Steve Roberts
Forum Replies Created
-
The mask only interpolates its shape between keyframes set by you.
Now in AE, you can use *motion tracking* to track a point (or four), but those points can only be attached to the position of solids or effect points (say, the center of a lens flare). This does not help you, since you want to attach the trackers to mask vertices.
I believe that there may be expressions to work with mask vertices, which may or may not help you. Try searching the COW for “mask vertex” and “expression”.
But I think you want one of the tools by Imagineer systems like Mocha. But those cost money. (Note that the cheaper little Mocha for AE just tracks — it doesn’t actually mask. )
If you don’t have that, I’m afraid you’ll have to rotoscope the talent, which means setting mask shape keyframes by hand. It’s boring work, but you can avoid a lot of wasted time by studying the talent’s motion carefully, and only placing leyframes when the motion of his shape changes direction, say from left-moving to down-moving. Let AE interpolate between those points. And if something (like a head) doesn’t change shape, mask it separately then just move the entire mask into position when necessary.
No, it’s not automatic, but for that you gotta spend the bucks on a dedicated tool.
-
At this point, I recommend making proxies (check the help). But for next time …
Forget 72 dpi.
Forget 300 dpi.
You need to calculate a specific DPI for that specific shot in that specific comp if you want to be precise.The dpi is the last figure in the chain. Here’s another way to do it.
-First is the size of your movie in square pixels, say 720×540.
-Next is the desired enlargement of the image to be scanned. If it fills the video frame, that would be 1x, whereas a big closeup into the photo might be 10x. If bigger than 1x, roughly gauge the enlargement by imagining a rectangle on your photograph enclosing the closeup area. If it’s a quarter of the photo, then the enlargement is 4x.
– Next is the desired size of your photo in ***pixels***. Take the first figure and multiply it by the second: (let’s just use the width) 720×4 = 2880 pixels wide. So your scanned image would have to be 2880 pixels wide to fill the width of your comp.
– Now we look at the paper photo. Let’s say it’s 10 inches wide.
– Now we calculate the scanning DPI, which is the same as “pixels per inch”: too small, and the image won’t hold up to enlargement in AE. Too big, and your computer will slow down. To calculate the DPI, take 2880 pixels divided by 10 inches = 288 pixels per inch, or 288 DPI.
– Put the paper in the scanner, and choose the next DPI up from 288. If it’s 300, so be it.But if you wanted to enlarge it 2x, the DPI would be 144, so you’d make it 150 dpi.
In that case, 72 would have been too small for 2x enlargement, and 300 would have been too large for 2x enlargement.Do you see? Calculate the DPI.
If you want to save time, and you only work in SD, make a list of rectangle-sizes-on-photos (3″ wide, 6″ wide, etc.) and what DPI it would take for them to fill an SD frame. Then you don’t have to calculate all the time.
But for now, use proxies so you don’t have to re-scan. Don’t ask me how big your proxies should be — that’s between you and your comp.
-
Sorry … but on set, on location, did you shoot video with one camera or two cameras?
-
No, I mean figure out how big the image needs to be in square pixels (full frame: 720×540 or zoomed in: maybe 1600×1200) which will be different depending on whether the image is small in frame, full frame, or zoomed in.
In the case of a Ken Burns push-in, you’d zoom into a group shot to show somebody’s head full frame, so that head would have to be 720×540. If the head area is 1″ by 3/4″, then you’d want that to be 720×540, so you’d have to scan at 720dpi (720 pix / 1″) or the next size up.
(image size in pixels) / (image size in inches) = DPI setting
https://forums.creativecow.net/readpost/2/912070
https://forums.creativecow.net/readpost/2/909256
https://forums.creativecow.net/readpost/2/901337
https://forums.creativecow.net/readpost/2/907860 -
Don’t use a standard DPI for all your work, guys! You care about image size in pixels, not DPI! Do the math: paper size (inches) X DPI = digital image size (pixels).
You don’t want it too small (fuzzy on scaleup) or too big (slows you down).
Scanning at the same DPI all the time is like saying “I always drive at 50 miles per hour wherever I go”.
-
Curling and bending like that has to be done with a plugin such as Forge Freeform or Zaxwerks 3D Warps. AE doesn’t bend things in 3D space without a plugin.
-
Excellent description, Kevin. 🙂
-
The layer that gets turned into a sphere has to be really big, because it seems to shrink when it gets sphereized, and you’re moving in on it. How big? Try double or 4 times what it was.
-
This used to be easy before AE has layer groups or whatever — you would import it in layers like a PSD. But noooo … now you have to take all those sub-layers or whatever and manually (in Illustrator) move them into separate Layers. Like real Layers, not those sub-thingies.
The “release to layers” command might work, but I’ve had trouble with it. Maybe searching the COW for “layers” and “Illustrator” might work, as this has been discussed before.
Bitter? Moi? 🙂
-
Yeah, I thought somebody would catch me on that. If I recall, the 480 has more non-active lines, rather than the 486 having less non-active lines. 😉
Want pain?
https://lurkertech.com/lg/big/howbig.html
https://lipas.uwasa.fi/~f76998/video/conversion/Look here under “active picture”:
https://www.quantel.com/resource.nsf/Files/Quantel_Digital_Factbook/$FILE/Quantel_Digital_Factbook.pdf… all this leading us to re-assert our long-held belief that a line is a pixel. If you need confirmation or denial (not me!) check out the Broadcast Video COW.
Now … the confusion may arise from past discussions that a font’s point size is not a line/pixel size. Set ups guides in the comp and work within those. Measure twice, typeset once. 🙂