Cooked another batch of nasi lemak today.
That requires the “high” burner on the stove cooktop, the one that JG King put far too close to the wall. So I put
our well-stained aluminium foil between to protect the wall. I didn't realize just how hot
it got: it caught fire! Sadly I didn't get a photo—after all, I had a fire to put out—but
the mess is obvious:
It's been 17 years since I (re)started this online diary!
Even at my age, that's nearly a quarter. Of course, another way to look at it is to say
that I left a 30 year gap since my last diary, which I stopped nearly 47 years ago. At that time I had been keeping a (paper) diary for over a
third of my lifetime, from age 14 years, 3 months to age 22 years. And I really regret that
I stopped. Reading that old diary now is really interesting.
But then it occurred to me: I have now kept a diary for a larger proportion of my life than
I had then. In October 1970 I had kept it for 93 of 264 months, or 35.2% of my life.
Adding in the additional 17 years, I've now kept a diary for 297 of 827 months, or 35.9%.
Not completely by coincidence, today was spent tidying up diary files. Obviously I
finished the file for August 2017. Less obviously, I
finished entering the file for March 1965. I'm entering 4
days per day, and I had one left over, 1 April 1965. And that was
the only day remaining in April 1965, so I finished that too. Another couple of months and
my entire diary will be on line.
Picking up my “new” Sansui San110 wasn't easy. The seller responded to my messages quickly
but inaccurately, and by this morning (intended date of pickup) I still didn't know much in
the way of details. He appeared to refuse to divulge his address. The correspondence went
something like this:
Groggy: How about Saturday late morning? That would be a whole lot easier for me.
Seller: ok will have to be the am
Groggy: OK, give me a time and an address.
Seller: just call that number as im out and about in the am for the directions about 10 to pickup by 11 then I have to shoot off
So I called at 9:00 and got voice mail. OK, try again at 10:00 and got him with a barely
intelligible mobile phone connection, but was just about able to understand the
address, and some garbled statement about meeting him in his car near mumble. Finally
off and got there. Car was there, with cat in it, and finally the seller emerged from the
house. Took the amplifier and remote control from the car, handed it
to me, drove off, and that was that.
The pickup itself was straightforward enough, but the arrangements were really irritating.
Still, I have the amplifier—possibly the biggest I have ever seen, at 39 × 43 × 13 cm (21.8
litres). By contrast, my old Technics amplifier is 30 × 41 × 12 (14.8 l), and the receiver
in the lounge is 33 × 44 × 13 (18.9 l). Certainly the volume/price ratio is good, but it
means that it won't fit where I had the Technics. One of
the SPARCStation pizza boxes is
going to have to go.
Over 50 years ago we lived in Kuala Lumpur, in a street (lane) called
Lorong Kuda (“Horse Lane”) on the edge of the racecourse, on the edge of the town. Some
decades later one of our friends from the time, now a big businessman, got involved in
replacing the racecourse with an enormous building complex, which for a while included the
tallest habitable buildings in the world, and called it
“Kuala Lumpur City
Centre”. Why Centre? It was several kilometres from the accepted centre of town.
Clearly that has made its mark. To get
from Dereel to the western suburbs
of Ballarat we take a way that, although
obvious, is completely ignored both by Google
Maps and all navigation programs I know. The last part involves going down Cherry
Flat Road, across the Glenelg Highway into Wiltshire Lane.
Over the last year or so there has been considerable building work on the corner of Cherry
Flat Road and the Glenelg Highway. And on Thursday the complex was opened: a shopping
centre called DelacombeTown Centre.
OK, you could hardly call Delacombe a city—it's only a suburb of Ballarat, after all, with a
current population of 6,297—but you could also hardly call the location a centre. Today it
had a sort of second opening (“celebrations”) with lots of events. Since we were in town
anyway, we decided to go and take a look, particularly since there was the chance of special
offers.
Getting in isn't the simplest thing in the world. It's on a corner, but the entrance is at
the southern end of the complex, down Cherry Flat Road. And where are the parking spaces?
There are a few when you arrive, but most are on the north side. And to get there every
single car has to go between the two buildings, and across a pedestrian crossing. Here's
the current aerial (“satellite”) image from Google Maps, made some time before completion:
Not surprisingly, things were seriously clogged up. I wonder what it'll be like in a
month's time. The car parks were full to capacity, and I chose to park outside on the
highway:
And sure enough, we found lots of special offers, mainly fish. But somehow the whole place
looks dingy and unfriendly. Each of these two images was taken in a different building:
Where did that come from? It seems to be from shin height, clearly optimized for something
else than “customer experience”.
Across the road to Kmart. Yvonne was concerned about whether we would be allowed in with our
Woolworths trolley. I had no such concerns, but she asked anyway. And they advised her to
leave it outside: it seems that they automatically lock up if they're taken further than a
certain distance from the shop. While that's potentially a good idea, it would probably
mean that we couldn't get it back to the car out on the highway. Not good.
At Kmart we found a frying pan “frypan” that we had been looking for: 24 cm,
supports induction, and cheap. But where do you pay? Here's the entrance:
That's my car on the extreme right. I had been the only car there when we arrived 40
minutes earlier.
Since we were pointing west, and the traffic in town was particularly bad, decided to take a
long way home via Ross
Creek. I wish I had an overview of what the GPS navigator thinks it's doing. In the
end carried on
through Napoleons, and saw a
garage sale near Sullivans Road, just south of Napoleons. I don't think we've ever been to
a garage sale where Yvonne didn't buy something, and this
time it was two rubbish bins for $30 together. Normally you pay about $100 each. But of
course we were in the wrong car, so we had to go home, and then I out with
the Commodore to pick them
up.
After picking up the rubbish bins, headed down Misery Creek Road again to see how things
were progressing. Once again I saw plants that I can't identify:
The Grevillea bedggoodiana
are still not really in flower. It's interesting in passing to note their preferred
habitat, hanging down the edges of the road:
On the way home, down Peppermint Track, which I thought might lead back to our house. After
reaching East Dozed Track (where do they get these names from?), found a barrier:
That's a pity. As a check on the map shows, I was only 1.6 km from home, and the track went
on for at least another 300 m (if you believe the map), or possibly even joins with Stones
Road (if you believe the aerial view).
This week we ate trout that we found in
the Delacombe Town Centrethis morning. There were a
couple of differences from normal. Firstly, Chris Bahlo wasn't there:
Finally got round to installing the amplifier today, requiring a bit of rearrangement of my
desktop (the area above the desk). But apart from the usual issues with cables, everything
Just Worked. For one system, anyway. Connecting to dischord (nomen est omen?)
didn't work, but that seems not to be due to the amplifier; the inputs worked when connected
to eureka.
What really got me was the difference in sound quality. This is no
high-end Hi-Fi system, but with the old
loudspeakers it's much better than the toy Logitech speakers I've been using so far.
Well worth the trouble.
About the only issue is the user interface. When I turn it off, it forgets its settings:
which channel? How loud? What balance? My old Technics had mechanical memory (switches
and potentiometers), but here it's all digital. I wonder when digital interfaces will
attain the level of maturity that analogue devices had 20 years ago.
Or did we? The recipe that Yvonne found included both grapes and
raisins. It also wanted the quails wrapped in smoked bacon, something that we didn't have.
In the end we faked it:
We were too lazy to peel the grapes, especially as the instructions (“put into boiling water
for 30 seconds to loosen the skin”) didn't have the desired effect. The result was not bad,
but I can see us improving on it further.
CJ Ellis has problems with his MyNetFone connection again! It's the third time now: for reasons that nobody has been to
explain, outgoing calls function normally, but incoming calls divert immediately to voice
mail. This first happened two
years ago, then repeated six
weeks ago. CJ is non-technical and deaf, and has difficulties understanding
non-Australians (including Yvonne), so I offered to talk to
support for him.
On each occasion the MyNetFone support people told me that
the ATA (a Mitron MV1) was not registered. Wrong! The ATA was registered, otherwise CJ would
not have been able to make outgoing calls, including the ones he made to report the problem.
What they meant was that the ATA was showing up in their system as “not registered”: a
network problem, not a hardware problem. And they should really notice the difference.
OK, just because it has happened twice doesn't mean that the same symptoms this time have
the same cause. First, check the settings on the MyNetFone web site.
Login failed! How come? Checked again and in the end asked for a password reset (“Enter
email address...”). But it didn't like the (custom) email address, the one to which they
send all notifications! Why does this sort of problem happen so often?
In the end gave up and called the MyNetFone number. For further reference: select 1 for
“residential”, and then 2 for support. Once again I had to put up with really flaky hold
music, which sounded as if it had a poor network connection, and the classic announcement:
While you are waiting, please consider this: Please stay on the line for the next
available operator.
Finally connected to Shakeel, who laboriously spelt out his name, then asked me for my name
and date of birth, and after being given the ticket number asked “How can I help?”,
presumably the next line of his script. I suggested he looked at the ticket, after which he
told me that the cable to the ATA was flaky, and that the ATA itself—sold to CJ 2 years ago
as part of the bundle—was obsolete, and should be replaced. While writing this up, I
checked: it really seems that CJ got the last of the stock, as this article, posted at 4:48 PM 2 years ago, describes.
With a bit more encouragement he decided that the ATA was not registered. I explained once
again the one-way nature of the problem, but he kept going on about the flaky cable. To
quote last time, literally, I asked him to escalate it, but said he couldn't. Asked to
speak to his supervisor, but he couldn't connect me, and I could expect a call back in 24 to
48 hours!
Finally I lost my temper when he repeated things yet again, and he hung up on me. At the
time, completely by chance, CJ was standing in front of me, something that happens no more
frequently than twice a month. I told him “He says the ATA isn't registered”. CJ: “Then
how can I make outgoing calls?”. If a 78-year-old non-techie can see that, why can't the
support people?
So I called back and asked to make a complaint. They wanted details all over again, but I
refused to give them: too much time wasted already. Connection to the complaints department
took a while, but finally I was connected with Ash(ley), who sounded more sensible, and who
knew all the details. It seems that somebody must have checked the calling number and done
a lookup, something more complicated than I would have expected. Ash arranged to have a
senior techie call me back again.
Finally that happened round 16:10: Cory called and we discussed the matter, in particular
the discrepancy between the support view and the view from the ATA. He understood, but
there was a problem: the ATA was now registered, which I was able to confirm. Clearly there
was nothing he could do under those circumstances, but he promised to have somebody keep an
eye on it from time to time, and gave me a number to call if it happened again.
Why do these bugs go into hiding just as you're about to catch them? People often bring up
conspiracy theories: first they fix things, then they claim that there has never been a
problem. I think it's more likely that they fix things while investigating, and don't
notice the fact.
When taking panoramic photos, alignment is very important: you rotate the camera about
the entrance pupil of the lens.
How do you determine where it is? If you can find them documented online, and you trust the
data, use it. For Olympus
lenses there's a table on the German
Olypedia, but it hasn't been maintained for a long
time, so modern lenses aren't there. I've adapted this to my equipment. Panotools also has an entrance pupil “database”, but
it currently doesn't include any Micro Four Thirds cameras.
Then there's the alternative of measuring it yourself. This page seems to be the easiest to
understand. But the method relies on successive approximation and is slow, tedious and
inaccurate.
But then it occurred to me: how does parallax manifest itself when stitching a panorama?
Larger “error” values from the control point detector. And they should be proportional to
the distance from the entrance pupil. So if I take comparison shots with the lens
positioned at two different positions, I should be able to interpolate between the two to
find the correct position. To do that, I positioned two tripods in front of each other up
against a brick wall:
These photos were taken with my
Olympus Zuiko Digital ED 8 mm f/3.5 fisheye lens, which according to Olypedia, after
my adjustments, should have an entrance pupil setting of 84 mm. So I took two shots at 80
mm, then another two shots at 90 mm:
The first thing that occurred to me is that I really only needed one tripod close to the
camera. The wall was at a constant distance, so the error was very small, and the front
tripod really stood out. The corresponding control point displays were:
In each case, the control point on the tripod head is highlighted, and the error is more
than 10 times any of the others.
Next, there's an assumption: I assume that the entrance pupil is between the two positions.
The alternative is that it's outside the two positions. In either case, the error is
proportional to the distance from the entrance pupil position. If the entrance pupil is
between the two points, the 10 mm distance correspond to an error of 14.39 (6.21 + 8.18)
pixels. The entrance pupil should thus be at 80 + 6.21 ÷ 14.39 × 10 mm, or 84.3 mm—as close
as practical to what Olypedia says.
What if I'm wrong in my assumption? Clearly the entrance pupil is closer to 80 mm than 90,
but it could be less than 80 mm. In that case, the 10 mm correspond to an error of -1.97 mm
(6.21 - 8.18), so the entrance pupil position would be 80 + 6.21 ÷ -1.97 × 10, or 31.5 mm.
That's clearly ridiculous in this case, but in others it's not so clear. I don't have any
information at all for my Leica Summilux 25 mm f/1.4, so I guessed it would be somewhere near the front
element, at about 40 mm on the scale, and took photos at 30 mm and 50 mm. The results:
error 1.34 for 50 mm and 3.45 for 30 mm. In this case the entrance pupil could either be at
50 - 1.34 ÷ (1.34 + 3.45) * 20 mm (44 mm) or 50 - 1.34 ÷ (1.34 - 3.45) * 20 or 62.7 mm.
Which is it?
Based on position, I'd go for the former, but it's easy enough to check: more photos at that
position. I need to do that anyway to prove that my method works. And I also need to take
more photos to do a proper write-up, which will include real distances rather than the
markings on my focus rail.
What I've learnt so far:
The method seems to work.
You only need one tripod.
These photos should be taken at minimum aperture (f/22 where available, f/16 with the
Leica lenses).
It's worth being a little more generous with the distances between the measurements.
Off to Ballarat for a dental checkup.
Somehow I seem to be going to the dentist and periodontist all the time lately—today was the
fourth time since March. As
then, he found nothing wrong, but cleaned my teeth for me. I was dubious about the
necessity, as I was last time, but he handed me a mirror and showed how he cleaned the scale
off the inside of my bottom teeth. That's convincing, but why don't toothbrushes do their
job?
On the way home, as planned, took a look down Misery Creek Road, on the off chance that
something might have changed. About the only thing that has changed, apart from the weather
(for the worse), is all the Acacia bushes
in flower. Still, it looks like I have 3 months of observations ahead of me.
I've commented from time to time on the importance of shipping to the information age. I
can buy things in China and have them
shipped here for a total price of less than normal letter postage
in Australia. Not so from
the USA: I seldom buy things from there
because the postage makes it uneconomical. And this statistic bears it out:
At the time I linked to
https://infographic.statista.com/normal/chartoftheday_10922_parcel_shipping_volume_and_parcel_spend_in_selected_countries_n.jpg,
but that's an old, worn-out chart. They have updated it, but it's not the same thing:
Analysing this information, it seems that the average cost of a package in the USA is $7.40,
while in China it's $1.94. And presumably a far higher proportion of Chinese packages are
international, further skewing the relationship.
The other thing I don't understand is the speed. I really wish I understood why postage was
so slow. Decades ago I read a book
by Thomas
Hardy—Jude the Obscure, I
think, in which there were mentions of letters passing
between Dorset
and London by train. Letters were sent in
the morning, and a reply received the following morning. That can't possibly happen any
more, because there is only one postal delivery per day. But if things could have been so
fast then—late 19th century—why can it take over a week for packages to reach me
from Melbourne, a shorter distance?
The clothes host is about 2 m from the ground. Upright, he must have been at least 1.7 m
high. And the posture clearly shows how dangerous his claws are:
Last week ALDI had some special offers:
a dashcam and a wireless reversing
camera. I've had a reversing camera before, which connected to my GPS receiver, but it
required running wires through the car, and I never got round to installing it. This one
least obviates that problem.
The reality, as so often with ALDI special offers, is different: the camera part can be
mounted either wired to the reversing lights, or unwired with a switch. So when you want to
reverse, you get out and press the “On” button. If it's wired, it only works when it's
powered, either by the reversing lights or the tail lights. Neither is really satisfactory:
what if you want to keep an eye on your trailer load?
The dashcam is pretty bare-bones, as you'd expect from something costing $40. It doesn't
even have a storage card, something that ALDI have always included in the past. It came
with protective film over the camera lens and the display. After removing the lens
protection, I discovered that the protection had been stuck to the lens with a
not-easily-removable adhesive:
More to the point, though: where do I put the devices? They're all supplied with a suction
cup for attachment to the windscreen, but that takes up much of the field of view:
Round about this time I need to ask myself what I need the devices for. For me, the dashcam
is basically just another way to record my route; it would be nice going down Misery Creek
Road, for example. But for that I need GPS logging.
I have GPS logging! It's just in a different box, and never the twain shall meet. And then
there's the reversing camera: all the display on the windscreen does is show the images from
the camera. My last GPS receiver could do that too, module wiring. So basically, there's
no reason to need more than one display on the windscreen.
And that's exactly what a mobile phone or tablet can do, sort of. Mobile phones are too
small, and the quality of the software available subpasses even these cheap dedicated boxes.
One day, maybe? With a GPS receiver that gives fixes at least once a second?
Call from CJ Ellis today: Gmail was preparing to
delete his email account:
From: gmail.com <info@libdb.stromungunmittelbar.com>
Date: 6 September 2017 at 13:04
Subject: Your account will be -deactivated
To: cjhimself@gmail.com
Hello cjhimself,
This is to inform you that your request on: 9/3/2017 1:30:15 p.m. to remove
your account from gmail.com server has been approved and will initiate in
one hour from the exact time you open this message.
Regards.
ignore this message to continue with email removal or
Please confirm if this is genuine as i have not made any such request.
Regards,
Google
NOTE: If You Receive This Message In Your Junk or Spam Its Due to Your
Internet Provider
That's what I see after it has been forwarded, of course. Gmail tastefully avoids
displaying domain names like stromungunmittelbar.com (it could at least contain an
ö), so there was almost no way for him to know that it was spam. Even the out-of-sequence
date is no indication. About the best was that gmail itself had removed it as spam.
What did it want? Scare people? Receive confirmation for spam lists? The last line of the
text suggests that the perpetrator didn't think it through very well.
I've been buying things online now since the end of the past millennium, and it would be my
favourite way to buy things—IF. You'd think that by now people would have got their
act together, but in fact, on reflection, there seem to be problems Every Single Time.
Today I finally came to the conclusion that I needed a new lens, the Olympus M.Zuiko
Digital ED 8mm f/1.8 Fisheye PRO. What do they cost? B&H are my price benchmark: not necessarily the cheapest, but reasonable. They
want US $999, but of course they're in the USA, so the very cheapest shipping is another
$39. USD 1,038 corresponds to about AUD 1,260.
What about eBay? The cheapest I could find
was $870, including shipping. That's a lot better, and it's from my old friends eGlobal, who offer it for $869 outside
eBay—with another $19 shipping!
OK, there's this other company, DWI,
who clearly don't do business in the USA:
DWI is American for "Drink Driving”. I'm
told that they're really another face of eGlobal, but with different prices and conditions.
And how about that, they're offering it for $839 with free postage! Still better, they
offer me a $10 discount for my first purchase. $829 sounds a whole lot better than $1,260.
And I got an empty display, repeatedly. OK, accept the “chat” offer at bottom right (are
they all run by Zendesk)?), where I was
told to do what I had just done. When I explained that it didn't work, I was then given a
code, possibly a generic one: RMN2016. 15 minutes fighting web site breakage, $10
gained—maybe.
The following day, while writing this article, it worked:
At least that suggests that the problem wasn't at my end.
Next, create an account, sign in, create profile. Like eGlobal, they have differentiated
shipping charges. In their case, it's free to big towns, and some unspecified cost,
probably also in the order of $20 more, to ship to places
like Dereel. But we've seen that before,
and Chris Bahlo works in Ballarat
(really Mount_Helen), so I can enter her address as the shipping address.
But that doesn't work! Yes, they maintain both billing and shipping addresses, but when I
update one, the other gets updated! After 10 minutes of frustration, more chat.
Checkout did accept the promotion code I was given, and showed a total of $829.
Over a period of 10 minutes, I was able to enter Chris' work address and have it stick,
only for the shipping address, after only 6 attempts. But then the price had gone back
to $839!
Going back to correct the price showed nothing wrong—until I discovered that it had
forgotten the shipping address! Jane on Chat said “don't do that”, and that I should
ignore the information until checkout.
One more entry of Chris' (long and complicated) address, and then to “payments”. 1.8%
surcharge for anything except bank transfer! It would have been better if they had said
that up front.
Finally I got to checkout (still without paying, since I needed an invoice number). All
went well, and finally my order was placed. Time so far: 60 minutes.
Next I needed to perform the bank transfer. Where do I enter the invoice number? The only
place I could do that was the “Description” field. The minimal help suggested that I would
get that description in my bank statement, and that the recipient may get it from his
bank. Thought about this for a while: probably it will work, but just to be on the safe
side, I included the invoice number in my name field as well. Done? Almost.
Press “Submit”. “This session has timed out”. The whole bloody thing all over again. This
time it went through. Finished, at 16:30.
Total time: about 60 to 70 minutes. Yes, this one was particularly painful, and I still
don't have the warm, fuzzy feeling that the incoming money will be paired correctly with my
order, but we can hope. But why must things be so difficult 20 years into the Age of the
Web?
Apart from the pain involved in my order from DWI, it's worth considering what the
alternatives are for online camera purchases. Here's my view:
Suppplier
michaels
B&H
eglobal
DWI
Price, with shipping
$1,320
$1,260
$888
$829
%age of michaels
100%
95.5%
67%
63%
Warranty
1 year Australian
1 year international
2 year seller
1 year seller
Web site quality
bearable
acceptable
painful
unbearable
michaels is a leading traditional camera
shop. I have bought from them in the past, but not recently. I've downgraded their
otherwise useful web site because I can't find shipping prices until I create an account.
I'd guess that this is negligence on the part of the web programmers rather than any malice
on their part. My price includes a guess of $21 for shipping.
B&H's price of this lens is over
$1,000, so it's quite possible that I would have to
pay GST on it.
I also cooked (baked?) them in the oven, mainly to ensure even heating. I set the oven to
140°, but that was too warm, and later dropped it to 130°. Maybe that is also too warm.
Next time I'll try 120°.
Off to Ballarat again this afternoon to
have my neck kneaded. Gradually I'm getting the feeling that things are improving, but it's
been a slow and gradual process. The sessions—still $0.80 each—are getting less frequent,
and next time will be 10 October 2017.
We're planning
raclette for dinner tomorrow, and I had
to buy the cheese. The cheese shop that I once found so good has gone downhill, and when I
went there yesterday, they only had Heidi raclette. Their “Swiss” cheese tastes good—a fair imitation
of Gruyère—but I've had bad experiences with it for fondue, so I have certain reservations. Asked the saleswoman, but she didn't
understand “raclette”; she pronounced it sufficiently differently that the real
pronunciation didn't ring a bell.
That appears to be what foxes leave of a lamb, though I'm told they usually leave more, just
eating eyes and liver. I'm surprised it was left in the middle of the road.
At some point I should transplant them. Peter Jeremy, who gave me this tree as a sucker,
tells me that they should be frost hardy, but possibly that's his observation
in Sydney.
Apart from the obvious suckers, there's another plant there:
That's clearly not a curry tree. It looks for all the world like
a Hibiscus rosa-sinensis,
and I think it must be. We'll see when it gets a bit bigger. I hadn't expected a hibiscus
to self-seed.
Raclette for dinner this evening, and I
discovered that the cheese I bought yesterday had not been cut into slices. It hadn't even
been cut reasonably straight:
We've been looking for German TV series on the web for some time. It's not easy. There
seem to be three main sources:
Old series: on YouTube, generally in a
quality that varies from bad to execrable. I've found some that had compressed 16:9
down to a mobile phone format, something like 6:9. That can't look good even on a
phone. And the resolution typically starts at 360p—the resolution of low-resolution
British TV 50 years ago—and goes down to an unrecognizable 144p.
Then there are torrents of dubious legal status. They're uninteresting for another
reason: there's almost no German-language content.
Then there are the „Mediatheken“ of various German TV channels. They're thoroughly
confusing: first, the structure of the public stations confuses the matter, then the
interface is confusing, and access is haphazard. Sometimes it works, sometimes it
doesn't, and sometimes they refuse to play for copyright reasons. And here, too,
resolution is not what you'd want. About the best thing about them is that you can be
sure that they're legal, if you can watch them.
Spent some time today looking further, and I'm still not sure I understand everything. But
there's a web site fernsehserien.de that
at least gives an overview of what's on on TV at the moment, and may help find it online.
It's at least another piece in the jigsaw puzzle that is online TV.
Our old raclette grill is showing its age (probably about 20 years), and if we could find a
replacement, we would. It must have heard us talking: the power switch failed when I tried
to turn it on. How to fix that?
Where can I find a corresponding bit? I used to have lots of bits with things much stranger
than Torx, but that was in another life. As it happened, though, I found one pretty
quickly, and was able to bypass the screw relatively quickly. I was back at the dining
table and testing for live chassis in 13 minutes:
Apart from that, it seems that the Heidi cheese really isn't the best choice. Like many
other kinds, it melts so completely that there's hardly anything left:
What's the best way to take the dinner photos? I asked myself this question two weeks ago, but the method I used
there only really works when there are two or three people. The week after next we're
having another Arne Koets clinic, and
apart from Chris Bahlo, we'll also have Margaret Swan and Chris' sister Melanie.
The table's round, so the answer's clear—or is it? A circular panorama. Tried that this
evening, with the Vanguard
ALTA PRO 263AT tripod contorted differently:
That's a cylindrical panorama, so I can extend it as far as I want round the table, and it
will look as if people are sitting in a straight line. Once again the question arises as to
what we should be doing. I certainly shouldn't be displaying the remote control.
That's a particularly difficult scene to expose; this was done with 2.3 EV overexposure. Arguably it should
be done with no composition and spot measurement:
While making a regular backup to my external backup drive, I had a strange experience: at
the very end, it didn't complete. Checking, I found that the backup drive was no longer
mounted. But umount was the very last thing that happens in the script. Further
investigation showed:
Yes. The disk was /dev/da2, and I was seeing round 140 32 kB reads per second! This
went over a long period of time, and umount used considerable CPU time in the
process:
What caused that? ps showed that the process was started at 9:32, and it didn't
complete until 9:35. I suppose I'll have to climb into the code and see why it's reading at
all after umounting.
More investigation of online German TV today. There's lots of stuff to read, and the
signal-to-noise ratio is
low. But I came across onlinetvrecorder.com, which, despite the name, is German. If I understand things
correctly, I can download anything recorded in the last few days—for free. Well, only
during the “Happy Hour”, the third of the day between 0:00 and 8:00 German time,
corresponding nicely to 8:00 to 16:00 here in the winter, or 10:00 to 18:00 in the summer—in
other words, the time I would choose anyway. By the time I found this (the help is well hidden), it was
past 16:00, so put the rest off until tomorrow.
Not a good idea: it has been raining a lot, and the middle paddock is sodden. But it turned
out that there was a lot to see, starting with a mob of kangaroos:
It wasn't until I processed the images that I saw the flower stalks, out of focus in this
image. Clearly I need to go back with more appropriate equipment.
Yvonne has bought another saddle
in Austria. How do we transfer the
money? Last time I only
mentioned the arrival, not the hoops we had to jump through to pay for it.
The main issue seems to be
that Australia hasn't quite made it to
the 17th century when it comes to money transfers: cheques are still the preferred method of bank payment.
Last time Yvonne had to go to a physical bank to perform a transfer which cost an arm and a
leg. For this reason, amongst others, I've been trying to open a bank account
in Euros, so far without success.
I looked again today, and how about that, there's progress. Both our banks now allow
international (”overseas”) transfers from web banking! I had to apply for the privilege in
each case (why?), and in the case of ANZ I was
told to wait until tomorrow.
OK, Bank of Melbourne. Somehow I
found a way to calculate the exchange rates and costs before actually making a transfer,
though I can no longer find it:
Their modern rendering make it difficult to see that I had entered the sum of 800 € for this
comparison. OK, how does that compare to real exchange rates? According to OANDA,
the Interbank rate is 0.67085 €.
Factoring in the fee (why a fee?), the exchange rate is effectively 0.6213 €, a whopping
7.3% more! Even a credit card only costs 3%.
I've been here before, and I've been frustrated. Try again. The Currency Shop gives me “5 Cheaper Ways To Transfer Money Overseas”, none of which
seem very cheap; they even mention the banks. To get any idea, you need to select “Compare
Exchange Rates”, which came up with this top savings:
It's not until you read the small print that you discover that “it's not for you”. Still,
it was worthwhile following up on, and in the end I was able to get an exchange rate of
0.66638 €, only 0.67% below the Interbank rate. Amusingly, the rates changed while I
watched them, reflecting changes in the the Interbank rate. In addition, the first two
transactions are fee-free, so that's all I paid. The actual sum I transferred was 852 €, so
the costs were: 5.71 €, compared to (coincidentally) $99.99 from the Bank of Melbourne:
Company
Rate
Fees
Total
Total cost
(Interbank)
0.67085
1,270.03
0
Bank of Melbourne
0.62130
20
1,370.02
99.99
HiFX
0.66632
0
1,278.66
8.63
For once, that looks good, though the extra $15 from the third transaction make it less
attractive.
This was taken with my macro ring flash. Apart from the inherent difficulty in keeping a
dog still for the photo and the unexpected difficulty in focusing the M.Zuiko
Digital ED 30 mm f/3.5 Macro, how do you set the exposure? If he moves only a few
centimetres forwards or backwards, the light intensity will change enough to require a
different aperture. This image was in fact underexposed, and I had to compensate to get a
useful photo.
The technical solution has been with us for decades. Starting in the 1970s, flashes
acquired the capability of terminating the flash when enough light had been reflected from
the subject. And then TTL flash was
introduced (it seems by Olympus), involving the camera too. But most after-market flashes seem to target
Canon and Nikon, and there's now precious little available for Olympus.
My current ring flash, a Viltrox JY-670 (URL likely to rot) doesn't have significantly more intelligence than the flashes of the
1960s. Yes, I can set the light output, but that's all. What automatic ring flashes are
available? Metz (mecablitz) make one, the
mecablitz 15 MS-1. I've discussed this topic in January 2010 and September 2014.
Reading the instructions (PDF) is always instructive, if not in the manner that the manufacturer
intended. The good news seems to be that there's only one model, and it can handle all
kinds of camera. But nearly all of the instructions relate to remote operation,
where the flash is not mounted on the camera. What kind of confusion could have caused
that? Who uses a ring flash anywhere except on the lens? Of 46 pages, only about 2
describe specifics of using it as a ring flash: the reflectors can be set at 0°, 10° or 20°
to adapt to the distance from the subject, and there's a bounce diffuser for distances under
10 cm. That's good to know.
But the price! The cheapest I have seen is about $430. By contrast, my current ring flash
cost about $100. Apart from the swiveling reflectors, the only thing missing is the
automatic exposure function. And the mecablitz comes with only 3 adapter rings. I can pay
up to $60—close to the cost of my current ring flash—for each additional piece of metal!
Then Andrew Richards on the M43 Tech Talk Facebook group came up with the Dörr
Macro TTL Blitz DMF-15 für Olympus und Panasonic. It looks pretty close in specs to
the mecablitz, and it's slightly cheaper. But where do you get one? eBay doesn't seem to know them, though I can find a
couple of suppliers in Germany. But do I want a unit from an almost unknown manufacturer?
The name Dörr rings a bell, but now they describe themselves as „Foto Marketing“, and their
“about us” page suggests
they have only 50 employees. Do they make their own equipment, or is it a rebadged (and
unknown) Chinese product? The appearance suggests that it could be: like my Viltrox, it's
clearly a modified on-camera flash.
More to the point, though: is a ring flash the answer? Kev Russell, moderator of the
Facebook group.
sang the praises of his Olympus STF-8 flash unit (clearly a model name thought out by a non-native English speaker). He says
it gives better lighting. It's certainly much dimmer: guide number of 8.5 instead of
15 for the mecablitz, my Viltrox, or even the current crop of cheap LED “flashes”. That's
only 32% as bright. Once again it comes with few adapters, suited only to the M.Zuiko 12-40 mm
f/2.8 “Pro” (which I have now sold) and the M.Zuiko Digital ED 60 mm
f/2.8 Macro and the M.Zuiko
Digital ED 30 mm f/3.5 Macro lens. And, of course, it's even more expensive than the mecablitz.
But how do you expose it? I really need HDR images. Can I be
bothered to get a tripod? No. So I took it hand-held (with my standard +3 EV/-3 EV bracketing, offset by +1.3
EV. And how about that, it wasn't bad. I should do more of this.
For reasons I forget, I accessed Gmail from the
browser on my :0.1 display instead of the normal :0.2. Same browser
(outdated firefox), same
computer. But I still get a message:
I suppose things are getting better: now at least it gives a (correct) IP address. But how
does it recognize the “device”? It's the same one I use all the time, and so is the
browser. I wonder if I will get new messages when I upgrade the browser.
Discussing yesterday's money
transfer article on IRC today. Peter Jeremy came up with a couple of alternatives:
payeer.com and paysera.com. At first glance, both look
usable, but of course the first glance is often misleading. Certainly I'll take another
look at them before my next transfer.
Now wouldn't it be nice if I could transfer my pension into it and convert at their rates?
I've been experimenting with various Asian food for breakfast. One thing I have wanted was
something Korean, mainly so that I can have something to eat kimchi with. Today I tried faking something with fish,
dofu and spring onions, along with rice, fried egg and the
oh-so-Korean ikan bilis. Yvonne was eating poached eggs, so I tried using our fried egg rings to
fry an egg that would fit nicely on top of the rice. Not a success: the yolk broke, and the
egg stuck to the ring. What a mess:
What's wrong with this questionnaire? There are so many things I can think of:
How can I know whether the law needs changing? They haven't stated what the current law
is.
What do they mean by “marry”? My understanding of marriage is a religious and cultural
one. I got married because I didn't want my daughter to be a bastard. That reason
(presumably) doesn't apply to same-sex marriage. And the religious aspect has nothing
to do with the law.
Why is there any legislation at all about marriage? Should we not be looking at the
definition of marriage itself? I can't see any merit in legislating who may and may not
marry.
What's a “dark pen”? Does it need a nib? What colour should the ink be? If this is a
requirement, why don't they supply one (“Australian Bureau of Statistics Same Sex
Marriage Memorial Pen”)?
On the whole, I don't feel qualified to decide on a matter that is obviously of extreme
importance to the people affected, but which has absolutely nothing to do with me. My
feeling is that they should be allowed to do what they want. But enshrine that in law, and
there will be loopholes and unforeseen side-effects. I'll be interested to see if the new
law (and I'm sure it will be passed) will allow people to marry themselves.
So what do I do? Abstinence is one option, but I'd rather abstain explicitly than just not
return the form. Maybe I'll deface it, thus at least making some kind of point.
Where is Chris' office? It has one of the longest addresses I know outside
the United Kingdom: Greenhill
Enterprise Centre, Ballarat Technology Park, University Drive, Mount Helen, Victoria 3550.
And still it's missing a street number!
Maps can't find Greenhill Enterprise Centre, Ballarat Technology Park, University Drive,
Mount Helen, Victoria 3550.
Make sure your search is spelled correctly. Try adding a city, state, or zip code.
OK, how about just Ballarat Technology Park? Yes, it finds that:
But there's nothing there! It's on the side of the road. The right-hand building is IBM,
but the left-hand building isn't marked. Technology Centre? Off to take a look. No, some
Government department.
But it wasn't that bad: Yvonne came with me, and we found
Chris on the corner of Enterprise [sic] Grove, talking on the phone and holding the
bag with the lens. I didn't even get the chance to say thanks: Yvonne grabbed the bag, and
we drove on to Caledonian Transport, despite the name a local company, in Daveyduke Drive. To our surprise, it was 17.7 km
away! It's only 27 km from home to the university. But Google Maps is (almost) equal to
the challenge: its preferred route is 26.2 km:
Finally we had the thing and headed back home, stopping off at a couple of furniture shops
(the real reason that Yvonne came) to look at lounge room furniture:
It wasn't until I got home that I looked at the Google Maps again. The place that it had
given me (with pin) was a “bus interchange”, as the intuitive marker indicated:
What's that? American for bus stop? I can't see that more than one bus would stop there.
In any case, the real Technology centre is a couple of hundred meters east-south-east, but
that's not where Google Maps puts its pin:
The difference in size is all the more amusing when you consider that the new (small) one
has a maximum aperture of f/1.8, while the old one only has f/3.5, two stops less.
Spent some time trying to determine the position of the entrance pupil, using the method I
worked out last week. It didn't quite work out the way I expected: the first pair of images didn't align at
all well. I think I need to be more careful with my control points, but that will have to
wait until tomorrow.
The new raclette grill looks quite different from the old one. It's rectangular rather than
round, and it comes with a stone
for Teppanyaki, something I've never
tried, and which Yvonne forbids me from trying indoors:
The pans are considerably deeper than those of the old grill—not a disadvantage, I think—and
the distance from element to pan is also nearly double the distance of the old one. As a
result, the food gets browned more slowly, but more evenly:
We tried some ham on the stone, which was at 270°, but it didn't seem to cook as fast as on
the metal surface of the old grill. And it left quite a mess, which was almost impossible
to remove:
House photo day again today, but first
I needed to determine the position of the entrance pupil of the new Olympus M.Zuiko
Digital ED 8mm f/1.8 Fisheye PRO. Last week I had worked out a
way to triangulate the errors between two positions to find where the no-error position.
Based on the position of the front element, I chose 60 mm (behind the element) and 80 mm (in
front). At 80 mm all went well:
The control point detectors had found things close to the edge (where the illusion of an
“entrance pupil” starts to break down), and also things not on the plane of the wall.
Basically, the results were useless. So I tried again, this time masking out everything
except the wall in one of the images. The results then were much better:
Conveniently, the difference in the error was almost exactly 80 (55.84 + 24.10), and that
spread over 20 mm. So I interpolated and came to the conclusion that the correct position
was at 66 mm. Tried that and... got an error of 16.91!
What went wrong there? There's some reason to believe that I collected the wrong data.
I'll have to repeat the whole procedure. In any case, for today I decided that I could try
a bit further, at 70 mm, and how about that, it worked: an error of 4.15, which is easily
good enough. But I still need to find out how to avoid this kind of problem.
And the results? Still to be analysed. The difference in position on the rail makes it
visible at all times, and it limits the height of the resultant photo. Here full height and
trimmed:
Mark (or is that Marc?), a plumber
from Cape Clear, along today
to see about fixing the taps that we've been waiting for for over two years. Problem:
they've been stored badly, and the chrome surfaces have developed blisters. Given the
price, it makes sense to buy new ones. He also does tiling, so we may finally get round to
covering all the bare concrete areas.
Middle of the month is time for the monthly flower photos. There's really not much this
month. The last month has been unseasonally cold and wet, and not much has come up apart
from spring bulbs. The frosts have taken their toll:
ALDI has
had quail on special again, so Yvonne bought some more for freezing. Under the circumstances—once
again Chris Bahlo wasn't here for a Saturday night dinner—we decided to eat a couple of them
fresh, using the cailles aux
raisins secs recipe.
This time I wasn't as happy. They need to be cooked in a pot for 30 minutes, and though
they tasted good, they still weren't as tender as I would like. Possibly this is a
candidate for cooking sous vide.
For some time I've been fighting the online content („Mediatheken“) of German TV stations.
They have just about everything there that they broadcast, but hidden behind such an
unpleasant interface that you'd think they were trying to stop you from seeing it. Recently
we watched a film from the
ARD
(Bayerischer
Rundfunk). „Falsche Siebziger“. Quite good, and even the image quality was good
(ABC take note), but
the data rate wasn't sufficient to watch without interruption.
In the past couple of days I've done some investigation and discovered on both the
CHIP and Heise-Verlag a program called MediathekView, a program for finding, viewing and downloading TV programmes
from all German and many other European web sites.
Is it legal? It must be, otherwise two leading publishers wouldn't refer to it, at least
not without mentioning the legal status. So I installed it and tried it out. Typical
undocumented user interface (and one that somehow
shouts Java), but it
works, and I was able to download three films with no problems.
Now why is the normal web interface so painful? Lack of care?
It's been nearly two years since Olympus came out with firmware support
for Focus stacking, and I have yet
to completely make friends with it. Every time I
use it, I have to go back to the instruction manual and read what little it can tell me.
Today's photos of
the Hibiscus
and Phalaenopsis were done with
focus stacking, but they were only the result of an incredible amount of fiddling. Time to
write up some notes:
What Olympus calls focus bracketing, where the camera takes up to 999 (!) images with
marginally different focus settings, starting at the closest. The default is an equally
unrealistic 99 images.
What Olympus calls focus stacking, where the camera takes exactly 8 images, starting at
the mid-range and moving first closer and then further away. After taking the photos,
they are merged into a JPEG image
covering about 90% of width and height. It also saves the original images in the format
specified in the camera settings (in my case raw) and also in JPEG, whether you ask for
it or not. The merged JPEG image is in addition to the other 8, so in my case I get a
total of 17 images. The result is like this, first the JPEG image, and then the same
photos processed with DxO Optics “Pro” and
merged with external software:
Kim Holst summarizes the in the following image (click to get full size):
The original was on Facebook, but it seems that their links rot quickly.
The distance moved is controlled by a value that Olympus calls “focus step”, a number
between 1 and 10 that has some relationship to the size of the step. The larger the number,
the larger the step. Apart from that, there is no reliable information. Some say that the
size of the step also depends on
the depth of field, and some say
that step 1 means a step of exactly the depth of field, so that for any two images the near
limit of the second image is the far limit of the first image. That doesn't make sense to
me for two reasons: first, that would make the larger steps far too large, and secondly
“depth of field” is a matter of opinion: how unsharp can something be? In addition, it
doesn't line up with my inaccurate observations. One thing is clear from observation is
that the step is not constant: it gets larger with focus distances.
In each case, the camera uses the electronic shutter. This places severe (and undocumented)
restrictions on the shutter speed with flash: 1/8 s maximum for the Mark I, and 1/50 s for
the Mark II. The camera will happily shoot at higher speeds, giving images like these (E-M1
Mark II, 1/250 and 1/60 s):
When using flash, you can extend the time between shots to allow the flash to recharge.
There's no feedback from the flash, of course, so if you choose it too short, you'll get
uneven exposure.
Setting the parameters
There's nothing obvious about where to find the settings for focus stacking. They're in the
camera 2 menu, under Bracketing.
This time last week I bought a disk on eBay.
Unlike just about everything nowadays, there was no PayPal option, so I had to pay by bank transfer. But this morning I received a mail
message from the seller, and discovered that he had opened a non-payment case.
How could that happen? No idea. The eBay site showed that it had been paid, but after
checking my bank accounts, I found that no transfer had been made. Either I forgot it
(unusual), or the bank messed up, which wouldn't be the first time. There's no way to tell
which it was any more. Either way, the seller was right that he had almost certainly not
received any payment. And clearly eBay messed up by marking it “paid”.
OK, pay up, in the process running into a bug on the ANZ web site: I copied and pasted the details to be sure I got them right, but it
managed to get a leading space, and claimed that the account number wasn't a number. It
accepted it after I removed the (almost imperceptible) space, but by then it had truncated
the account number.
Back to eBay. So I had paid the bill. What next? I had no less than three messages from
the seller, and all I could see were the headers. How do I display them? There's no
indication on the index of how to do it. But there
is a link “Tell us what you think”. Clearly they're not interested: the link is
broken.
Finally I found that the message I had received was sent by real email, and not eBay's
concept of email. How did he get my (eBay) email address? Maybe from the communication
after I purchased the item?
Back to the unpaid item case. The instructions say:
When paying by electronic funds transfer, include the transaction code that you were given
when you checked out.
What transaction code? There's an email ID, too long for the banks' antiquated systems, and
containing # characters, which aren't in their
(48
character BCD?) character set. And that's all.
Anyway, the item page also includes the information:
OK, Pay Now? I've already done that in real life, but maybe I need to tell eBay too. Nope,
nothing there, just bank details and the Seller notes:
Please include your USER NAME. Otherwise your deposit may be difficult to identify.
Thanks.
It also included an RIB key, whatever that may be (identical to
the BSB number, which they choose to
call “bank code”), and IBAN
and SWIFT information (not used in
Australia, and thus “not specified”. What happened to the transaction code? An old,
worn-out magic word?
OK, how about “contact seller”. That's a generic contact page, including their irritating
“We'll look for an answer for you, whether you want us to or not”:
I don't bake as much bread as I used to, and it's been nearly 2 months since the last loaf.
Fortunately the sliced bread keeps well in the freezer, so we always have some. But the
last starter smells funny, and there's a dark grey patch that seems to indicate an infection
at the top of the pot.
That's not because of the length of time. This starter was two months old, from the last
loaf I made. That's comparable in age to the length of time I left starters when I was
baking more frequently. And though the bread rose, it was barely enough, and it took nearly
5 hours. I've been cultivating this starter for 8½ years. I think it's time to buy an alternative
starter and see how this it works out.
It seems that I'm not the only person to have done so. But why not? It says so here.
Ah, but that's fake news. Real official statements come from Twitter.
More seriously, though, what is the purpose of the bar code? And what are the dangers of
publishing it? I can only think of two concerns: that the results of the survey could be
falsified, or that my personal details could be compromised. I'm not overly concerned about
my personal details, but it does beg the question as to whether they can identify my by the
bar code. That shouldn't be possible. And if they're checking the bar code, it's trivial
to reject all but one (or all) of the forms with a specific bar code. And the instructions
say that the form must be returned in the envelope provided, so submitting a duplicate would
mean that another vote would go unregistered.
And if they don't want us to publish the code, why don't they say so on the accompanying
“information” sheet that came with the form?
I've mentioned in the past that I'm quite happy with the Mendhak GPS
logger. It works well, while the corresponding function in OI.Share is fiddly to turn on, and it
frequently turns itself off again for reasons that I haven't established.
But today I had no logs at all. The entire directory was empty. But the logger was
working:
Only nothing got written to disk. Why was the
message Gpx10FileLogger.write in red? Is that an even more polite
way of saying “failed, but I don't really want to tell you”? I had deleted the files in the
directory via WiFi File Transfer. Could it be that it set the directory permissions to something
that GPS logger can't access? Who knows? Everything about Android seems to be
designed to hide details. In the end I removed the directory, restarted the device and GPS
logger, and it worked. But why is this stuff so painful?
Arne Koets is, if not in town, then at
least in the area, and he's holding a clinic in Clarendon,
near Scotsburn, where there
have been a number of such clinics. He has 8 participants, 4 of whom came for dinner this
evening: Yvonne, of course, Chris Bahlo, Margaret Swan and
Chris' sister Melanie. More fun with dining table panoramas:
Those are only two of 15,552 combinations of the individual images. Things are getting
better, but I'm still not happy with the lighting. A bit of fill-in flash from the camera?
And maybe marginally more serious poses?
Oh, and food. Grilled tuna with baked potato, pumpkin and sweet potato, followed by a
“Quark Sweet Cheese Strudel”, whatever that may mean.
There's something funny about MediathekView:
sometimes it doesn't find things that are listed in the Mediatheken, and sometimes it can't
load things that I can watch directly, claiming “forbidden”
(HTTP error 403). Why?
One reason could be something that I've frequently observed: my network block (“Class C”)
was allocated in Germany in the dark ages, round 1992. When I moved here, it moved with me,
but much software still thinks I'm in Germany. In general it's a minor nuisance—some broken
sites of Big Business decide to redirect to their German subsidiary when I access their
site, for example—but in this case it appears to tell the Mediatheken that they can display
the content—if I use a Class C address and not the gateway address.
OK, how do I work round that? eureka.lemis.com is the gateway machine, but for some
reason I can get the content there, currently only via a web browser. How about installing
MediathekView there? Of course it doesn't say that it works on FreeBSD, and there's no port. But it's
in Java. How hard
can it be?
Not very hard. As the instructions said, I needed to install Java 8, and then it ran—for a
fraction of a second. Then it died with the message:
JavaFX wurde nicht im klassenpfad gefunden.
Stellen Sie sicher, dass Sie ein Java JRE ab Version 8 benutzen.
Falls Sie Linux nutzen, installieren Sie das openjfx-Paket ihres Package-Managers,
What does that mean? Basically, that it couldn't find JavaFX, whatever that
is. ktrace showed that it didn't look for anything of that name (and that the output
of the error message was messy, with the newline characters output with a
separate write system call). And that I should ensure that I
have JRE version 8, along with some
Linux-specific information. I had that. Configuration error?
Spent a lot of time looking for details of how to configure Java, along with lots of web
sites dealing—mainly inappropriately—with the error message. Finally I discovered that the
last line of the error message was relevant: in FreeBSD terms, it meant that I should have
installed the java/openjfx8 port. Problem: there is none,
only java/openjfx8-devel. What happened to the stable version? It seems that
there is none.
Installing Java 8 was relatively straightforward, but openjfx8 is a whole different kettle
of fish, and it took forever, paradoxically including a new version
of ruby. But then it
ran, which means I no longer need to fight Microsoft to use it.
But I still can't get my content. Whatever is failing the test, it's not the system
itself. Wouldn't it be nice to just store things from the web browser?
Yesterday's dinner panorama wasn't bad, but Margaret Swan's face came out a little
underexposed: she was sitting side-on to the flash unit, and the reflection from the ceiling
wasn't enough. OK, we can do that: a little bit of fill-in flash with the toy flash that
came with the Olympus
OM-D E-M1 Mark II, and use it to trigger the room flash. Here's the arrangement. The
whole thing is mounted on the Vanguard 263
AT PRO, with the centre column spread out horizontally, with the end over the
centre of the table:
The end of the column is a minor problem: I need something to turn things through 90°. I
used an old ball head, turned through 90°, and with the broken Sunwayfoto rotator on top:
And sure enough, the test photos worked fine. But not the for-real photos. It looked as if
the room flash didn't go off, which is more likely to mean that it went off too early,
probably a pre-flash. Here one of my tests, and then the real thing, without any
correction:
It's tiring work, and I was glad when the 45 minutes were over. Margaret Swan didn't have
any such problem, though she did manage to fill up
an SD Card with her continuous video:
The whole horse photo affair was somewhat frustrating. I had taken my two big bottles, the
Zuiko Digital ED 14-35 mm f/2.0 SWD and the Zuiko Digital ED 35-100 mm
f/2.0. Unlike last time, I was closer to the action this time, so I didn't need the
35-100 mm lens at all. For once, most of the recording was video, and I discovered the hard
way that the Four Thirds
system lenses don't support continuous autofocus, which made the 14-35 as good as
useless. Instead I used the Leica DG Vario-Elmarit 12-60 mm f/2.8-4.
And even then, I couldn't find a focus system that worked. Continuous autofocus with
tracking seems to ignore the AF targets. It works well enough as long as it can see the
horse's head, but when the horse turns away, the focus point wanders off into nirvana, and
it's almost impossible to catch it again. And continuous autofocus without tracking gets
lost if I let the horse leave the target area. Somehow I'm left with the feeling that
there's no good way to do this. I wonder if other cameras do it better.
On the way home, drove down the first couple of kilometres of Misery Creek Road. As
expected, things have changed considerably since last time: it's warmer now, and these
orange pea-like flowers are everywhere, interspersed with at least two kinds
of acacias. One kind is a bush:
In the forest, they really stand out, but they're barely recognizable in the photos. How
can I emphasize them?
Once again I'm impressed by how many of these flowers are close to the east (highway) end of
Misery Creek Road.
This Hardenbergia violacea
is directly on the turnoff from the highway, and it's the best specimen I found:
Why can't I download Mediathek videos that I can watch online? It all sounds like a web
proxy issue. So how do I set a web proxy for MediathekView? Once again Google to my aid. It
pointed me at this forum
discussion (!), which states that you should start MediathekView like this:
On Monday I discovered I
hadn't paid for an eBay item, so I rectified
that quickly. Sent a message to the seller. No reaction. Sent another message to the
seller. Still no reaction.
OK, time to take further steps. Where eBay had claimed that the item was paid when it was
not, it now claimed that it was not paid when it was. Back to eBay's confusing web site.
“Take action”. What action? I've been there before on Monday. OK, get eBay to
call me “valued customer” and walk me through. Maria called back pretty quickly.
Problem: she had a strong fake American accent (probably originally from
the Philippines), and I really
couldn't understand her. I don't like doing it, but I had to ask her to get somebody else
to call me back.
That was Michelle, not much more intelligible. I'm beginning to wonder if it isn't the
telephone. I must try changing to the other phone some time. In any case, she was not able
to walk me through the matter: there is no way to “take action” in this case. “Take
action” means “Pay with PayPal”, and if I pay
by other means, the system can't handle it. I need to contact ask a question of
the seller and get him to “answer” it by closing the case. She was prepared to give me his
phone number. No, sorry, I want some kind of record. OK, she would contact him. That
sounded like a reasonable workaround of the site breakage, so we left it at that.
And that's where it stayed. No evidence of any activity. OK, I can open cases too. Sent
the seller a message asking for response by tomorrow evening.
Some of there things are inevitable, but eBay's system makes it much more difficult.
By coincidence, received a message from eBay sent to my second account, the one I set up
trying to fix the breakage they caused three years ago. It seems that they're
celebrating their 18th birthday! I've been with them since July 1999, so I must have signed
up with them some months before they were born. There's an explanation, of course: this is
eBay Australia, but they didn't tell me that.
Six months ago I
finally gave up on TV, mainly because of the appalling usability of the software available.
We still have live TV reception, but I gave up on that in late 1983 when I got my
first video recorder. Clearly technology is overtaking the concept of transmitting
information only at fixed times.
So it was interesting to see that I'm not
alone, at least if I pretend to be a young adult:
Here they distinguish between a TV set (“only”) and a digital device (“only”). Apart from
the fact that I'd expect the vast majority of people to use both, what do they mean by
“digital device”? Everything in a modern TV set is digital with the exception of the tuner.
I suppose they're lumping mobile phones, tablets and computers (remember them?) into the
latter category. But there are enormous differences and considerable overlaps. My TV is a
real TV, with tuner and toy digital functions, but I use it only as a large monitor. And
that's why: I can get a decent sized image. In an age where people are rushing to
resolutions of 3840×2160 and larger, displayed on screens with diagonals of 160 cm and more,
who wants to be limited to a mobile phone with a diagonal of, say, 10 cm and an aspect ratio
that makes much of that useless?
Finally a day that was warm, sunny, and not too windy. Just what I need to spray the
weeds. And the battery of the spray unit was flat! OK, put on charge. Came back half an
hour later to find the thing rattling to itself and producing smoke. Had I forgotten to
turn the power off? No, the switch was off. But this thing has a stupid knob
(potentiometer?) that bypasses the
switch. It looks as if the unit is dead. I think that's worth a return to the seller.
I've done many attempts at focus
stacking over the last couple of years, and even bought a tripod to help. But I've
never used both outside the house. Today was the day, taking photos of
the Pterostylis:
And this time I definitely hit the limits of in-camera processing (“focus stacking” as
Olympus defines it).
Here both foreground and background are out of focus.
OK, “focus bracketing”, the main difference of which is that I can take more than 8 images
(a ridiculous 99 by default). But what's the correct number? I guessed at 20, but when I
looked at the images, I found I had 35! That turned out not to be such a disadvantage: I
used 24 of them. But how did I get that number? Did I set the number incorrectly?
Checking the camera, I found it set to 9.
There's something funny here. It seems that the camera stopped because focus had reached
infinity—maybe. According to the Exif data,
the last 7 images had a depth of field reaching infinity. Theoretically I could have failed
to save the new number of shots—the menu system makes that easy—but that doesn't explain why
the camera was set to 9 when I checked. And it also doesn't explain why the camera carried
on taking 6 further images after reaching infinity. There's more stuff to puzzle about
here.
The weather was good today, and tomorrow they're promising rain and wind, so I took my
house photos a day early. After
last week's experience, it
was clear that I needed to remount my rail so that it wasn't so visible, they way I did
at the end of July.
That worked, up to a point. I still have the issue of the side of the panorama rail in the
right hand side of the image:
That will be difficult to remove. It really does stick out in front of the horizontal axis,
by about 2.5 cm. The horizontal angle of view of the lens is only 144°, so I can minimize
it by moving it closer to the camera, but that brings the rail back into view. I wonder if
any specialized panorama hardware addresses the issue, but I rather suspect that not.
Fortunately it's not that important. Today I had another problem. I left the cable for the
remote shutter release hang forward:
On the whole, I think that for flare the old fisheye is (barely) better than the new one.
But I think both are very good. There aren't many lenses that can deal with direct sun in
the image. I must compare with some other lenses.
CJ Ellis has problems with his MyNetFoneVoIP service again! That's the fourth
time that he has had the same issue: outgoing calls work normally, but incoming calls divert
immediately to voice mail.
This time it happened at a particularly inopportune time for me, with
the Arne Koets clinic taking up most
of my time, so by this morning I had had time to consider what I could do without getting
MyNetFone support involved. Clearly there's something wrong if
the ATA thinks it is
registered, and can make phones to prove it, but the network thinks it isn't. But what's
causing the problem? Could it be the ATA itself? In my experience, they're particularly
unreliable, and this one is a cheap, old model. How about power cycling, long enough to
drain the electrons allow the registration to time out (a maximum of 20 minutes) and then try again. So we did
that, and how about that, it worked!
What next? Is it necessary to let the registration time out? Next time I'll get CJ to try
a normal power cycle (2-3 seconds) and see if that does the same thing. And maybe we should
be lining up a new ATA.
Yesterday Petra Gietz brought us a couple of large bulbs, of a name that she has forgotten.
They should produce large white trumpet-shaped flowers. I've planted them to the right of
the house entrance:
Tuesday's dinner panorama
wasn't an unqualified success. In fact, it was a barely averted disaster: somehow the
on-camera toy flash caused the room flash to fire at the wrong time, probably too early.
Time for more experiments. Last time I had the toy flash set off the room flash as a slave.
Is it possible that the timing was insufficient? The wireless flash triggers also have a
cable input, so I tried that.
For the fun of it, I also tried using a different lens, the Panasonic Lumix G 20 mm f/1.7, which doesn't see much use. Testing it before the
event was surprising: it hunted for a while and then took a completely out-of-focus image:
Disaster! After more playing around, it really seems that the focus on this lens is much
worse than on the Leica-branded
15 mm and 25 mm Summiluxes. That's disappointing, and it makes the lens much less valuable to
me. But at least this test (without toy flash) confirmed that the cable solution works,
even if it looks messy:
OK, back to the issue with the flashes. There's a known problem with on-camera flash that
they make a short pre-flash to eliminate red eyes. That would trigger the room flash too
early. So I checked the settings, and no, no pre-flash. The result:
No flash at all! How could that be? Somehow I need to do more checking. On the off chance
that it might be flashing in advance to help with autofocus, I turned that off, but it made
no difference. More experimentation needed.
Today's real results weren't so different from before:
We buy lots of stuff at ALDI, but not fresh
meat: past experience has made us wary. But last week Yvonne saw a good-looking piece of beef filet (“eye fillet”, as it's called in Australia):
After dinner, Chris said that Piccola—now
pushing 9 years old—had become much friendlier, and that she always used to run away when
she was young. Not my recollection, so I checked:
That was on 11 April 2009, and the diary showed that on 19 April 2009 we had gone to Mount Wallace to
pick up a colt. A colt? What colt? Chris couldn't remember. But then, she doesn't keep a
diary. So we went looking for videos. And yes, Yvonne had
taken horse videos in those days, with my Nikon
“Coolpix” L1, not a camera I bought for its video capabilities. For some
reason, Yvonne hadn't processed them, but we looked at them directly. And all of them had
the noise of an old dot matrix printer in the background. That was all the more surprising
because this was in the open air:
After a while we came to the conclusion that this was the autofocus of the camera. How
things have improved since then!
And no, by the end of the evening we still didn't know who the colt was.
An indirect result of our discussion about Piccolayesterday, I
noted an entry describing the pain of
taking house panoramas. Back to look at the images again today. Things have improved
considerably, and I was through with the whole thing in less than 30 minutes, with
considerably better results. Here before and after:
Strangely, the old images don't look as bad as I thought. Clearly I had expended a lot of
time lightening them up, but still the gradation is better with the new one, notably in the
sky on the right. And for some reason, there's a pink tinge in the highlights of the new
images, possibly a problem with DxO Optics “Pro” that I
don't care about enough to pursue.
More pain with eBay today, trying to get my
money back from the person who took money for the disk I ordered two weeks ago, but didn't
deliver. It seems that eBay won't help much because I paid by bank transfer, so I'll have
to take legal channels. But eBay told me what they thought of what I did today:
We have reason to believe that your eBay account has been used fraudulently without your permission. We’ve reset your eBay password. If you had your PayPal account linked to your eBay account, we've disabled your PayPal link to protect your funds. Any unauthorised activity, such as buying or selling, has been cancelled and any associated fees have been credited to your account. Any listings that we removed are included toward the end of this email. We assure you that your financial information is securely stored on a server and cannot be seen by anyone.
I've seen this too often, most recently in April, May and July. And though they claim to have
reason, they don't provide it, even when asked. I see this as reason to believe that it's a
bug that they don't want to fix. It would be nice to understand what it is that triggers
the bug. In the meantime, I have a whole lot of old, worn-out magic words
passwords of increasingly insulting nature.
It's no secret that Donald
Trump is a liability to his country. He's not making the USA America great
again: he's gradually eroding its importance on the world stage.
The USA has two main concerns on the nuclear
arms stage:
the Iran nuclear
deal, which Trump calls a “terrible deal for America” or some such drivel, and the
current escalation in North Korea.
The whole world is calling for a diplomatic solution for North Korea. But that's what they
have with Iran, and Trump is threatening to not honour the US obligations under the treaty.
What incentive is that for the North Koreans to negotiate?
His behaviour over the last week has been particularly appalling. It seems that he is doing
everything to provoke North Korea,
including calling their Supreme Leader “Rocketman”. I can only presume that he didn't call
him “Mr No Dong” because he wasn't aware of
the relevance.
Even without the crisis—it's hard to call it diplomatic—there should be things that are
beneath the dignity of a statesman, notably appearing at
the United
NationsGeneral
Assembly and threatening to obliterate another member state! And there were many more.
In the process, the North Korean responses show a finesse and use of language wholly unknown
to Trump. A dotard, he is—a word that the Washington Post thought appropriate to explain to its readers—or a barking dog.
And Ri Yong-ho showed
much more insight than Trump has ever done. But the real danger is that this kind of
insult-flinging may be OK in a school playground, but not between grown men who have their
fingers on atomic triggers. I don't see any great likelihood that it will come to war, but
somebody like Trump is the right kind of personality to drive the world into catastrophe.
He has to go.
The US constitution was written at a time when the American States were very concerned about
accountability, and it provides ways to deal with rogue leaders. But they're slow. After
the Watergate affair it took
over two years before Richard Nixon
saw no other way out than to resign. Does the world have that much time? The alternatives
are outside the law, but that's the stuff of revolution. Maybe it's time to consider the
options.
A new Washington Post-ABC poll found that two-thirds of Americans oppose launching a
preemptive military strike against North Korea, with a majority trusting the U.S. military
to handle the escalating nuclear crisis responsibly but not President Trump.
Apart from the somewhat ambiguous wording, only two-thirds oppose the idea of
a preemptive strike? So 100 million people are either for it or undecided? That's a truly
horrifying statistic.
Apart from that, though, it's clear that people in
the USA have different concerns. Here the
headlines from the Washington Post for today:
NFL Week 3: Steelers plan to skip anthem; Ravens, Jaguars protest President Trump’s
comments by linking arms, kneeling
Across the country, players decide what they will do before kickoff. And Colin Kaepernick
isn't even playing.
Colin Kaepernick vs. Tim Tebow: A tale of two Christianities on its knees
One is kneeling in private prayer. The other is kneeling in public protest.
Jaguars owner joins players during anthem protest in first game since Trump’s NFL
remarks
Shahid Khan, who had contributed $1 million to the Trump inauguration, locked arms with
his players in what is believed to be the first visible participation in relation to
anthem protests by a league owner.
Names I have never heard before. It appears to be a kind of football, so I suppose NFL
stands for National Football League, but that's so US American that the rest of the world
would have difficulty understanding it. It seems that the players are protesting against
police brutality, something laudable if it weren't apparently divisive.
How about “ever”?
The Federal Republic of
Germany came into being on 23 May 1949. This is the first time in the history of the
republic that a far-right party has been admitted to the Bundestag.
And not to be outdone, Statista produced a chart showing how the far right is on the increase in Europe:
Into the office this morning to notice an eerie silence. All computers were off. Power
failure? No. The RCD
had tripped. Why is not longer important, but there's a good chance that it's related to
the fact there's only one for the whole house. Next time we have an electrician here, I
should get a separate RCD installed for my office, in place of the current circuit breaker.
Back for the eternal fsck (I can hear friends
chanting ZFS!ZFS!ZFS!). Used the time to do a much-needed
tidy up of papers to be filed, and in the process noted that /home took something
over a hour to complete. And as on previous occasions, when it did, it wasn't clean, and
required a repeat fsck. Interestingly, the second fsck “only” took 20
minutes. But after starting, I had a problem I've seen before: the system crashed, this
time with a panic and a dump. I've been suspecting for some time that there is some
breakage in the /home file system that fsck doesn't detect nor repair, so I
wasn't surprised when I saw the backtrace:
(kgdb) bt #0 doadump (textdump=<value optimized out>) at pcpu.h:219
#1 0xffffffff80946582 in kern_reboot (howto=260) at /eureka/home/src/FreeBSD/svn/10/sys/kern/kern_shutdown.c:455
#2 0xffffffff80946965 in vpanic (fmt=<value optimized out>, ap=<value optimized out>)
at /eureka/home/src/FreeBSD/svn/10/sys/kern/kern_shutdown.c:762
#3 0xffffffff809467f3 in panic (fmt=0x0) at /eureka/home/src/FreeBSD/svn/10/sys/kern/kern_shutdown.c:691
#4 0xffffffff80d4968b in trap_fatal (frame=<value optimized out>, eva=<value optimized out>)
at /eureka/home/src/FreeBSD/svn/10/sys/amd64/amd64/trap.c:851
#5 0xffffffff80d4998d in trap_pfault (frame=0xfffffe085ba6d2b0, usermode=<value optimized out>)
at /eureka/home/src/FreeBSD/svn/10/sys/amd64/amd64/trap.c:674
#6 0xffffffff80d4902a in trap (frame=0xfffffe085ba6d2b0) at /eureka/home/src/FreeBSD/svn/10/sys/amd64/amd64/trap.c:440
#7 0xffffffff80d2f032 in calltrap () at /eureka/home/src/FreeBSD/svn/10/sys/amd64/amd64/exception.S:236
#8 0xffffffff8091ae54 in prison_priv_check (cred=0x0, priv=<value optimized out>)
at /eureka/home/src/FreeBSD/svn/10/sys/kern/kern_jail.c:3711
#9 0xffffffff8092fe86 in priv_check_cred (cred=0x0, priv=320, flags=0)
at /eureka/home/src/FreeBSD/svn/10/sys/kern/kern_priv.c:97
#10 0xffffffff80bacee9 in chkdq (ip=0xfffff80301960580, change=104, cred=0x0, flags=<value optimized out>)
at /eureka/home/src/FreeBSD/svn/10/sys/ufs/ufs/ufs_quota.c:196
#11 0xffffffff80b8efe6 in softdep_setup_freeblocks (ip=0xfffff80301960580, length=0, flags=2048)
at /eureka/home/src/FreeBSD/svn/10/sys/ufs/ffs/ffs_softdep.c:6862
#12 0xffffffff80b79afe in ffs_truncate (vp=0xfffff802de9bd938, length=0, flags=0, cred=<value optimized out>)
at /eureka/home/src/FreeBSD/svn/10/sys/ufs/ffs/ffs_inode.c:347
#13 0xffffffff80bb1b89 in ufs_setattr (ap=<value optimized out>)
at /eureka/home/src/FreeBSD/svn/10/sys/ufs/ufs/ufs_vnops.c:629
#14 0xffffffff80e710d1 in VOP_SETATTR_APV (vop=<value optimized out>, a=<value optimized out>) at vnode_if.c:799
#15 0xffffffff809f6d55 in vn_truncate (fp=0xfffff80036504780, length=0, active_cred=<value optimized out>, td=0x0)
at vnode_if.h:335
#16 0xffffffff809f1006 in kern_openat (td=<value optimized out>, fd=<value optimized out>, path=<value optimized out>,
pathseg=<value optimized out>, flags=1026, mode=<value optimized out>) at file.h:312
#17 0xffffffff80d4a05f in amd64_syscall (td=0xfffff802ddbde000, traced=0) at subr_syscall.c:141
#18 0xffffffff80d2f31b in Xfast_syscall () at /eureka/home/src/FreeBSD/svn/10/sys/amd64/amd64/exception.S:396
A call to truncate or ftruncate, running into trouble somewhere
inside UFS. I no longer have
the sources, though maybe I can reinstate them, but the “solution” (workaround) is obvious:
new, larger disk for /home, so I ordered one, along with one for Yvonne's machine lagoon, since it's clear that this rogue
eBay seller is not going to deliver. And
under the circumstances it's worth wondering if I shouldn't
use ZFS after all.
So what caused Saturday's
flash problems? I've been thinking about it for a while, and it still didn't make
sense. Today I took a more methodical approach. I mounted the camera on a tripod on the
other side of the table and took individual images of myself sitting at 90° to the position
of the studio flash, on the chair on the left of the image below. It was twilight, and the
curtains were closed, so outside light was not an issue.
Yes, the framing is wrong, but that's not the point. It's slightly underexposed. I later
discovered that the camera was 2.4 m away, and it has a guide number of 13, so the minimum
aperture I could use would have been f/5.8. The image looks in keeping.
That wasn't obvious at the time, so I took another image with -2 EV exposure compensation. It came
out darker, as expected (first image). Next, turn on the studio flash and try again (second
image):
That's the problem I saw on Saturday, and it makes no sense at all. If the studio flash
went off at the wrong time, I would expect no change. But somehow the effect of the toy
flash has been nullified!
Moving on, I took a photo with only the studio flash:
That's as expected. Some sanity, anyway. Add the toy flash again. And once again, no
image.
Can it be that the shutter speed is limiting the flash? That could happen if it responded
too slowly. So I set the speed to 1/20s and tried again. The only difference was that the
light shining through the curtain was brighter:
Could it be a problem with the toy flash? I replaced it with my mecablitz 58 AF-2 and tried again, first with the studio flash, and then without. In
each case the camera flash had -2 EV exposure compensation.
Still too dark! Could this have something to do with manual exposure mode? Tried it on
aperture priority, but the difference was marginal, although for some reason the camera
always selects 1/60 s under these circumstances:
It chose f/3.9 (maximum aperture, I think) at 1/60 s. And it's still too dark! At this
point it's worth noting that all I'm doing here is taking a flash photo under completely
normal conditions: on-camera only, TTL, let the camera do all the work. Why is it wrong?
Moving on, I tried the combination aperture priority, no exposure compensation, and studio
flash on. No difference from the attempt I previously made with studio flash off (first
image):
Whatever's going on, it seems that the studio flash is going off at the wrong time. None of
the images except the first show any evidence that it contributed to the photo. I had
already considered that the problem might be something that triggered the studio flash too
early via the optical trigger. I can't turn that off, but I can cover the receiver. First
a test photo with only the studio flash:
That's slightly underexposed (different illumination of each cheek). But as I mentioned
above, but didn't consider at the time, I was too far from the subject for the flash to be
able to illuminate it properly. At least I know what to do next Saturday.
What I now know is:
The problem is related to the studio flash being triggered too early.
The problem isn't related to the on-camera flash, since it happens both with the toy
flash and the mecablitz.
When taking these photos, I should not use any exposure compensation for the on-camera
flash.
But I'm still left with a surprising number of questions:
What is triggering the flash too early? It's not the autofocus: that uses the LED, and
it's far enough in advance that I can confirm that it doesn't trigger the studio flash.
Is there still something in the equation that is causing the completely unexposed
images? One hypothesis would be that the exposure measurement in the camera is being
confused by the studio flash going off, so that it turns off the on-camera flash almost
immediately. I should try this with another camera.
Why is there no image at all when the studio flash is triggered by the on-camera flash?
Could it be that the flash confuses the camera's exposure logic?
Why do I always have problems with on-camera flash, even under simple circumstances?
I've complained about this on numerous occasions, with numerous (Olympus) cameras and just about
all on-camera flashes. On 3 May 2012 I analysed the
histograms of the images and confirmed that they were underexposed. And on 19 April 2014 I had similar issues
with what should be perfectly normal flash exposure.
There are many more. It seems that the photos I took of Leonidyesterday are some of the very few
that have been properly exposed.
I've been meaning to keep my software packages up to date on all machines, but it's not
easy. I was going to catch up today on lagoon, Yvonne's machine, but the upgrade process wanted to remove Emacs and not replace it. That looks
dangerous, so I've put it off until I upgrade the machine, which should happen when I get my
new disks next week. Today I upgraded teevee, where the upgrade process wanted to
remove Samba. I can live with that: I don't use
Samba on that machine. But why does the upgrade remove software and not replace it? Is
this a political statement?
The other thing that becomes increasingly evident is how much bloat compilers have
developed. In the Good Old Days a compiler was something small, and so were libraries.
Today I installed, on a system on which I do no development at all, the following:
llvm 3.5 was enormous! Over 18 MB compressed files! But it's dwarfed by version
4.0, with over 300 MB of compressed data. What's in there? And why do I need to
upgrade 6 different compilers (2 gcc, 4 llvm or clang) on a
system on which I don't do any development? They'll never be used, but they require about
770 MB of downloads.
Clearly Donald Trump has
been too busy ranting about the allegiances of sportsmen to notice that there's another
natural disaster in US territory. The whole
of Puerto Rico (or is
that Porto Rico? That's what I hear
people saying) is devastated, and will be without mains power for the rest of the year.
What does Fearless Leader say? Nothing.
On the positive side, it's good to see some progress: Twitter is planning to make the President's life easier by doubling the maximum
length of tweets to a whopping 280 characters (coming close to the length of this
paragraph), Now can Fearless Leader's attention span keep up?
That's very different from a
couple of days ago. That's because of the marginally different question. But once
again it's horrifying that so many people can countenance mass destruction in Korea when
they were equally horrified by the killing of less than 3000 people in
the World
Trade Center16 years ago. I wonder what the
response would have been if they were told that a Korean rocket would then inevitably hit
the USA.
Into town for what seemed to be an endless list of things to do. One by one they fell
aside. Buy a lettuce. Return a defective hair trimmer—which proved still to be functioning
when I demonstrated it. Look for two-seater reclining sofas, of which we found none. Buy
some tool stuff at Bunnings, in the
process marvelling at the opaqueness of their stock layout. One of the things I wanted was
a 6 mm high speed drill bit. I gave up even trying to find them all and bought one which
didn't look too expensive. Then they had screwdriver bit sets. Price? Elsewhere. None
had the torx bits that I was looking
for—they were on the other side of the department.
Foie gras for dinner this evening,
something almost unknown in Australia.
I had managed to find some a while back, real goose liver (foie gras d'oie), not the duck
liver (Foie gras de canard) that is so prevalent nowadays. Opening the jar wasn't easy—I
needed a pair of pliers—and then I saw:
Last month I worked
out a way to use Hugin to
determine the entrance pupil of a lens based on the parallax error in the control points.
But two weeks ago I tried it
in earnest, and it didn't work out the way I had hoped.
What caused the problems? I don't know for sure, but I had a couple of ideas to follow up:
just plain sloppy measurements, inappropriate choice of control points, phase of the moon...
One thing was sure: the entrance pupil was in a position that caused that the end of the
focus rail to intrude into the image. Time for a bit of old-fashioned hardware: drill
another hole in the rail 2 cm from the end, so that the rail is positioned further back.
I'm really no handyman. I measured my 20.0 mm with a digital caliper with a 10 μm
resolution. Drill the hole
(in aluminium, with a cordless drill),
and suddenly the distance was about 22.3 mm. That's bad enough, but how do you measure
sub-millimetre distances between 6 mm holes? In any case, knowing the distance should be
good enough.
Back to the same measurements as last week, this time with initial settings on the rail of
100 mm and 80 mm. This time I positioned much closer to the wall, so that most of the image
was on it:
Then I masked off the tripod, and separately everything else except the wall. The mask
lines are very fine, so this is really only recognizable at full size:
Initially I did this on both images, but that's not necessary: one is enough to confine the
control points to that area. Ran the control point detector and got my results, not all as
good as I thought, so I removed the worst ones and tried again. The maximum error was 1.95:
That's on the other side, so the total error is 221.13 over 20 mm, close enough to 11 pixels
per millimetre. So the correct position should be 80 + (106.74 / 11), or 89.5 mm. Tried
that, and got:
37.49 mm! That's more than I could explain by experimental error: it corresponds to about
3.5 mm on the rail. There's something here that I still don't understand. More
head-scratching needed.
MediathekView, a program for finding,
viewing and downloading TV programmes, proved to be quite a find. But it didn't
quite live up to my claims “from all German and many other European web sites”. In fact, it's only some German sites,
and also ORF and SRF. These last two also only have a very limited
number of programmes available, and MediathekView doesn't even seem to be able to find them
all.
Is that really all there is? More looking brought up a number of alternatives: this one seems to have something to do with Australia, and this page describes an apparently legal, if somewhat dubious, way of downloading videos from ORF.
“This page” had the
URL http://www.chip.de/artikel/ORF-Mediathek-Download_47276441.html, but the
publisher has decided to (re)move it. It could be one of these results, but youtube-dl works just
fine.
I'm sure I'll come up with many more. But why is this such a mess? You'd think that
somebody would have come up with a consolidation site.
Saturday dinner again today, and with it the panorama photo. Tried again, taking into
account on my experiments with flash illumination on Tuesday. And it worked:
Somehow I'm still not happy. Is the complete blackout that I saw previously a bug in the
Olympus OM-D E-M1 Mark
II, or does it happen with other cameras too? Another thing to try out.
Do you have a comment about something I have written? This is a diary, not a
“blog”, and there is
deliberately no provision for directly adding comments. It's also not a vehicle
for third-party content. But I welcome feedback and try to
reply to all messages I receive. See the diary overview for more details. If you do
send me a message relating to something I have written, please indicate
whether you'd prefer me not to mention your name. Otherwise
I'll assume that it's OK to do so.
