Greg
animals brewing
food and drink gardening
general health
history language
music multimedia
opinion photography
politics Stones Road house
technology
Greg's diary
recent entries
Translate this page
Select day in February 2019:
Su Mo Tu We Th Fr Sa
1 2
3 4 5 6 7 8 9
10 11 12 13 14 15 16
17 18 19 20 21 22 23
24 25 26 27 28
Select month:
2018 May Jun Jul Aug
2018 Sep Oct Nov Dec
2019 Jan Feb Mar Apr
Today's diary entry
Diary index
About this diary
Greg's home page
Greg's photos
Network link stats
Greg's other links
Copyright information
    
Groogle

Thursday, 21 February 2019 Dereel Images for 21 February 2019
Top of page
next day
last day

Another power failure
Topic: general Link here

Another power failure today while we were cooking breakfast. And I didn't even notice until I tried to heat something in the microwave oven.

When did it happen? Both the microwave oven and the conventional oven are too polite to display anything until I tell them, so they were both showing “power on”. But when I went to look later, the second microwave oven in the pantry was showing 4:15 or so. What does that mean? What does it reset to? I've decided that it must reset to 12:00 (a likely value) and then count in 12 hour notation, so the power must have been reset 4 hours, 15 minutes previously. But I had thought that it reset to 1:00 (another obvious value). But that would give me a time later than when I discovered the issue, so I'll have to go for 10:10 and the oven resetting to 12:00. I'll have to keep an eye on it next time.


What does “ISO” mean?
Topic: photography, opinion Link here

Tony Northrup has brought out another video: “ISO is totally FAKE. Seriously”. I've already found fault with his videos; last time I accused him of deliberate falsification. This time I'm not sure: maybe he just doesn't understand the issues.

His argument is that “ISO” (he means the arithmetic sensitivity measurement according to ISO 12232:2019, or maybe the older ISO 12232:2006) is not consistent between makers, and that it doesn't mean anything anyway because it can be compensated for in post-processing. He gives examples with the Olympus OM-D E-M1X, which, he says, is out by 1 EV from the stated ISO, as demonstrated by photo from a different camera. And yes, indeed, the two images appear offset by 1 EV or so.

But what is ISO? He seems to think that it has its roots in the ASA system, which is only part of the truth. ASA is one of the most recent of a plethora of sensitivity systems and goes back only to 1943, over 60 years after Leon Warnerke brought out his first system.

He also seems to think that a typical rating has always been 100/21° ISO, when in fact for much of my youth that was a relatively high sensitivity. My father took ciné films with Kodachrome I, with a sensitivity of only 12 ASA, or (coincidentally) 12° DIN, and I normally used films with sensitivities between 25 ASA/15 DIN and 125 ASA/22 DIN.

But how do you measure photographic sensitivity? I have a book somewhere, but I can't find it. And there's not very much information about it on the web. Determining film speed states:

Film speed is found from a plot of optical density vs. log of exposure for the film, known as the D–log H curve or Hurter–Driffield curve. There typically are five regions in the curve: the base + fog, the toe, the linear region, the shoulder, and the overexposed region. For black-and-white negative film, the "speed point" m is the point on the curve where density exceeds the base + fog density by 0.1 when the negative is developed so that a point n where the log of exposure is 1.3 units greater than the exposure at point m has a density 0.8 greater than the density at point m.

It also shows this graph:

Which system? The graph refers to ISO 6:1993. And these slides state that for DIN 6-1974, the sensitivity is defined in the same way:


This should be DIN-1.png.  Is it missing?
Image title: DIN 1          Dimensions:          1803 x 1350, 175 kB
Make a single page with this image Hide this image
Make this image a thumbnail Make thumbnails of all images on this page
Make this image small again Display small version of all images on this page
All images taken on Thursday, 21 February 2019, thumbnails          All images taken on Thursday, 21 February 2019, small
Diary entry for Thursday, 21 February 2019 Complete exposure details

 

That's completely the other way round from the way you'd do it nowadays, where you go from the highlights. And it means that the sensitivity depends on the dynamic range of the sensor. Increase the dynamic range by 1 EV and you're also increasing the sensitivity by 1 EV.

Is that the way ISO 12232:2029 still defines it? I can't find out. I would have to pay CHF 118 just to look at the standard. The best I can do is this preview, which shows a tastefully truncated table of contents:

6. Determination of ISO spee
   6.1 General
   6.2 Saturation-based calcu
   6.3 Noise-based caclulatio
   6.4 Method of reporting

Another detail from the preview is interesting:

this document specifies an exposure index (EI) that corresponds to the focal plane exposure of a typical mid-tone.

This page also contains interesting detail. I haven't finished digesting it yet, but it basically confirms what I have written above, and goes into detail about the saturation-based and noise-based calculations. What I can say so far is: there's no single way to look at things.


Tony Nortrup's understanding of image capture
Topic: photography, opinion Link here

But Tony Northrup wasn't done with saying that some makers lie about camera sensitivity. As the name (“ISO is totally FAKE. Seriously”) suggests, he says that the whole concept is wrong. Got a problem with an image exposure? Fix it in post-processing.

That's an overly simplistic view of how it works. What we have is:

  1. Light falls on the camera sensor and causes a voltage in rough proportion to the light intensity.
  2. The camera electronics convert this voltage to a digital value, typically between 0 and 4095 (12 bit images), though some cameras have 14 bit sampling (0 to 16383).

How does this conversion take place? Tony doesn't go into detail, maybe because he doesn't understand it. What he does say is that you can take these numbers after storage and manipulate them. For a correctly exposed image, the highlights should be round 4095. For an image that's underexposed by 2 EV, they'll be round 1023. No worries, multiply by 4 and Bob's your uncle. He even shows an image that he corrected like this, and it looks OK on his video.

OK, I can do this too, in the process checking what he claims about sensitivity. Here's a boring photo of the back of my office, taken with my Olympus OM-D E-M1 Mark II with automatic exposure (P setting):

 
This should be Office-200.jpeg.  Is it missing?
Image title: Office 200
Complete exposure details
Dimensions: 600 x 450, 98 kB
Dimensions of original: 5184 x 3888, 2229 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Friday, 22 February 2019:
thumbnails    small images    diary entry

I was going to follow this with a manual exposure using the settings shown by my exposure meter, but they were the same. Whatever issue might exist with the camera, it wasn't the exposure metering. And does this image look underexposed? Not to me. Yes, there are issues in the shadows—that's why I chose this motive—but make the highlights any higher and they'll be burnt out.

OK, now let's try with 6400/39° ISO, followed by the same image (run the cursor over an image to compare it with its neighbour):

 
This should be Office-6400.jpeg.  Is it missing?
Image title: Office 6400
Complete exposure details
Dimensions: 600 x 450, 101 kB
Dimensions of original: 5184 x 3888, 4951 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Friday, 22 February 2019:
thumbnails    small images    diary entry
 
This should be Office-200.jpeg.  Is it missing?
Image title: Office 200
Complete exposure details
Dimensions: 600 x 450, 98 kB
Dimensions of original: 5184 x 3888, 2229 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Friday, 22 February 2019:
thumbnails    small images    diary entry

Yes, there's a minor difference in brightness there. Is it maybe related to the dynamic range? That would almost make sense. But it's certainly nothing like the 1 EV difference that Tony Northrup claims.

Next, let's look at his claim that you can compensate for exposure in postprocessing. Here's the same image again taken with the same exposure as the 6400/39° one, followed by the same image corrected in post-processing, and finally the correctly exposed 6400/39° one:

 
This should be Office-200-6EV.jpeg.  Is it missing?
Image title: Office 200 6EV
Complete exposure details
Dimensions: 600 x 450, 55 kB
Dimensions of original: 5184 x 3888, 881 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Friday, 22 February 2019:
thumbnails    small images    diary entry
 
This should be Office-200-6EV-corrected.jpeg.  Is it missing?
Image title: Office 200 6EV corrected
Complete exposure details
Dimensions: 600 x 450, 102 kB
Dimensions of original: 5184 x 3888, 4756 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Friday, 22 February 2019:
thumbnails    small images    diary entry
 
This should be Office-6400.jpeg.  Is it missing?
Image title: Office 6400
Complete exposure details
Dimensions: 600 x 450, 101 kB
Dimensions of original: 5184 x 3888, 4951 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Friday, 22 February 2019:
thumbnails    small images    diary entry

Yes, the results are surprisingly good. But what's that colour cast? My take is that that's a non-linearity in the sensor curves between the different colours. The problem is that multiplying all values by 32 means that the low-order pixels have values 32, 64, 96, 128 etc., and nothing in between. And that is the difference between setting the correct ISO sensitivity when taking the photo, and faking it after the event. Like converting to JPEG, you're losing detail with underexposure.


Kangaroo jawbone
Topic: animals Link here

While walking the dogs today, found this:

 
This should be Kangaroo-jaw.jpeg.  Is it missing?
Image title: Kangaroo jaw
Display location on map
Complete exposure details
Dimensions: 600 x 450, 137 kB
Dimensions of original: 5184 x 3888, 4777 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Thursday, 21 February 2019:
thumbnails    small images    diary entry

It's the lower jaw of a kangaroo, which Nikolai found on the side of the road and then decided that it was uninteresting. Leonid got it a little later and also decided that it was uninteresting, but not before breaking it: the join on the sides is quite fragile.


Friday, 22 February 2019 Dereel Images for 22 February 2019
Top of page
previous day

Contact details in Malaysia
Topic: general Link here

It's been 4 days since I sent a mail message to Tengku Razaleigh, and so far I haven't had much feedback. So how do I contact people? And should I contact the Sultan of Kelantan or not? Maybe the Malaysian High Commission in Canberra could give me advice about correct protocol. Called 02 6120 0300 shortly before noon and was waiting for an answer before it occurred to me that that was probably the worst time of the week to call, so put it off until Monday.


More system update stuff
Topic: technology, opinion Link here

As a result of other recent issues, I've lost track of updating eureka. Back to work today, wondering how to migrate as painlessly as possible. And then I thought of x2x, a program to join X servers. I used it for years in Wantadilla, but stopped 12 years ago when I moved to Dereel in the assumption that four displays are enough, and I can do that with a single system.

But I have a fifth monitor, and in fact the display card in eureso can drive two displays—I think. Why not connect the left-most monitor on eureka (eureka:0.0) to eureso as eureso:0.1, and connect eureso:0.1 to eureka:0.1 via x2x. That way I can carry on using the old eureka where I have issues with the new one.

To my surprise, x2x is still installed on both machines. But I've forgotten how to use it. While playing around, discovered that the options on the two versions are different, and in particular the one on eureka won't allow me to go up or down (“north” and “south” in x2x-speak). I'm sure that that wasn't always the case.

And where have all the mice gone? I couldn't find one for eureso. While messing around and trying to run X on eureso without a mouse, discovered that it's not a good idea to run x2x on both sides of a join: it seemed to hang both servers (maybe just x2x looping), and I had to shoot them down before I could continue.

In the end got it working, but the X configuration on eureso is messed up: only one display works, and that at 1280x1024. Should I fix it, or just move on to the planned update without the safety net?


This page contains (roughly) yesterday's and today's entries. I have a horror of reverse chronological documents, so all my diary entries are chronological. This page normally contains the last two days, but if I fall behind it may contain more. You can find older entries in the archive. Note that I often update a diary entry a day or two after I write it.     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. 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.


Greg's home page This month Greg's photos Greg's links

RSS 2.0 Valid XHTML 1.0!