HDR & ACES: a 1-day Technical Seminar

with Charles Poynton

hosted at Mr X Gotham, Manhattan

On Thu Aug 11, in collaboration with Mr X Gotham (Manhattan), I’ll present a one-day technical seminar on HDR and wide colour gamut for d-cinema, HD/UHD/4K/8K.

ACES is important in high-end production, and is becoming increasingly useful in CGI/VFX and DI. The ACES block diagram serves as the theoretical underpinning for CGI/VFX/DI pipelines, and ACES is HDR-capable. So, we’ll start with a detailed review of ACES. After approval, at mastering, HDR material needs to be coded in either PQ or HLG code. We’ll describe these encodings.

BT.709 is best understood as an acquisition standard. For HD/UHD standard dynamic range, BT.1886 is the important display standard. We’ll give details, and explain why “709” is not usually the correct term. HDR in practice comes with wide colour gamut (WCG), often DCI P3 in a BT.2020 “container.” We’ll give details of the colour transforms – both the colorimetric transforms (which are conceptually simple, but have math) and the gamut mapping transforms (which ideally involve a colorist).

For distribution and consumer display of HDR/WCG, metadata is necessary. We’ll describe today’s HDR/WCG metadata standards, and outline what is likely to be required for dynamic metadata in the near future.

HDR presents challenges for test signals, instrumentation, and calibration. High (peak) luminance is an issue. Emergent displays have different spectral characteristics than usual for studio reference, potentially leading to metamerism. HDR displays typically have average power (“APL”) limitations. Finally, consumer displays may incorporate highly nonlinear signal processing. We’ll outline recent developments.

The seminar will take place at Mr X Gotham, 10:00 to 17:00. Registration fee is USD 375.

Registration details are forthcoming. Meanwhile, if you’re interested in attending, contact me by phone or email.

Practical Aspects of HDR (in Oxfordshire, July)

On Tue July 12, at Heythrop Park in Oxfordshire (90 min or so from London), John Watkinson presented a 5-hour seminar, Practical Aspects of HDR: Implementation and Deployment, as part of the SMPTE/HPA UK Tech Retreat. John covered the material that I presented at HPA Tech Retreat in Palm Springs in Feb., with an update concerning emerging developments.

I had planned to be there; however, a very minor medical incident – now completely treated – prevented me from flying. In the next week or two, I’ll be offering a webinar to the attendees. I also plan to offer the material as a (paid) webinar. Stay tuned.

Emergent Technology in D-cinema and UHD/4K

High Dynamic Range, Wide Colour Gamut, and ACES 1.0

A Seminar with Charles Poynton

hosted by SKYLAB HQ/Screen Dragon, Vancouver, BC

Tue Feb 23, 10:00 – 17:00

Three important emerging technologies touch acquisition, post production/DI (especially colour grading), studio display, mastering, and distribution for cinema and UHD/4K: high dynamic range (HDR), wide colour gamut (WCG), and ACES 1.0. We’ll address these topics by outlining the theory of psychophysics and colour appearance, then applying that theory to high-quality cinema and video from production through consumer display. We’ll outline the theory of “picture rendering” that adapts image data to varying scene, display, and viewing characteristics. We’ll describe the commercially important colour spaces in cinema and film, and discuss the transforms among them. We’ll describe the need for colorimetric transforms (for example, to place BT.709 or DCI P3 material into a BT.2020 container), and we’ll discuss colour gamut mapping that is necessary to adapt wide-gamut material to a conventional-gamut display. We’ll discuss system-level issues: OECF/OETF, EOCF/EOTF, their inverses, and OOTF (also known as rendering transform, or RRT).

Building on these concepts, we’ll then discuss specifics of our three important new technologies: HDR, WCG, and ACES. We’ll explain the ACES block diagram and its various transforms. We’ll outline open standard and proprietary proposals for HDR, and discuss the SMPTE ST 2084 perceptual quantizer (PQ) and the BBC/NHK hybrid log/gamma (HLG) transfer functions that have been proposed. We’ll discuss several emergent coilourspaces including ACES AP0, ACES AP1, and BT.2020. We’ll discuss accompanying image data with tone- and colour-mapping metadata that allows a consumer receiver to adapt the image data to diverse luminance and colour capabilities.

The fields of digital cinema and UHD are rapidly evolving in technology, business, and standards arenas. Expect discussion of last-minute developments.

The seminar will be suitable for cinematographers, colorists, digital intermediate and post-production technical staff, VFX craftspeople and programmers, mastering/distribution engineers, and consumer electronics (CE) television experts.

Registration: CAD 450.

Contact Charles Poynton, charles@poynton.ca, +1 416 535 7187

“Windows Technical Support” scam

At least once a week I get a call from a caller claiming to be “Windows Technical Support.” It’s a scam. The routine is to convince you that your computer is “infected” by dozens of virii. You give a “technician” permission to take control of your computer (!); after a payment of from $100 to $360, the “technician” repairs the alleged damage. Actually, they have no idea whether anything is wrong with your computer; the whole “repair” scenario is faked-up. Their goal is simply to extract cash.

The latest call to me was from +1 203 793 8906. Early in the call the caller tries to substantiate that he’s associated with “Windows” because he knows my unique “CLSID/CLS ID” number, 888DCA60-FC0A-11CF-8F0F-00C04FD7D062. You can read about how CLSID not unique.

You can read anecdotes from people who played along for a while: Phone scammers call the wrong guy, get mad and trash PC, and A tech support scammer dials Ars Technica. These guys are dangerous: Apparently if they get as far as convincing you to grant remote access to your machine, but then you fail to pay, they’ll remotely delete your user files and trash your system. Uggh.

The scammers attempt to identify themselves with “Windows.” Microsoft has published a few web pages about the scam, Avoid tech support phone scams and 5 things you need to know about tech support scams. Computerworld has an article, “Aggressive, persistent Windows tech support scammers continue to stalk consumers.” A detailed description of the scam is available: Hanging on the telephone (PDF, 1.5 MiB).

I describe this because a friend of mine (who should have known better) was taken in by the scam. It is clearly generating enough cash that their operation has been sustained for several years. It must be the case the the front-line worker (my caller) knows that the operation is a scam; it is a sad sign of the times that apparently you can convince hundred of workers to lie, cheat, and steal to make (presumably) a pittance as a wage. I admit curiousity about whether you could construct an operation like this maintaining a veil where the front-line callers don’t even know that what they are doing is illegitimate. My tentative conclusion in this case is that the front-line workers here know that they are lying.

Windows 7 System Update, 80246002 error

I’ve been using Windows 7 (SP1, 64-bit). I use it rarely, and keep it very very close to factory-standard state. I use it to run Adobe FrameMaker and nothing else. I frequently run Windows Update.

But yesterday, Windows Update failed with an error code 80246002 (which I gather is in hexadecimal). None of the remediation options offered in the Windows Update succeeded; no further information was available. The Windows Update troubleshooter was unsuccessful in clearing the problem.

There are several relevant pages among the Microsoft Developer Community pages (for example, this one).

I did two things. I reconfigured DNS lookups to and (Google public DNS resolver, see also Wikipedia). Also, I downloaded and ran the appropriate System Update Readiness Tool (KB947821) appropriate for my system. One or both of these acts resolved the issue on my machine.

The incident should point out to Microsoft that system update machinery should be very very carefully checked. If Microsoft breaks the update machinery, how do they execute the update to fix it?

Phishing for Bell Canada

During the past four weeks or so, about once a week I have received a voice landline call where the caller begins, “Hi, I’m calling from Bell Canada Promotion Department.” The caller ID does not indicate BELL CANADA, that’s your first clue that something is amiss. The second hint (sad to say) is that the caller has an East Indian accent, and the high level of ambient/background noise suggests a low-Q call centre. Your third hint that something is amiss is that when you request a callback number the caller explains that callback isn’t possible.

Given the chance to continue, the caller explains that the call is regarding home phone, internet and TV services.

After collecting as much information as I could on these four calls, and contacting Bell several times, I have concluded that this is an identify theft phishing scam aimed at obtaining personal credentials. The call starts out sounding like it’s an offer to upgrade your Bell service, and provides details such as “home phone, internet and TV services all for just 92 dollars.” The phishing starts when you are asked to provide details to “verify your account.” They may bail out when they have the details they want – or, more likely and more insidiously, the call-centre employee may complete the “transaction.” With the latter M.O., the call centre employee has no reason to doubt that the operation is legitimate, sort of like the fake Apple stores in China.

Bell says that they are aware the scam, and also of similar instances where the caller identifies himself as being from Rogers or Videotron.

My latest call indicated 1 816 311 0245 (MO), but it is likel;y that the caller IDs used by the scammers are fake. Why Bell allows faked-up caller ID numbers (e.g., 0000000000, 0123456789) to enter their network is beyond me.

Bell’s own Fraud Department (1 855 558 2355) is fully aware of these attempts, but explained to me at great length and in great detail why it’s not their job to do anything about it. The Fraud Department exists, they tell me, to protect Bell from fraudsters (and not, as I had hoped, to also assist in protecting their customers from fraudsters). The Fraud Department seems completely unconcerned about the potential (huge, in my opinion) to damage of the Bell brand. [This is despite Bell’s own How to protect yourself from telecom fraud page saying, “If the phishing scam involves the false representation of Bell, email the situation to abuse@bell.ca”.]

Of course, Bell has already rendered significant damage to its own brand by scraping customer traffic from the network and selling mobile subscriber browsing histories and other personal information to advertisers. But that’s another story.


It surprises me that Adobe Illustrator (mine is 15.1, a.k.a. CS5.1) doesn’t respect figure space. A figure space (which I would prefer to call digit space or numeral space) has the same width as the digit zero, but (obviously!) makes no mark on the page. Most typefaces (“fonts”) have digits that set to the same width, and in those typefaces the figure space is very useful to align numbers. 

FIGURE SPACE is defined in Unicode as U+2007. In this notation, 2007 is the hexadecimal form of the Unicode code point. Inserting this character into an Illustrator file produces a space, but not of the same width as a digit. 

You can work around the issue in Illustrator by using the digit zero, but using stroke None and fill None so as to make no mark. 

By the way, you can enter Unicode characters directly in most Mac applications by choosing the Unicode Hex Input source. Access System Preferences → Keyboard, choose the Keyboard tab, and enable Show Keyboard & Character Viewers in menu bar. Then access Input Sources (also available from the Language & Text preference panel), scroll to the bottom, and enable Unicode Hex Input. 

Upon closing the preferences panel, you’ll see a flag (Canadian, in my case) at the right-hand side of the menu bar. That’s a pull-down menu, one of whose choices is Unicode Hex Input. Once you choose that input source, you can enter a Unicode character by holding Option down continuously while you key four hex characters (0–9 and a–f or A–F). Obviously, normal entry of text is impaired when this input source is selected – for example, é, e-acute, can’t be entered directly. I recommend that you choose your normal source as soon immediately after entering Unicode. 

 By the way, if you see the term font designer in a document, in my view that document cannot be definitive on typesetting. The person to whom that document refers is known in the profession as a typeface designer. 

Bell Sympatico pop/smtp

If all of a sudden (in my case after 10 years of service) you’re having trouble with Bell Sympatico email service, if you find that your credentials are not recognized – even immediately after resetting your password – I’m here to tell you that Bell has turned off pop email access by default.

Yes, crazy, I know, but they’re almost a monopoly, they get to do stupid things.

Change the setting in – get this – webmail. Log-on to webmail, click on the “gear,” under Options, Managing your account, Connect devices and apps with POP, Enable.

Nuts. Crazy.

Xerox Phaser 6500 fan replacement

As the title suggests, my Phaser 6500 colour laser printer – which I’ve been pretty happy with – had its main cooling fan fail, a few months after the warranty expired. The failure was indicated by a fairly explicit message on the built-in display.

So I tore the printer apart and fixed it. Should you find yourself in a similar circumstance, you could give it a try.

I have to give three warnings: 1. Read all of these steps before executing any. 2. Should a step here prove too difficult to execute, reverse everything, reassemble the printer, and seek qualified help. 3. If upon finishing you have any screws or other parts left over apart from the failed fan, then actually you are not finished.

The fan itself bears Xerox part number 127E86270. You may be tempted to read the manufacturer’s data from the fan itself (Nidec D08K 24TS1), and replace accordingly, but you would be disappointed: The Xerox fan bears that Nidec part number, however, Xerox has used a custom version of that part having three wires [black, brown, yellow] instead of the usual two [probably black and red]. I guess that the third wire enables sensing of proper fan operation (or, if you’re a pessimist, enables detection of fan rotation failure). I’m confident that the two-wire version won’t work without some engineering (if it were to work at all). So, procure the Xerox part. I found one at the curiously-named ItemInc.

If you order from ItemInc, be explicit about shipping. For me, in Canada, they shipped it out FedEx Ground (fine) but lacking the commercial invoice necessary for Canada Customs (not fine). FedEx discovered this and sent the shipment back to ItemInc. ItemInc augmented the shipment with the required invoice, but then inexplicably re-shipped it UPS (not fine – in fact, bad). UPS charges CAD 20 to file customs paperwork, collected COD (VISA or cheque please, no cash) for a service that (as I understand it) FedEx provides at no charge as part of their service. There are many other reasons to loathe UPS.

Now you have to disassemble the printer. Refer to your user manual [PDF] to make sense of this paragraph. Release and lower the front cover (as if clearing a paper jam), and drop the transfer belt unit (ditto). Then, remove the fuser, located just inside the unit at the exit tray. Its location is described in the user manual, as are the two green tabs that release roller pressure for paper-jam clearing. However, removing the fuser is not described in the user manual.

You need to find the grey tab that releases the right end of the fuser. The fuser is effectively hinged at its left end, so swing it toward you keeping it horizontal (with the right end in the “hinge”), then at 30° or so, lift it out.

When you lowered the front cover, you exposed the front of the printer mechanism. The bottom remained on the table. There are four other faces of the cube. Your task is now to remove the plastic covers on those four faces, that is, the top, the left and right sides (as viewed from the front), and the back. The top cover is also known as the “exit tray.”

Having removed the fuser, you’ve exposed two Philips screws that attach the front end of the exit tray assembly to the body of the printer. Remove those screws. On the back edge of the exit tray is sort of a hinge, invisible to you at the moment. Lift the front of the exit tray, rotating around the imagined hinge at the back edge. At 30° or so, you’ll be able to lift the exit tray clear. You can now see “hinge” mechanism; it’s simply moulded into the plastic parts (at zero marginal cost). Removing the exit tray has revealed the fan that needs to be replaced, but it will be obvious that more shrouding needs to be removed, and there are more screws to do so. All of the screws tap into plastic, not metal: Remove them carefully. When reinstalling, take care not to overtighten and strip threads; it’s not clear to me how you could recover from that. All of the screws that you will remove are identical.

When you lowered the front panel, in addition to exposing the front of the printer engine you exposed two Philips screws that attach the front edge of the left-hand side cover to the main unit. Remove those two screws. There is a moulded plastic hinge on the back edge of that part; rotate the front edge of the panel outward, and at 30° or so, you’ll be able to lift the part clear. (Is this hinge business beginning to sound familiar?)

When you lowered the front panel, you also exposed two Philips screws attaching the front edge of the right-hand side panel. Open the toner access cover to reveal several more screws; remove all of the screws. (Sorry, I forgot to count.) Then rotate and remove the right-hand side panel (along with the toner access cover). The “hinge” for the side panel is, of course, at the back.

Now, access the back of the printer. At the bottom of the back panel are two screws; remove them. There is no hinge here – just pull the back cover straight out toward the back. Now the machinery is bare – and now you can access the fan.

You’ll notice three small wires forming a bundle that leads from the fan to the power supply board. Carefully (!) pop out the cable connector at the power supply end. There’s a very small plastic latch involved; gently hold it in the release position while you gently tug all three wires of the cable straight upwards. Unthread the wiring from the (black plastic) fan duct so that the wires emerge directly from the fan. The fan duct has four black plastic spring tabs that hold the fan itself inside the duct – splay those four tabs outward from the axis of the fan, and pull the fan straight backwards to remove it from the duct. Remember the fan’s orientation with respect to the emergence of the wires. (The fan sucks air into the printer; it does not exhaust air.)

There. You’re halfway done.

Snap-in the replacement fan. Dress its wires through the wire channel, and insert its connector onto the power supply board.

Then, perform the inverse of all of the cover/shroud instructions in opposite order. Done!

If you disassembled everything before receiving the replacement fan (as I did), you should be aware that the imaging unit can be damaged by light. If you leave the unit for several days with the covers off (as I did), I recommend that you cover it up with an opaque green polyester bag or some such.

I admit that I did have access to a maintenance manual while I executed this procedure, but you shouldn’t need it. (In fact most of the service manual information that I used was for the closely related model 6125/6130; you could use search terms like Phaser 6125 30 Service Manual Part1 filetype:pdf.)

Contacts – Linkedin/Mac Address Book

I wouldn’t want Linkedin to think that I’m abandoning the “platform,” on the other hand, with every dot-dot release they’re making it more difficult to stay.

So: How to get Connections out as contacts? On the web, access Connections, look under the gear icon at the top right (Settings), then at the top right of that page is an item Export LinkedIn [sic] Connections.

You’re given five choices for format: three choices for csv format, and two for vcf format.

In my experiments (who knows what might happen under a full moon, or on Tuesdays) all of these choices export eight-bit text files in MacRoman encoding. That is, e-acute and a handful of other accented characters work (if you read the file on a Mac), but trouble awaits for Greek or Cyrillic or lots of other characters. Any character unavailable in MacRoman is silently exported as a questionmark (?). Should a Linkedin member place a question mark in any field of his/her contact data, that would be exported indistinguishably from a character outside the 7-bit set; to a programmer, this means that any “?” must be treated as missing or unknown data. If you read the exported file in Windows, I don’t know, but I would expect you’d need to re-encode to a Windows-friendly encoding, or better, Unicode.

It seems that the Outlook Express and Microsoft Outlook format choices differ only in the sort order of the connections; I could find no other difference.

The Yahoo variant has all the essential fields, but a few dozen obscure fields are missing compared to the Outlook formats.

The two vcf formats both have CR/LF lineends, but there’s a spurious LF at the end of the file. The contents are apparently identical except for sort-order of the connections.

Linkedin’s code to produce the csv variant is faulty: The Linkedin programmers apparently forgot that backslash is used to “escape” certain codes. The programming error is that they do not check to see if a character in a Linkedin field contains a backslash. If a Linkedin user puts a backslash in a field, then the exported csv file almost certainly breaks.

If you want to retain accented, Cyrillic, and Greek characters, then you might try running the Linkedin Android app, syncing with Contacts, and then exporting Contacts to an SD card. I assume you’ll want to do this operation without contaminating your main (non-Linkedin) contacts; for me this was easy because I use an iPhone as my main device, and I have an auxiliary Android device whose contacts database I could devote to the Linkedin export operation. I have found bugs of various sorts; for example, the contact sync on the ANdroid side fails for various reasons at various points. Of my 2500 or so connections at the moment, 1800 or so are transferred, and the remainder is in limbo. No photos transfer at the moment. Don’t expect Linkedin to fix any of this.

Such an export operation exports a VCF file in UTF-8 (with CRLF lineends); accented, Cyrillic, and Greek characters are retained. However, a field having any character outside the 7-bit range is coded as quoted-printable (that is, in hex code) of UTF-8 encoding of the [Unicode] character. In fact, in my test, in a field that contains a character outside 7-bit ASCII, every character is coded in q-p hex, even those that don’t have to be (the ranges 0x21-0x3c and 0x3f-0x7e would be more sensibly and more conventionally represented literally as “!”-“<” and “>”-“~”).

When exporting connections to storage, an image may be coded as part of a connection’s entry; images in my test contain BASE64-encoded JPEG data, sized 80×80 pixels.

Phone numbers are not represented in canonical form; they are apparently retained in the form entered by each of your connections. So, there may or may not be a country code, with or without a + sign; there may or may not be a “(0)” or other designation of a dialing prefix; and there may or may not be hyphen, space, or other separators between country code, exchange, and local number fields. Chaos.

Export now, while you have the chance! There’s no telling when Linkedin might withdraw this feature.

[p.s, Should you export vCards from Mac AddressBook (or presumably, from Contacts), these seem to be in UTF-8 without encoding declaration, despite the vCard spec saying that any character encoding other than ASCII must be declared field-by-field. Lineends (for example, in notes) are represented as backslash n. Curiously, comma and semicolon are also backslash-escaped.]