Building a Better Scorpion

Following the Gakken mechanical grasshopper kit, we have the scorpion (1,100 yen = $10 USD). Not including the springs, nuts and bolts, the kit consists of 30+ pieces, and a small sheet of sandpaper to smooth down any burrs if you are so inclined. Again, I didn’t see a suggested assembly time, but it took me a little over an hour to finish it, largely because the little nuts and bolts kept loosening up or falling to the floor.

There’s nothing fundamentally difficult about building this one. The sheet metal is very soft and malleable, and the instructions are very clear even if you don’t understand the language. You won’t see it marketed in the U.S. to anyone under 12 because of the sharp corners and small part choking hazards, but those aren’t as much of a safety concern in Japan. The two included tools are basically useless – it’s better to have your own screwdriver and hex driver set. The only thing worth mentioning is the forming of the face-spine strip. The long thin piece has kind of a half-moon section at one end. This half-moon needs to be curved before mounting it on the back body pieces, so that the bolts holding the half moon to the lower face plate don’t damage anything. If you use the included tool as a pliers to bend the half moon, you’re going to scratch the metal in places that are going to be visible when the kit is done. What I should have done is to use the tool to bend the outer tabs, and then keep pressing the center tab against the table or floor until it was curved enough. As it is, I got sloppy, so now the “forehead” section is scratched. It’s not overly obvious in the photos, but I could have done a better job on that step.

The springs don’t serve a functional purpose, they’re just there for appearances.

It’s fun making these kits, but they’re kind of big and I don’t have many places to display them. I’m thinking of getting one a month, but there are only two right now that I want to assemble. The next one will probably be the tarantula.

The pamphlet has the instructions for the kit, advertising for books on insects and dinosaurs, and a list of the other kits in the series. Plus, there are three pages of photos of five species of scorpions, and a small explanation of why scorpions are not insects.


Otona no Kagaku newsletter #163

Just got the latest newsletter in email. Unfortunately it’s very skimpy on the kind of news that I’m hoping for; specifically announcements of any plans to release the next Adult Science kit.

1) Rainbow Loom
The first item is for the Rainbow Loom kit for girls. There’s a new starter kit book coming out April 21st, and a few events on the 12th and 25th.

2) Denshi Block EX app
Gakken is releasing an app for the iPad for emulating their electronics building block EX-150 kit. Due out on Mar. 30, for 700 yen ($6 USD). Looks to be Japanese only.

M.C. Escher Kiri-e

(Finished kiri-e)

I really like M.C. Escher, and I think that a number of his pieces lend themselves to the kiri-e approach (cut-paper pictures). So, I was looking for other projects to try, and his self-portrait caught my eye. The challenge was to translate the pencil/woodblock shading effects into what’s essentially a simple stencil design. The result is that Escher has a lot more solid black in his hair and beard than was in the original drawing. This makes him look almost 20 years younger in my version. But, I think that the overall finished picture looks close enough to the original for government work.

In the original, the white portion of the shirt collar was the same as the white of his skin, his eyes were black, and the detailed edges of his black jacket were the same color as for the rest of the jacket. I debated over how to handle this and how much color I should introduce. I have a sheet of pink construction paper that would have been perfect for the skin tone, but it took me too far from what Escher had been trying to do. So, I just used brown for the pupils, an off-white for the shirt collar just to differentiate it a bit from the skin of his neck, and then for the jacket I took the black paper I used for the cut-out and turned it over so that the unpatterned side would contrast with the patterned face of the cut-out. The jacket effects don’t stand out as much in the scan as they do in the real thing, so I’ve included the cut-out below to show what that looked like before I glued the other stuff in.

(Raw cut-out)

Bokaro P ni Naritai, vol. 13

(Images used for review purposes only.)

I want to be a Vocaloid Producer, vol. 6, 1,500 yen, plus tax.

New magazine features:
In the 4-panel comic, Rana is stuffing herself with the candies she got for White Day (Mar. 14) when Robo-panda yells at her to get to class. Teachers Jasmine and Satchan then eat the remaining candy. The classroom section covers how to upload a finished music video to youtube. Next, there’s a discussion of New Pops music and an interview with Band Sound genre composer Last Note. There’s a bit more talk about notes, scales and chords, then the last page highlights the Cyber Diva package, and the pickup artist is Nana again.

New DVD Features:
The pickup song is a bouncy little piece by Nana, called Electrotrain. Very buzzy, and it sounds like Rana has been heavily autotuned.

The MMD accessory file this time is another dance stage – Class Room.

It looks like the publisher has decided to stop including voice sample files. Instead, we get the finished video of Rana doing the sidestep dance in the Chocolate Room from volume 12.

(MMD tutorial detail for Rana doing a dance spin turn.)

The primary goal this time is to get Rana to whisper. This is pretty easy, in that you right click on each phoneme in the music editor, choose Note Property, and then append “_0” to the end of the pronunciation part within the Phonetic field. It’s something of a pain, because you have to do each and every phoneme one at a time, though. Afterward, the volume will be too low, so you drag the portion of the track you want to edit into a separate blank track, bring up the Mixer and then change the Compressor settings for that track. Unfortunately, this does add noise, and you need to remove that as part of the SSW tutorial. The next part is to add a bit of hissing to the beginning of a word that starts with “Shu”. To do this, go to the other blank track, add a new phoneme in the same place as “Shu”, set the new phoneme to “Shu” and open Note Property to change Phonetic to “Shu_0”. This causes “shu” to be sung twice, but the second one is in a whisper, making the overall effect more sibilant.

And, I finally figured out why I can’t work with anything shorter than quarter notes in the editors – for some reason my version of the demo song files are set to “Quantize” 1/4 notes. I changed this to 1/8 and everything works better now.

I was hoping for the use of a noise removal filter. Instead, the tutorial discusses the use of Automation to zero out the volume at points in the whisper track where the pops occur. Automation is a misnomer, because nothing’s actually being automated here. Instead, you have a parameter (i.e. – volume) level than can be segmented and dragged so that a particular track will be affected more or less by that parameter at specific points. It’s automatic in that you can create linear ramps between points and SSW will figure out how to take the steps between those points, but that’s about it. In any event, you can use the Volume parameter track under each instrument, or click on the Controller Entry button in the toolbar, select Volume from the pulldown and then manually enter start and end points for the ramp and if you want the ramp to go from 0 to 100, or whatever other values you need. And this works for both analog samples and MIDI tracks. So, what you do is look for “bursts” in the whisper tracks, which represent a noise pop, and set the volume to 0 only for that burst.

In the last volume, there was a discussion of the Chip Tune genre, which uses the sounds from 80’s video games. To get something of that “chip” sound, you can turn the Equalizer (EG Bandpass) effect on and off. This can be done from the mixer for the entire song by assigning EQ to Output 1 and then clicking the On/Off button. OR, you can go to the automation track for Output 1, change that to EQ Bandpass, and set the levels as you like throughout the song. In the tutorial, you set EQ to 0 for the first 10 bars, and jump to 100 for the rest of the song. This gives the feel of listening to the beginning of the song on a cheap transistor radio and then switching to a good stereo amp.

Ok, honesty time again. I didn’t even try to do the exercises for MMD for this volume. In part I was just too busy over the last two weeks to get to it, and in part I’m at the point where I want to do something other than recreate dance video poses. The tutorial concentrates on creating a pirouette cycle for Rana. There’s a motion file for the foot movement, but we’re supposed to set the key frames for Rana’s arm and head rotations, and modify the motion curves to speed up and slow down appropriately at the keys. There’s nothing really new for MMD itself; instead, the majority of the lesson concentrates on dance video, and animation, theory. Eventually, all of the dance portions – the sidestep dance, the turn, and another movement cycle coming up in the next volume, will be stitched together to make a full video.

(Rana starting the pirouette, in the new classroom stage.)

As mentioned above, I’m starting to think of a project I want to do for myself, so rather than following through on the future tutorials, I’m going to try to make time for my project, and hope that a future volume will have the kind of accessory dance stage that I need.

Menus, part 2

What I really use the Firefox menus for is reading online comics. I’ve written before about my VBScripts for grabbing the image URLs for the GoComics strips that I like reading every day. I have a folder on my desktop that contains a batch file that I run after GoComics rolls over their comics for the next day at 10 minutes after midnight CST. The file runs my VBScript, which gets the links for each of the GIF or jpg files for those strips, and then builds up an HTML page. Once that’s done, I go to my Firefox root menu screen, click on the Comics link, and then I’d click on the link “GoComics shortcut”, which would open the page created by the VBScript.

This system was fine when I only had 20 or so strips that I liked, but now I’m up to 40, and it doesn’t work at all with the strips that are hosted on the Seattle newspaper site (such as Funky Winkerbean and Bizarro), since that site stopped allowing hot linking of their comics. In the case of GoComics, I just made a second script and a second generated HTML file, but this means that I have to manually open both generated HTML pages one at a time if I want to read all of the GoComics comics at one sitting. And, I have to open every single Seattle comic one at a time every day. It’s not a particularly crippling problem, but I’m a programmer, and I’ve had this itch at the back of my mind to address this situation at some point.

I finally got around to doing something about this. Initially, I was hoping there’d be something in HTML that would allow me to open multiple pages with one link, but it doesn’t look like this is part of the HTML standard, yet. So I started rooting around on the javascript tutorial sites. And, it’s surprisingly easy to open new pages in .js. So much so that I rewrote all of my pages to adopt a new format. The links are all image buttons, and either open a sub-menu page, or go directly to the outside page I want.

(New root menu.)

As I made my changes, I figured that I’d try my hand at cascading style sheets (CSS) as well, which is what gave me my above 2-column layout. The results have been much more mixed, though…

Just about every tutorial site I visited starts out the same way – “HTML, javscript and CSS are the three prongs that every web developer needs to be familiar with.” Modern developer, maybe. I started HTML programming in the 90’s, when the internet was just starting to catch on. There’s not that much to HTML, and since the obsoleted tags still work in backward-compatible browsers, I haven’t needed to “get with the times”. Javascript isn’t necessary in simple pages, and I’ve never liked CSS.

Actually, I’ve always viewed CSS as an unnecessary complication that doesn’t work all that well. Yes, I get the idea that having a master CSS file in one location that lets you update 200 .html pages in one fell swoop is a nice thing, but this works best if all the pages follow similar structures, or if you work within some development environment like Dreamweaver. For what I do, with only a few pages that are all different, CSS just gets in the way.

But… since I was changing everything anyway, I might as well at least try to see if CSS would let me do something more cool than I was planning on.

First off, having a “navigation” bar and a “main body block” is nice. I can more easily add a “back” button to the sub-menus. That part’s ok. So, I made my “mystyles.css” file, put it in the html directory, and linked to it from the other pages.

The next step was to turn the text links into image buttons. According to several of the tutorials, putting “img src” tags inside the button definition is a “bad thing” because of unresolved security issues. Because all the files, image and html, are on my PC, and no one else uses it, I’m not worried about this. Rather, I don’t like the work-around, which is to make separate calls to 3 different functions, for hover, mouse out and onclick, with different images for each button state. That would put several hundred little thumbnail images in the pics directory, and additional stress on my hard drive read head. Naw, I’ll keep things straightforward.

<button onclick=”openSingle(‘;)” type=”button”>
<img src=”pics/yahoo_com.gif” width=”100″>

I put the javascript code itself at the bottom of the page, just before the end of the body block:

function openSingle(fn) {
var myWindow =, “_self”);

I could have just wrapped an “a href” statement around the “img src” statement, and that would have been fine, except that I want to be able to open multiple links with one mouse click. Having the button calling my function works for this, and it gives me a “hover over” effect because the button border changes color. If I want to open multiple tabs, I can make a second function and stick all of my links in that.

function openMany() {
var myWindow =“;, “_self”);
myWindow =“;, “_target”);
myWindow =“;, “_target”);

This works really well for opening all of the webcomics pages I want to read all at one time.

What’s still missing, though, is some kind of reminder for what a particular image represents, should I forget along the way. Apparently, this was a problem for other newbie javascript programmers, because several people asked about this on the forums, and most of the replies were “I dunno”. I did track down one solution that was exactly what I wanted – span. Span pops up a tool tip-kind of text message when the mouse stays in one place over the hot spot for 1 second or so.

<span title=”English Yahoo”>
<button onclick=”openSingle(‘;)” type=”button”>
<img src=”pics/yahoo_com.gif” width=”100″>

This gives me my functionality. The next step is look and feel.

I recently read a paper from some computer guru that made a big deal about “separating presentation from content”. And I saw similar comments in the CSS forums. What this means is that the code for defining how a page looks (the presentation) should be separate from the code controlling what the page does (the content, or the domain). The reason is that you can then easily port your page between browsers, or platforms (Mac, PC, tablets, smartphones). And this is where having a dedicated .css file comes in. By specifying styles for each part of your HTML page, and just calling the style by its ID, you don’t need style information (presentation) as part of your content (presentation code then goes in the .css file).

It’s a noble conceit, but I couldn’t make it work. If you look at my above screen cap, I have two columns, each of which is one division (<div>). To specify where each of the buttons is going to go, I need to make additional div’s, one for each button. The only positional choices are “absolute” and “relative”. Using multiple embedded “absolute divs” kills “a href” links because the location of the link, and where the browser THINKS the link is, don’t line up anymore. And, using “relative” forces the buttons onto separate lines. In other words, I couldn’t get the buttons to work if I used <div> instead of tables. Now, I know that CSS programmers hate tables for anything other than displaying tabular data, and that tables are presentation code that’s mixed in with content. I don’t care. For what I want – image-based menu buttons in a nice layout on my screen – tables work fine and CSS doesn’t.

On the other hand, if someone can tell me how to do this better, I am willing to listen.

Big Hits

Generally, I only get 40 or so hits a day on this blog, with occasional spikes at 80 about once every one or two weeks. So, I was pretty surprised on March 6th to see that I’d gotten 242 hits with about 130 unique visitors the day before. Looking at the search key listings, I found that there were 35 searches on Momofuku Ando, the creator of cup ramen, which all linked to the 50 Famous People magazine write-up on him. Turns out that Google had a doodle celebrating the 105th anniversary of Ando’s birthday on the 5th. That still doesn’t explain why so many people found my blog…

But, things are back to normal now. I only had 30 hits yesterday.

Menus, part 1

Maybe I’m weird, but I don’t like having a lot of bookmarks in my internet browser, and I especially dislike having to click through a bunch of pages to reach websites that I visit all the time (as much as 3-4 times a day). I like having pages grouped logically, and with simple navigation. For me, this means VBScript, batchfiles and (most recently) javascript.

A long time ago, I hand-coded a set of HTML files that I placed on my PC for my most common tasks. For simplicity’s sake I just named them c1.html through c8.html. I made the main menu, c.html, the root screen, and set that as my home page in Firefox. This means, every time I start Firefox in the morning, c.html opens automatically. To keep things readable, I use tables, with each link in one cell. Within the menuing system, the root links open the next file within the same window. The code looks like this:

<title>3 Steps Over The Edge –Back Door</title>
<meta content=”text/html; charset=unicode” http-equiv=”Content-Type” />
<body bgcolor=”#00ffff”>
<div align=”center”>
<table border=”6″ cellspacing=”3″ cellpadding=”3″ width=”600″>
<tr align=”center”>
<td colspan=”3″><h3>Root Menu</h3>                                     </td>
<td width=”350″><a href=”c1.html”>Links 1</a>                             </td>
<td colspan=”2″><a href=”c3.html”>Links 2</a>                              </td>
<td width=”350″><a href=”c4.html”>Links 3</a>                        </td>
<td colspan=”2″><a href=”c8.html”>Links 4</a>                            </td>
<td colspan=”3″>     </td>
<td width=”350″ ><a href=”c2.html”>Links 5</a>        </td>
<td width=”150″ ><a href=””>Site 1</a> </td>
<td width=”150″ ><a href=””>Site 2</a> </td>
<td width=”300″><a href=””>Otona no Kagaku</a>       </td>
<td colspan=”2″> </td>
<td colspan=”3″>  </td>
<form method=”get” action=””&gt;
<input size=”30″ name=”p” /> <input value=”English Yahoo!” type=”submit” /> </p>
<form method=”get” name=”sf1″ action=””&gt;
<input size=”30″ name=”p” /> <input value=”Japanese Yahoo!” type=”submit” /> </p>

I also like having direct search forms for and Unfortunately, the form strips out kanji, so that’s not particularly useful. I’ll probably remove the direct search forms at some point soon.

The menu screens are pretty boring, and I could change them up to use graphics, but for my purposes this has been adequate, and it’s easy for me to maintain. I do recognize that this HTML doesn’t follow all of the more modern syntax rules, such as having the slash after the “br” tag, using “b” instead of “strong”, and not using a .css file, but these pages are for my personal use. I don’t care if the syntax is textbook or not.

At least, the above had been true until a couple weeks ago. I finally decided that I wanted to make the menuing system more visually interesting. The approach I settled on is to change the text links to image-based buttons, for something that’s closer in appearance to Microsoft’s Window’s image icons, combined with sub-folders.

(To be cont.)

Bokaro P ni Naritai, vol. 12

(Images used for review purposes only.)

I want to be a Vocaloid Producer, vol. 12, 1,500 yen, plus tax.
New magazine features:

The 4-panel comic has Rana handing out homemade chocolates. Unfortunately, she used Tabasco sauce with white chocolate to make the pink coloring. The classroom section has Rana panicking over how to start out writing her own music, so Jasmine and Robo-panda tell her to pick an “image” first, such as an up-tempo piece that makes people smile. Jasmine adds that there’s a walkthrough PDF tutorial in the Singer-Song Writer program install directory that she can use as well. There’s an overview of 80’s retro “chip tune” music and an interview with Sasakure.UK. The tutorial section goes into the synthesizer plug-in for SSW in some depth. Then finally there’s a mention of the Vocaloid 4 Editor, and a brief talk with the pick-up artist, Takamatt.

(SSW section covering the main features of the Alpha 3 plug-in.)

New DVD Features:
The only accessory for MMD this time is a new room model, for the Chocolate Room.

The pick-up demo song is “Zenga”, by Takamatt. It’s a relatively subdued J-pop piece that sounds like it’s been slowed down by 50%, and has kind of a Euro-Latin feel. I like this one better than most of the other demo songs so far.

The main task is to change Rana’s enunciation of specific syllables. Unfortunately, this highlights a flaw in my version of Vocaloid, which might be connected to it being the 32-bit package (the install directions said to use 64-bit, which my little laptop doesn’t support). The problem is that I can’t select small sections of the piano roll (16th notes(?)) to do what the tutorial says to do. When I try, I get an error message indicating overlapping notes. The task is to insert a new phoneme between two existing syllables and use Note Property to change Phonetic and Lyric to “Sil”. The last part of the video is just the playback of the work song, “Morning Discussion”. This has that retro “chip tune” video game sound going for it, along with a perky, bouncy lyric track.

Finally, something that grabs my attention from the get-go.
As mentioned above, the SSW section of the magazine talks about the synthesizer plug-in. Specifically, this is the Alpha 3 module from the German company LinPlug. The first portion of the tutorial walks through the steps for changing the regular base guitar track to the Alpha 3, then loading preset patch files to get the 80’s retro video game sound that represents the chip tune genre (i.e. – lots of squarewaves). After this, we change three more tracks to synth variants, and the demo song is done. The second half of the tutorial is a brief overview of the OSC, Filter and Amp sections of the main interface. The LFO section is skipped, and the student is directed to the Japanese manual in the SSW VSTPlugins folder. Personally, I wanted the English manual, so I went to the LinPlug website for that.

(The Alpha 3 interface. Much coolness.)

If you’re familiar with synthesizers (or if you read my blog entries for my Java app) then you’ll recognize most of the controls here. The interesting things are that Osc 1 and Osc 2 both have A and B waveforms, balance and detune controls. Then, there’s the mixer for Osc 1 and 2, so essentially you’ve got 4 waveforms to play with, not just 2. The Filter has 3 low band cutoff switches, band pass and high band cutoff, as well as resonance and an envelope generator. Amp has volume and velocity, and its own EG. The enhancements are in the form of depth (how big the filter EG envelope is, spread (size of the amplitude envelope) and fade times. There’s a good 30 Osc waveforms (not just the plain square, triangle and sawtooth waves) as well as a noise generator. I haven’t done much with the LFO, and what little I did try didn’t have much of an effect, but that’s only because the notes for the sample work song are all so short. I need to start with a different song, probably something written from scratch. Overall, the Alpha 3 is a fun little toy in the same category as the HackMe Rocket, but without the Drone and Loop modes.

One more comment about SSW – all of the instruments from Sonar (I’m assuming these are VST instruments) are showing up in the SSW VST plug-in selection window. Meaning that the Cakewalk products might be compatible with SSW (I haven’t even started learning how to take advantage of Sonar yet) and vice versa. I really wish I had more time to mess with this stuff.

Ok, I’m cheating a lot here, but the tutorial says it’s ok and I’m not going to argue with the tutorial. The DVD comes with 2 motion files, one is the cleaned up version of Rana’s lower half sidestep dance, and the other is the completed upper half arm swing. The tutorial uses the pre-created keyframes to demonstrate how Rana’s motion was determined, specifically to make her look cute. There’s a discussion of what motion curves to use where, and then instructions to copy-paste the keyframes to the end of the song. There are some suggestions of how to add a few variations to her movements, and then having her finish the dance with a “victory pose”. I’m considering whether to start work on a new project, which isn’t a dance video, so I’m ducking the “homework” for this volume in favor of spending the time on this new thing. But, I did move some of the accessory files around on the hard disk to make them easier to find in the future, and renamed a few others for the sake of readability. And I figured out how to delete accessories from an MMD project to replace them with something else (deleting Rana’s Room to use the Chocolate Room).

One thing I finally realized is that the room door texture files are intended to be used both in the Warp Room AND the room the door comes with. That is, all of the doors have two sides, one facing the interior of the warp room so you know where you’re going, and the other facing the interior of the room you’re in. Technically, it makes more sense for the inside of the door to say “Warp Room” so you know you’re looking at the exit. If you’re in the Chocolate Room, and the door reads “Chocolate Room”, how likely is it that you’ll look at the door and say, “OH, THAT’S where I am!”? Anyway, each of the door textures have to be copied from the folder of the room they come with, and then renamed to “dor1b” or “dor2b” in the Warp Room folder. So far, there are 5 rooms, but only 4 door texture files. Rana’s Room apparently isn’t accessible from the Warp Room (yet).

I’m really hoping for a desk accessory in an upcoming volume. There should be 8 more rooms before the series ends, but there’s nothing in the magazines saying what the future sets or instruments will be. Speaking of which, the sax model mentioned in the magazine a couple of issues ago still hasn’t been included on the subsequent DVDs. I don’t think I missed it, so maybe it will be on a later DVD. I hope so, because I want more instruments for Rana to play with. I’m really looking forward to a keyboard accessory.

Additional comments:
Note to self – To activate the Rana voice bank, go into the Program Files->Vocaloid3Tiny folder and run the Activate3 program. This is necessary about once every 2 or 3 months to input the next serial number.

Stacking Stars

(Raw shot of the brightest object that had been in the sky on Feb. 13th.)

I don’t know why I keep hoping that I’ll be able to do something new for the first time and have it come out right. I should know better by now.  Anyway, I’ve been wanting to do astrophotography for years, and recently I decided to actually see what I could accomplish. Initially, I did have an 8″ Newtonian telescope, but only a cheap pocket camera that I couldn’t get rigged in front of the eyepiece. Now, I have the CoolPix, but no telescope. So, the best I can manage is to set up a tripod and just accept what I can get.

(Stacked result.)

The first step was to learn more about stacking. Ian Musgrave has a decent tutorial on this. The basic idea is simple – stars are shy and hate sticking around for people with cameras.

Well, maybe not. There are two big problems with taking star shots at night – the lack of light, and the annoying tendency for the Earth to rotate. These two things work against each other if you have a fixed tripod (instead of a telescope with a motorized mount). To get enough light, you need long exposure times. But, the longer the exposure, the more streaking you get from the shot. One way to get around this is to limit exposures to 30 seconds (or less), and take LOTS of photos one right after the other.

(Raw shot of the big dipper.)

I’m told by a sales clerk at Bic Camera that my particular CoolPix can’t support a remote trigger, that I need a more expensive camera. And, after fiddling with the settings, I know I can’t save photos in RAW format (I’m stuck with jpeg compression). Finally, I can do manual focus with auto exposure, or vice versa, but I want manual focus AND manual exposure. So, the only choice is to take lots of notes, fiddle with the photos at home, and then try again until I get it right. So, this is what I did:

1) Use a tripod. The bigger and heavier, the better.
2) Select Programmed Auto. This lets me change ISO and activate interval timer.
3) ISO (originally ASA) indicates how sensitive the digital camera sensor is to light. According to Photoxels, the higher the ISO number, the better the picture will look in low light. I kind of got this mixed up in my head, and I used ISO 100 for some of the shots, and 200 for a few others. Regardless, the exposure times automatically set to 30 seconds, which was what I was counting on. What’s important to keep in mind is that at higher ISO values, there’s going to be more sensor noise. I’m going to have to play with this setting to determine the best trade-off between light sensitivity and noise.
4) The CoolPix has a continuous interval timer, but it’s only available in certain programmed modes, not in the full Auto mode. Fortunately, one of those modes is Programmed Auto. In continuous interval, I can specify down to 30 second periods, but this seems to be 30 seconds from the end of the last shot, and not from the start of one shot to the start of the next. Since the shots take 30 seconds each, it’s closer to 6 minutes from the beginning of shooting to the end, if I only want 6 photos of a particular target.
5) Press the trigger and get away from the camera. Sensitive tripods can pick up jiggle from vibrations in the ground if you pace to close to the camera. Either way, Programmed Auto uses the same focus and exposure settings from the first shot through the subsequent ones, and turns off the display monitor between shots to save battery.

(Stacked result, plus color correction.)

That night, it was cold, maybe close to freezing. I’m pretty sure that’s why I got so many “stuck pixels” in the shots. These pixels are ones that don’t turn off when there’s no light hitting them. Generally, if the sensor has a problem with stuck pixels, you can correct for this by taking a couple practice photos with the lens cap on. I didn’t actually do this until the next day, when I was at home and the camera had warmed up. By that time, NONE of the pixels were stuck and I couldn’t cancel anything out in the stacking process. I’ll know better next time.

Another reason for taking the practice photos prior to, or immediately after the other pictures is that if you do have sensor noise, it can be averaged out when you do the stacking.

The idea behind stacking is two fold – remove stuck pixels, and some noise, from the raw photos; and, create a final image that is the summation of the light from the individual shots you took.

In Gimp, step one is to open your first raw photo. Then, drag in your practice shot (the one when the lens cap was on). Open the Layers toolbar, and select the practice shot and set the mode to “Subtract”. This will cause the stuck pixels to cancel themselves out, as well as remove some of the sensor noise (since the noise is random, you may want to repeat this step with more than one practice photo.) Use Layer->Merge Layers, then save the file back to hard disk. Repeat these steps for every raw photo.

Next, open the first corrected photo in Gimp, and drag in the second photo. This time, set mode to “Addition”. This will cause Gimp to increase the brightness of the total shots. You’ll also find that the Earth rotated enough that the stars in both shots don’t line up. You’ll need to zoom in and use the Layer Move tool to align the second photo with the first one. Then drag over the next few corrected raw shots one at a time and repeat the same mode and alignment steps. Ian Musgrave suggests using 5 photos, so that’s what I did.

(Raw shot of Orion’s Belt.)

Because I’m in the city, and I was only a few yards from some street lights, I had light pollution to deal with. It’s not a visible issue in the raw shots, but it does add up to be significant after 5 photos. Additionally, you may notice some streaking in a few of the photos. The more you zoom in on a star, the greater the impact the Earth’s rotation has on the shot for longer exposures. If you have a 30-second exposure, and a fixed tripod, you’re going to get streaking. The only way around this is to reduce the exposure times, which means messing with the ISO sensitivity, and the manual exposure settings. I don’t have a good handle on this yet, so I need more practice.

Also, the more you zoom in, the greater the impact of the Earth’s spin will be. By this I mean that the stars at the center of the photo may not move as much as the stars at the corners. You can try rotating the photos before stacking them, but this is very fiddly. And the amount of spin is going to depend on where you are and where you’re pointing. Being at the north pole and pointing straight up is not going to rotate the stars the same way as being at the equator and aiming the camera to the northeast. If you have a good camera with a remote trigger, taking photos in quick succession with short exposure times will reduce the amount of spin you get between shots. The alternative is to get a good software app that can automatically align the photos for you. (I don’t have this kind of an app yet.)

(Stacked result plus color correction.)

One temptation is to use Gimp’s Color tool to adjust the black levels of the finished stacked shots to get rid of some of the gray and make the pictures a bit clearer. This is kind of self-defeating, because you’re removing the light that you got from doing the stacking.

(Orion’s Belt closer up, raw.)

If you look closely at the stacked photo below, you’ll see a number of fainter blobs. These are what you want to be seeing with all the stacking you’re doing. These fainter blobs only show up in the camera if you stack many shots on top of each other. You’ll lose these if you color-correct the finished image by making the gray more black.

(Stacked result without color correction. Notice that the belt is showing some streaking.)

(Handle of the big dipper, raw.)

(And, stacked.)

Anyway, this is what I got after my first outing. You may ask why I didn’t experiment more when I had the chance. Well, first, there were some clouds during the night, and I had to shoot around those. Second, it was cold and I had been outside for a full hour; I’m not sure how much longer the battery would have lasted. Third, and more significantly, the authorities don’t like it when people stand outside in a park in the middle of a city at 2 in the morning, with a camera. I really need to find a photography club, so I can go with them out to the countryside. Or, for that “safety in numbers” thing.