Cooking and pano websiting

This morning I went out in the garden after breakfast to try and take some insect photos, but the only insect I could find was a fly that I'm pretty sure I already have a few photos of. While I was in the kitchen I saw a sparrow pulling up some moss and then flying away with it, and it also looked like the Robin might be making a new nest in the nestbox in our garden.

I did some vacuuming, then checked my email. After this I checked some photos to make sure they had been tagged correctly. I didn't find any missing the tags I thought might not have been added properly, but did find some that included the full tags and also a partial tag, which I then removed.

I then started work on the CSS for the buttons I made for my pano website yesterday.

After lunch I finished off the CSS needed to make the buttons to display properly, then did some testing on a 640px cube face QTVR versus cube faces in FPP.

For the tests I used IE8 with Fiddler 2 set to simulate modem speed, and compared loading speed of a QTVR produced by Pano2QTVR with cube faces produced by PTGUI. For both I used 640px cube faces with 70% jpeg compression. The QTVR was 422KB, while the PTGUI cube faces were 323KB, so the QTVR was a bit larger. All times in the below table are in seconds.

Run	Time from request until pano starts loading	time from pano starts loading until pano loaded	total
mov, loaderStreamed=1, no cache
1	11	63	74
2	10	64	74
3	10	63	73
mov, loaderStreamed=1, cache
1	0.25	0.25	0.5
2	0.25	0.25	0.5
3	0.25	0.25	0.5
cube faces, no cache
1	9	20	29
2	10	30	40
3	11	30	41
mov, loaderStreamed=0, no cache
1	10	63	73
2	9	63	72
3	9	64	73

The results are pretty similar to what I found when comparing large size JPEG cube faces versus QTVR - using a QTVR is much slower than using cube faces, and using streaming or not on a QTVR makes little difference to load speed.

For this test, since the file sizes are pretty small, I used Fiddler 2 to limit my download speed so I could get a better idea of how the different methods compare on a slow connection (which is what the smaller files are for, along with mobile users). With the download slowed down quite a bit, I decided to also time how long it took from requesting the pano until the loading text appeared.

When using a .mov (QTVR), FPP requires a mov decoder plugin to be used, so I wondered if this would make any difference to the speed from when the pano is requested until it starts actually loading. The movDecoder.swf plugin is only 8.12KB, but it obviously does involve another HTTP request to download it. But as you can see, there's not really any difference in the time until the pano actually starts loading.

One other thing I noticed is that the QTVR produced by Pano2QTVR appears greatly sharpened compared to the PTGUI jpeg cube faces. (Or should that be that the PTGUI cube faces are greatly blurred compared to the QTVR produced by Pano2QTVR)? I put an example comparing the two here.

Unfortunately, Pano2QTVR only allows cube face exporting at maximum jpeg quality, so I couldn't really compare Pano2QTVR cube faces with PTGUI cube faces. I did try doing a 'Save for Web' in Photoshop on a Pano2QTVR produced cube face, and increasing the blur setting, but with enough blur to make it the same quality as the PTGUI produced cube face, but the Pano2QTVR cube face resultant file size was still much greater than the same PTGUI cubeface.

The strong sharpening / lack of blur in the Pano2QTVR produced image also produces some artifacts not present in the PTGUI produced cubefaces. e.g. in my example above, if you look at the bottom left of the large building, there is a grate or window that displays what I guess is a form of very bad moiré in the PanoQTVR image, but looks okay in the PTGUI image.

In the evening I did some cooking with L - we made some Danish butter cookies (with hard marg), and a swiss roll. The Danish butter cookies don't really taste like Danish butter cookies at all, more like Christmas biscuits. The swiss roll is quite nice, though the sponge did split quite a bit. The recipie for the Danish butter cookies seemed to make loads, so we had to do a few batches in the oven, and ended up doing cooking most of the evening.

When we'd finally finished cooking all the biscuits and I'd finished washing up, me, Mauser, and L watched the latest episode of Lost, which explained where the Man in Black and Jacob came from.

The weather started off sunny, but was overcast by 9am. In the afternoon the clouds cleared up a bit, and it was a mixture of clouds and sun for the rest of the day.

Food
Breakfast: Bowl of Choco Moons Cereal; Cup o' Tea.
Lunch: 2x Cheese on Toasts; Salad; Satsuma; Rocky; Cup o' Tea.
Dinner: 2x Posh Pork & Herb Sausages; Jacket Potato; Baked Beans; 2x Tinned Plum Tomatoes; Grated Mature Cheddar Cheese. Bit of fresh baked bread; Coffee.
Evening Snack: A couple of home-made non-butter Danish biscuits; Coffee.
Supper: Slice of home-made Swiss Roll; Cup o' Tea.