conversation away from graphical elements and instead focus on aesthetics, or “the science of how things are known via the senses,” we learn that this distinction between how something looks and how it works is somewhat artificial.

Link:
In Defense of Eye Candy

The Secret History of Silicon Valley

2009-04-24T09:28:00.001+03:00

A friend pointed me to this YouTube video:

A rather interesting speech on history.

What Did I Think Today

2009-04-23T09:15:00.003+03:00

This is blatantly obvious, but it just dawned to me; Apple's App Store is, in essence, same thing a linux distribution repository. Apt-get/yum is just coated under pleasing interface and transaction support is added on top.

Few warnings about DIY Time Machine back-ups

2009-04-22T10:09:00.004+03:00

Since I gave some pointers how to make a linux machine to serve as a Time Machine server, I think it is also fair to write a warning about possible corruption issues. Article Why Time Capsule is doomed to suck may or may not be relevant for home-brew systems.

Proceed with caution, though, and before trusting your system it is good to test it throughout.

Woes with MythTV

2009-03-28T17:33:00.015+02:00

It was all going (relatively) well. As can be seen from the previous posts, Time Machine was running and AFP installed nicely. I got MythTV to record TV as well. Now, due to all this I was able to trust the system enough to start customizing directory structures to my liking (for easy migration etc) and modify other elements to make my life easier.

And then I find out that UPnP in MythTV does not work. Now, this was working on my previous install, so I took it almost granted - the last time I did this I had to make sure myth-frontend was installed before UPnP was installed (which is bad on itself already). I also knew that UPnP might be slow to detect as I saw with Playstation 3 tests. But no, now I don't get it to work at all.

So what is there to do? Perhaps my client (MediaCloud for Mac) is not detecting this? No, that's not it as Windows Media Player 11 doesn't work either. Just to be really sure, I installed MediaTomb and it was detected by MediaCloud within seconds. This sort of rules out firewall issues as well, what I gather.

So, I try to install weekly builds from Mythbuntu package, and suddenly the upgrade does not go well but ends up in dependency hell. Of course, the UPnP now works. So, some more forum browsing and I find this:

sudo cp -R /home/mythtv /home/mythtv-backup
sudo userdel -r mythtv
sudo apt-get install -f

And then my MythTV front-end on Mac starts to crash on start-up. Just lovely. While I do take all this as a hobby project, this has started to get to my nerves quite a bit. Most annoying things are usually related to GUI's, surprisingly. It is not too much of a pain to take a bunch of CVS code, compile and install it and make a few scripts when you can determine it is really working, but MythTV has been a painful exercise what comes to making it work post-install. UI does not really support configuring all of the features and forces myriad configurations over top level features. General information architecture could be made much more clear as well so that one would better know where to find a setting he's interested about.

Finally, it would not hurt to see about the everyday UI's either. But oh well, at least it comes free (as in beer) for us hobbyists. Okay, enough of ranting. I'm only upset since I care, by the way. What to do? Let's build MythTV front-end for Mac since the binary packages are out of date. Help is found here.

My suggestions to bring this to this century regarding usability would be to:

Make the back-end to be standalone

Integrate a http daemon that can be used for setup/maintaining

Don't use MySQL to store back-end settings.
Integrate MythWeb

Re-think the UI information hierarchy

Make it pretty

Anyhow, in the meantime the Myth-Fronted for Mac compiled and actually almost started to work. It does not show TV from the default capture card, but the other, identical one, works. Have to see about that some other day.

UpdateSo, I got the TV working by purging the MythTV database on the server and manually re-creating it. This, of course, destroyed my UPnP settings and those don't seem to be available anywhere to be set via, say, GUI. Last time I found these via MythWeb, but now there aren't any settings. Anyhow, here are a few directories that would need to be set:

VideoStartupDir

MusicLocation

GalleryDir

To set them (replace values that correspond to yours):

sudo mysql
use mythconverg;
INSERT INTO settings (value, data, hostname) VALUES ('VideoStartupDir', '/home/tempdir/Shares/Videos', 'Firefly');
INSERT INTO settings (value, data, hostname) VALUES ('MusicLocation', '/home/tempdir/Shares/Music', 'Firefly');
INSERT INTO settings (value, data, hostname) VALUES ('GalleryDir', '/home/tempdir/Shares/Pictures', 'Firefly');

A sample how to update a row if it exists already:

UPDATE settings SET data = '/myth/music' WHERE value = 'MusicLocation' and hostname = 'Firefly';

Home Lab

2009-03-25T10:06:00.007+02:00

I got an Antec NS2480 case yesterday to house the server that is splattered across the table in the picture above. It ought to be enough to host 4 3.5" drives for storage. At the same time I came to test how cool these new/old Intel dual cores (Core 2 Duo 7300) run an noticed that the hearsing wasn't even warm even after the computer had been on for a few days. So I don't expect major issues other than possible noise that this server will be generating in a closet.

The whole thing is actually working rather well. One can tunnel X11 (X.org) over SSH which works fine on a Mac. Power buttons on the server shut it down nicely thanks to Mythbuntu. Next item on the line is to limit the size of Time Machine bundles so they don't take over the whole disk. Additionally, I still need to get a stack of hard drives and get a RAID 5 running on this to give more storage and some security what comes to disk failures. That needs to wait a bit, though.

After that I might give a stab at theming MythWeb (I've already tried a bit and it's a pain!) and figuring out something else cool to use this for.

Setting Netatalk for Time Machine

2009-03-21T11:23:00.011+02:00

This geek life is killing me. Anyhow, I previously wrote about my initial experiences on getting MythTV to provide home media. The second major role for this little server of mine is to work as a back-up and file sharing solution inside the house.

As I'm mostly a Apple user (and these experiences reinforce that decision, let me tell you!) this backing-up would be done with Time Machine if at all possible. Of course, Apple has made some changes to their protocols so the repository packages don't work and are practically obsolete. Funnily enough, even 9.04 Ubuntu releases are obsolete in this regard, the little I understand of this matter.

So, that leaves me with 2.1dev branch that's on the CVS. First task is, of course, to get that branch and have it compile. So, here we go, this should take care of most of the dependencies:

apt-get install cvs autoconf cracklib2-dev libssl-dev
apt-get build-dep netatalk

The, let's get the branch itself:

cvs -d:pserver:anonymous@netatalk.cvs.sourceforge.net:/cvsroot/netatalk login
cvs -z3 -d:pserver:anonymous@netatalk.cvs.sourceforge.net:/cvsroot/netatalk co -P netatalk

And compile...

cd netatalk
./autoconf
./configure
make
sudo make install

And lastly, just in case that no package management system starts to think on their own:

echo "netatalk hold" | sudo dpkg --set-selections

A watchful eye might notice that this doesn't really help much. Netatalk ain't configured yet and you'd really wish to have init.d scripts for it as well.

Addenum

Netatalk installed various components to /usr/local/sbin, but thankfully I only care about afpd. Configuration files are installed by default to /usr/local/etc/netatalk, and thhere you can set shared volumes with AppleVolumes.default file. Better tutorials than mine exist for exact parameters.

For my own house keeping, here is a magic line to put to the volumes file (Leopard machines only, possibly newer OS X's too):

/home/username/TimeMachine TimeMachine allow:username1,username2 cnidscheme:cdb options:usedots,upriv

Here's a line for user authentication:

- -transall -uamlist uams_randnum.so,uams_dhx.so -nosavepassword -advertise_ssh

However, with this custom installation I did need to create a custom init.d file (I called mine simply netatalk, even if it just calls afpd). It turned out to be better done by downloading the netatalk deb-file and extracting it first via:

$dpkg-deb -x {deb-package name} {target directory}

Simply correct the paths inside the script to correct ones and use it from thereafter.

There was also a need to have /etc/avahi/services/afpd.service file. Even if this is somewhat easy to find, I put it here too to make it easier for myself to find it later:

<?xml version="1.0" standalone='no'?><!--*-nxml-*-->
<!DOCTYPE service-group SYSTEM "avahi-service.dtd">
<service-group>
<name replace-wildcards="yes">%h</name>
<service>
<type>_afpovertcp._tcp</type>
<port>548</port>
</service>
<service>
<type>_device-info._tcp</type>
<port>0</port>
<txt-record>model=Xserve</txt-record>
</service>
</service-group>

A little edit was also needed to /etc/nsswitch.conf file. Make sure you have a line like the following:

hosts: files mdns4_minimal [NOTFOUND=return] dns mdns4 mdns

After all this and some extras not mentioned here, system seems to be running. I've noted that there are already errors with the init.d script, for example, but I'll try to keep this post up to date with what I am running currently. Further, if you have comments, clarifications, corrections or general help, that's highly appreciated.

Experiences with MythTV

2009-03-20T18:30:00.011+02:00

I do try to avoid more geeky subjects, but every now and then one has to go down and dirty with technology to have something that really fits him. In this case it is my need of having a distributed media system at my place to deliver TV recordings, bought music and various podcasts around my house.

I'm not exactly a Linux newbie myself, even if I'm a far cry from a professional admin. I've run a few boxes, though, one for personal website and another as a render slave for the sake of experimentation. These started with Debian a long time back, but as Ubuntu had newer and more packages I switched to those a few years back. I did have a MythTV setup running on them already and given the experiences I had I chose to run with Mythbuntu this time.

At first I tried what's the state of the alpha release of Mythbuntu 9.04 and as it seemed to have issues to have MythTV interfaces show up correctly I short-cut the installation a little and reverted to more tested 8.10 release. After a while debugging the alpha, it was clear that I better use my time for something else and update later on when 9.04 is ready.

And the setup was complex indeed for what a consumer is used to, and this time it was even more pronounced as I was to do all the server configuration at once. So, here's a brief process description:

Intall mercurial
Get mantis driver code and REMEMBER to use take s2-liplianin branch and perhaps a needed patch
Merge the patch
Compile the drivers and install them
At this point I didn't want to remind myself how to load kernel modules, so just boot the machine and check lsmod that the DVB-C cards are detected

Now, after having compiled drivers you'd think things get easier? No, they won't. MythTV channel setup is nightmare what comes to usability. For me, it just started to work at some point. So at first, be patient. Likewise, the database filling seems to hang or do some strange things. But it sort of runs now, after having manually put channel numbers for the channels.

I still can't watch TV from the front end, though, but I'll tackle that later. Right now I don't want to dive into logs and see why I can't. Now, onto trying to make sense out of Netatalk 2.1dev branch.

Update: Turns out that there was a problem with channel tuning. These problems were plenty, but eventually I was able to create to create a channels.conf file with dvb-utils. However, I'm still not able to watch TV on local machine. It seems to record 4 programs simultaneously just fine, though.

Links:
Mythbuntu
s2-liplianin

Safari 4 UI Issues

2009-03-02T07:45:00.017+02:00

I've been running Safari 4 Beta for a while now, since it was released that is, to avoid making too hasty opinions about the changes Apple has made regarding the UI. Like many, I'm not totally happy with it, but let's start with positive changes at first.

A notable, positive change is the speed of the browser. This thing does feel noticeably faster and brings again the same fuzzy warm feeling I had when I tested out Chrome the first time. The speed alone, not to mention the added updates to WebKit are alone a reason enough for an update.

I personally also find "Top Sites" feature rather practical, but this might depend on one's usage patterns. Top Sites shows a page with a grid of web sites that one visits "often enough" to be considered important. What the exact algorithm is for this, I don't know and it's not very clear how to modify this to your exact liking, but the feature does give visual representation to your top sites and thus makes it easy to visit them. For me, this has pretty much replaced the whole bookmarks bar, saving space from the UI some more. Additional nice touch is that the feature does not get into your way if you do not wish to use it. With some polishing, this could be made to work really nice.

Another very nice touch is using Coverflow for browsing history. At first this might feel like a gimmick, but as many websites do not identify their sub pages with easy to remember titles this becomes really beneficial for those random searches through ones history. It is easier to remember visually how pages you look for looked like than how they were called in their titles. We actually pioneered this stuff in S60 MiniMap browser some time back already (*waves to Antti*), but personally I'm rather happy that one of the more-mainstream browsers get this functionality as well, even if in somewhat adapted form. Not that we were likely the first with our browser either.

Re-defined address bar is also nice, and I find it benefitting me a lot. A reminiscent of Awesome bar in Firefox, it really makes typing the same URL's faster. However, they could also add the Google search to the address bar as well. While the search field is also improved to include latest searches as well as Google suggestions, and is very handy nice little improvement too, Chrome has shown how this can be done pleasantly within the address field already. Such would save even more space from interface as the search field would become redundant.

As I'm moving towards the negative, I ought to highlight at this point that some of my issues are personal and due to my setup whereas other issues are more general. As a full disclaimer, I'm using a Wacom tablet at work (but a mouse at home) and it is increasingly fighting with OS X Ui choices. I am fully aware this is not what most people use, but more on this later.

Where using Wacom affects most is the tabs at the top of the browser window. The problem is that these new tabs have the behavior of a window title bar as well as that of a browser tab. They can be used to drag window around, to close a tab or move the tabs. Now, one of the behaviors of a window's title bar is that it can be double-clicked to minimize the window to dock. Since at least I tend to switch a lot between tabs, my Wacom tends to minimize the window to the level it is highly annoying. I've not noted the issue with normal windows, which could indicate that the algorithm for this particular behavior is not standard or that my usage patterns do not deal with clicking the title bars too often other than to move the windows around. This brings us to the second issue, which is the ever increasing utilization of click-though.

A lot has been written about OS X and how it mixes the model of click-though and click-to-bring front. I do not go into details here as others have explained the matter better a long time ago (John Gruber for one). However, I do consider the click-through lacks worse than the benefits gained from it. The biggest advantage of click-to-bring-front is that you do not need to guess and think what will happen when you click a window that is not active. This is not dumbing down the interface or making it for dummy users, but it makes it easier for the user to stay in "flow". He is doing his tasks and should be let to concentrate on the matter at hand instead of break his thought process to second-guess what the operating system will do depending where you press your mouse button. This is even further highlighted with tablets, as pens are very fast way to just click somewhere in a non-active window.

Safari 4's tabs that are on top take this pain to maximum. Not only do they behave as a mix of tabs and window titles, they also get crowded once you open more of them. Tabs have a way to close them, way to drag them, an indicator about loading content as well as the identifying method, the title for a given tab. As number of tabs increases, managing all this becomes excruciating exercise. We could argue that opening more tabs is not a common behavior for users, but then, click through utilization should be considered as a similar power user feature as well. I should add that imho this is a flaw in implementation, not in the idea. Managing information hierarchy is an windows in this advanced way is not a bad thing at all, but Safari 4 is not making it right. For instance, I do not have issues with how Chrome does what is basically the same thing. I do not have a concrete suggestions how to fix this either, but then, I'm not working for Apple anyhow.

Speaking of information hierarchy and tabs-on-top, another issue requiring polishing is the location of loading indicator. Namely, the location varies and depends on not only if the tab is on front but also what the user has been doing. On the selected tab loading indicator is at the end of the address bar. On background tab it is on the tab leaves. This causes eyes to jump around the interface as one juggles through different tabs. Having a fixed location for a loading indicator would be far better, or using multiple indicators of which one would be fixed. In fact, if the loading indicator would always be attached to the tab and the old method of using address field to show the progress of the selected tab this problem would already be solved.

So, for me the better way to go through this was to customize the UI which is thankfully possible. Here's a bit of command line magic that brings back a more usable interface (if it's not evident how to use these, don't):


 defaults write com.apple.Safari DebugSafari4TabBarIsOnTop -bool NO
 defaults write com.apple.Safari DebugSafari4IncludeToolbarRedesign -bool NO
 defaults write com.apple.Safari DebugSafari4LoadProgressStyle -bool NO

These changes can be reversed rather easily too:


 defaults delete com.apple.safari pref_name

I hope that helps somebody. In general, I am an Apple aficionado but one reason to use these machines is to be a productive designer/artist. If things start to turn to the point that I need to customize the systems too much I might just as well go back to Windows or Linux given they'd run the software I need. Same goes if the UI loses the factors that have made it different. There seems to have been increasing lack in details during the last few OS X updates that have quite frankly puzzled me.

Royal Pains

2009-02-23T22:44:00.003+02:00

Ever wondered what must be the hardest thing on CG graphics? Hair just have to be there somewhere in top 10...

I guess I'll be spending rest of the week to have anything worth to be proud of.

Mudbox Concept in 20 mins

2009-02-22T10:54:00.003+02:00

By Wayne Robson

Texturing

2009-02-19T10:55:00.004+02:00

So, I finally had time to try these texturing tests with my "real" model. Texturing is not something I especially enjoy. Fighting with the UV layouts is quite fine, but actually starting to paint the models is definitely not my cup of tea. Yet at least. I guess more practice is needed and possibly a little better tools than what I have at my disposal.

One more

2009-01-22T13:05:00.001+02:00

These are still OpenGL rendered previews.

3D modeling

2009-01-21T13:14:00.009+02:00

I have this very casual hobby of doing 3D on my free time. During the universities I was rather keen to it, even, and wanted to do my final thesis as a short animation. That proved to be too laborious.

However, I've not fully given up on that. Here's two texturing samples:

Nothing too fancy, just playing around to see how to texture more accurately based on photographs.

Decadence and New Start

2009-01-06T10:01:00.006+02:00

There is something strange in New Years. It is as if the time is starting a fresh once more and all that was before gets erased. It is like a new chance, even if the day is arbitrary. But each year it gives the possibility to reflect back to see what was done and how one has lived his life.

So, what did happen? I graduated from University of Art and Design Helsinki, for one. We launched Nokia Friend View and later on Nokia Image Exchange, marking an important milestone regarding how we've spent the last few years at work. Year ended in decadent moments when the road ahead for our team became unclear and despite the difficulties we had our celebrations. Thusly, it was indeed the last flight of the phoenix.

And in the midst of all this, it felt like a comforting idea to start this blog and once again concentrate on my personal thoughts and interests now that I have free time on my hands once more. Let's see what comes out of this, the year has the very least an intriguing beginning.