Tuesday 6 June, 2017
I got up at 07:30 and had a shower, then a banana and an apple for breakfast before leaving for the meeting. The day was sunny again, but hadn’t warmed up yet as I walked to the Intel office, where I arrived a bit early and was shown in to the meeting room with a few other early arrivers.
The meetings today got stuck into the technical ad hoc working groups, covering specifics for each of the standards we currently have under draft. I made comments on several of them, when I though something particular should be considered.
One in particular was the texture measurement with the dead leaves pattern test chart. The experiments had produced a puzzling result, with the spatial frequency response curves of test images looking somewhat reasonable when averaged with a ring-mean to average all orientations in the Fourier transform for each sample frequency, using the real part of the transform, but when the modulus was averaged it produced an obviously incorrect response as the frequencies grew higher, with the SFR rising anomalously and very obviously. Rob presented an analysis Imatest had done on this, showing that the problem was caused by nonlinear noise reduction, which had different effects aligned in the direction of the pixel axes and at 45° to them. The result was that looking at the Fourier values around the azimuthal direction revealed a sine wave artefact, with four periods, aligned to the cardinal directions. This caused the response to go negative for about half the cycle at high frequencies, so taking the real part averaged essentially zero with a small signal, while the modulus ended up averaging a series of larger and larger values as the frequency rose and the sine wave increased in amplitude.
Partway through the presentation I realised exactly what the problem was. But I let Rob finish his presentation rather than interrupt. When he was done, faces around the room looked glum and puzzled. So I took the chance to speak up. The problem was that they should have been taking the average of the complex Fourier values, and then taking the modulus, not the other way around. And then everyone was saying, “Oh yeah… of course, why didn’t we think of that?” Rob said he could redo the analysis that way and circulate the results.
 Salad for lunch at the Intel cafeteria |
We finished the morning session a bit late, giving us exactly an hour for lunch after Toru pushed the post-prandial restart back to compensate. Most of us just went to the Intel cafeteria today. This was huge, having a floor area something like three times the space of my entire company’s office back in Sydney. There were several hot food bars, with Indian, Mexican, burgers, sandwiches, and probably some others I didn’t look at, plus make-your-own sandwich and salad bars. Trying to keep healthy, I piled a bowl with salad ingredients, including chick peas, beans, mushrooms, chillis, what I thought were pumpkin chunks but might have been something else, and a few other things. I sat with Rob and we chatted about random stuff.
Back in the technical meetings, we continued discussing various standards drafts until about 16:30 or so. After this, beginning at 17:00, was an administrative working group meeting, which is a special thing held at plenary meetings to go through administrative issues for the whole technical committee. As the head of delegation from Australia, I had to attend this small meeting that only involved ten or so people. The major item was revision of the TC42 strategic business plan, and I ended up taking an action item to draft a paragraph clarifying the scope and exclusions of the committee, and circulating it before the second AWG meeting on Thursday.
With the AWG meeting done by 18:30, I headed with Jonathan over to Pedro’s restaurant, where we’d arranged to meet Ken and Margaret for dinner. This was just a ten minute walk from our meeting, and turned out to be a large hacienda style modern place, with a large car park around it. It seemed popular, but had lots of tables, and we got one by a window. After lots of complimentary chips and some very spicy salsa, Jonathan and I ordered a pitcher of margaritas while we waited for the others to show up. They both approved, and then we ordered dinner.
 Grilled ahi tuna on “tostadita”, at Pedro’s |
I selected a trio of tacos, choosing the chicken, pork carnitas, and pescado (fish) fillings. Margaret had the tacos too but chose two different vegetable fillings, one being cactus and one being some other plant neither of us had heard of, while Ken and Jonathan both had the chimichanga (essentially a deep fried burrito). We also ordered an appetiser of grilled ahi tuna with garnishes on blue corn “tostaditas”, which I said was “mini tostadas”, but turned out to be essentially just corn chips.
 Trio of tacos at Pedro’s |
The tacos were good, in house made soft corn tortillas. The fish in particular was nice. I could potentially have tried a dessert, but none of the others wanted any, so we simply settled up the bill and departed. Ken and I walked back to Jonathan’s car at Intel where I’d left my laptop bag, and then Jonathan gave us a lift back to the hotel.
I talked with M. a bit via FaceTime audio, since the WiFi kept killing any attempt to do video. Then read a bit and went to sleep.