[Dylan Reeve]

You know, based on the material I have received from the US in the past, I assumed they didn’t have any specs. It can be pretty wild.

Not like the BBC, who have rejected things for the most asinine reasons.

[Dylan Reeve]

The method in that tutorial is your best option really.

You can use an Animatte nested in V2 layer to limit the effect to a general area (get the green of the eyes but not the trees by only applying the Chroma key to a smaller area).

Another way you can achieve a similar result is with the Spill Suppressor in a Chroma Key – add a Chroma key to a clip, turn the gain on the primary key to 0. Enable the spill suppressor, select a colour that is OPPOSITE the one you want to keep on the colour wheel, turn gain right up and play with softness settings. This way is less precise.

[Dylan Reeve]

It’s not my expression – my work consisted of adding another zero to the existing expression, and looking at the way the numbers moved.

I love The Daily Show.

[Dylan Reeve]

It’s not my expression – my work consisted of adding another zero to the existing expression, and looking at the way the numbers moved.

I love The Daily Show.

[Dylan Reeve]

I think this is to do with ProTools – I know that older versions were unable to read any gain data from Avid. I’m pretty sure automation gain was a no go, and cuts and dissolves may have worked on some version.

I don’t know about newer versions of ProTools, but I think they read more of the timeline data these days.

There are a few ProTools forums on the web, it might pay to ask there about the limitations of ProTools.

My OMFs are always OMF2 AIFF Embedded. I leave audio data on the export, I don’t know how much of it survives to the audio session.

[Dylan Reeve]

It will depend on what you’re doing I guess, and what the market wants from you. Avid still has a pretty strong reputation, which can be it’s biggest selling point in some cases – people want to edit their program “on an Avid”.

I haven’t used FCP in a version or two, but it is a capable application.

Personally, as an Avid editor (and I have have considerable time in front of FCP) I can cut much faster in Avid than I can in FCP. Although with more flexability in keyboard on work mode in newer versions of FCP, I may be able to streamline it more these days.

In my neck of the woods there isn’t much demand for FCP editors, we are still a very Avid-centric industry.

I am comfortable with Avid to do everything I need, but FCP has been progressing in leaps and bounds since it started.

[Dylan Reeve]

I had a look at the comp.

The problems two-fold — it’s designed clearly for only three digits, although that can be changed, but it requires changing masks on a number of layers.

Second is that it is designed to count down, not up. And the way it works means that when you get to 162.0 you’re actually going to end up with a very weird and unreadable number, because the 1 will be rotated .6 of the way to 2, the 6 will be rotated .2 of the way to 7. But the 2 and the 0 should be okay.

Here are the expressions for the Offset effects of the four number layers (if you add another one):

Digit 1: [50, -40+(-1)*this_comp.layer(“SLIDER”).effect(“Slider Control”).param(“Slider”), this_comp.layer(“SLIDER”).effect(“Slider Control”).param(“Slider”)]

Digit 2: [50, -40+(-10)*this_comp.layer(“SLIDER”).effect(“Slider Control”).param(“Slider”), this_comp.layer(“SLIDER”).effect(“Slider Control”).param(“Slider”)]

Digit 3: [50, -40+(-100)*this_comp.layer(“SLIDER”).effect(“Slider Control”).param(“Slider”), this_comp.layer(“SLIDER”).effect(“Slider Control”).param(“Slider”)]

Digit 4: [50, -40+(-1000)*this_comp.layer(“SLIDER”).effect(“Slider Control”).param(“Slider”), this_comp.layer(“SLIDER”).effect(“Slider Control”).param(“Slider”)]

What happens with a real odometer (at least the one on my motorbike) is that the numbers don’t tick over until the one to their right reaches 9. That shouldn’t be too hard to achieve with these expressions, but I don’t think it would look as nice in the animation.

Maybe there is a nicer way you can cheat it? Or someone else can think of a better way of animating them so you’ll end up with a readable number at the end.

[Dylan Reeve]

I had a look at the comp.

The problems two-fold — it’s designed clearly for only three digits, although that can be changed, but it requires changing masks on a number of layers.

Second is that it is designed to count down, not up. And the way it works means that when you get to 162.0 you’re actually going to end up with a very weird and unreadable number, because the 1 will be rotated .6 of the way to 2, the 6 will be rotated .2 of the way to 7. But the 2 and the 0 should be okay.

Here are the expressions for the Offset effects of the four number layers (if you add another one):

Digit 1: [50, -40+(-1)*this_comp.layer(“SLIDER”).effect(“Slider Control”).param(“Slider”), this_comp.layer(“SLIDER”).effect(“Slider Control”).param(“Slider”)]

Digit 2: [50, -40+(-10)*this_comp.layer(“SLIDER”).effect(“Slider Control”).param(“Slider”), this_comp.layer(“SLIDER”).effect(“Slider Control”).param(“Slider”)]

Digit 3: [50, -40+(-100)*this_comp.layer(“SLIDER”).effect(“Slider Control”).param(“Slider”), this_comp.layer(“SLIDER”).effect(“Slider Control”).param(“Slider”)]

Digit 4: [50, -40+(-1000)*this_comp.layer(“SLIDER”).effect(“Slider Control”).param(“Slider”), this_comp.layer(“SLIDER”).effect(“Slider Control”).param(“Slider”)]

What happens with a real odometer (at least the one on my motorbike) is that the numbers don’t tick over until the one to their right reaches 9. That shouldn’t be too hard to achieve with these expressions, but I don’t think it would look as nice in the animation.

Maybe there is a nicer way you can cheat it? Or someone else can think of a better way of animating them so you’ll end up with a readable number at the end.

[Dylan Reeve]

dpi only applies to print. A 1280x720px picture will be 1280×720 pixels regardless of the dpi. However at 72dpi, it would print at about 18×10 inches, while at 300dpi it would be 4.3×2.6 inches. So ignore the dpi thing.

However, I would personally get the source pictures scanned at as high a resolution (pixel-wise) as possible and practical. You can always scale a photo down, but scaling it up presents problems. If you get them only at 2560×1440 then you’re limiting how far you can zoom into the pictures to only about 200% before you pass the picture’s native resolution.

Of course the aspect ratio will come into play, 1280×720 being 16:9 means you are going to have to reframe your images to fit the format. The more pixels the better I say.

[Dylan Reeve]

I agree with everyone here pretty much. Keeping the movement fairly consistant helps a lot, but you can get away with some differences (a small speed ramp to match them might work for example).

I’ve done practical and post approaches to this in the past. On one short film, for a shot that’s meant to look like to goes through the ceiling to an upper apartment, we just craned the camera up to the ceiling, where there was a small black object to obscure it, and then we craned from behind a similar object on the floor up to simulate the entrance into the upstairs. It was pure black in the wipe, and I used a small push to match the speed.

For others I have extracted elements from footage to use a foregroud wipes (cars, people walking past, a fridge once) and for others we have specifically shot wipe elements against greenscreen.

It can be as simple as a straight cut in a blackframe (ala “Rope”) or as complex as carefully rotoscoped foreground elements and precisely matched camera moves.