Darby Edelen
Forum Replies Created
-
I have to admit that I haven’t looked at this in depth yet, maybe I’ll get a chance to soon (I don’t work with green screen footage very much). However, given your ‘Math Node’ comment I can recommend that you look at some of the Channels effects. It might be a bit convoluted to get where you want to be, but the tools might be there.
I suggest looking at the Channel>Arithmetic effect. This gives you a range of operations to perform on the Red, Green and Blue channels using arbitrary limiting numbers. Also look at the Channel>Compound Arithmetic effect, this gives the same range of operations to perform on the RGB channels, but bases the operation on a 2nd source layer (add the values of the RGB channels from one to another, subtract, multiply, bitwise logical operators). You could also split the image into 2 pre-comps and use Shift Channels to create one containing the Green channel only and the other containing Red and Blue, or any other combination (if that would be useful).
[_Peter_] “If I want to keep the pixel values of blue and red unless it would be higher than green … how to do that in AE ?”
I’m not familiar enough with the actual formula used in despilling a green screen shot, although I believe I have some books at home I could look to for reference. If I’m able to figure out the operations generally used to despill, I might be able to come up with a more precise prescription for After Effects.
Darby Edelen
DVD Menu Artist
Left Coast Digital
Aptos, CA -
[Steve Roberts] “That’s one of those “darn useful” effects.”
Ha! I was actually going to post in reply to your ‘unsung’ effects post about the Grow Bounds effect =)
Darby Edelen
DVD Menu Artist
Left Coast Digital
Aptos, CA -
It’s hard to know what to suggest without knowing the details of a project. I suppose that’s my suggestion =) Always push yourself to come up with the best solution for the given source material, the source material will often determine how you approach your work. Also, use reference footage (even still images) constantly. If your composite doesn’t look right, ask yourself why the reference image does and yours doesn’t.
For creating mattes you can often create a hi-con matte by playing with the levels effect. If you’re willing to give up a little bit of control you can use the Keying>Extract effect which uses a histogram similar to the Levels effect to key areas of a luma range that you define.
Darby Edelen
DVD Menu Artist
Left Coast Digital
Aptos, CA -
If you didn’t have success with something like Xmult or Unmult (which will probably need to be applied to an adjustment layer above the text layer to work) then I can suggest a workaround. Duplicate the text layer that is receiving the light, turn off accepts lights on this layer and place it below your original layer, then set this layer to use the original layer as a Luma Track Matte.
This will only work well if the layer receiving the light is white, so if it’s not you’ll need to make it white. Also, if the text is animated I’d recommend removing those keyframes from the duplicated layer, going to the beginning of the composition and parenting the duplicate to the original, then make any adjustments to your keyframes on the original.
Darby Edelen
DVD Menu Artist
Left Coast Digital
Aptos, CA -
If they are separate layers then you simply apply a blur to the lower layer, but seeing as you’re asking this question I’ll assume that they are not separate layers =)
The best option would probably be separating them into 2 different layers, with an unaffected foreground on top and a blurred background behind it. Another option is creating a matte for the person you don’t want blurred and using Blur>Compound Blur, but if you’re creating a matte anyway you might as well just separate the foreground into a different layer =)
Darby Edelen
DVD Menu Artist
Left Coast Digital
Aptos, CA -
The way I initially envisioned doing this (maybe because I’m just a little sadistic) is by actually having the actor hit in the head with a bat… a foam covered bat to be clear =) This would give him/her something concrete to react to, then shoot the same scene (locked down camera would help) with a stand in for the head (a pole? manequin? something that the bat can ‘react’ to) and a real bat. Compositing the two should be pretty easy if you’ve managed to keep the camera still and maintain lighting conditions.
If this actually sounds appealing (I just like hitting things), you’ll want to consider the possibility of the foam bat going in front of the head/face of the actor (try to avoid it, if it happens make sure you can cover it up with the real bat, maybe use a very small foam bat). After that the compositing should be a simple rotoscoping job.
Your mileage may vary with this technique depending on the way you want to shoot the scene. Just another possibility =)
I also think sound effects would be essential on selling the shot, and I would suggest some ‘crushing’ effects. The sadistic side of me shining through again i guess (;
Darby Edelen
DVD Menu Artist
Left Coast Digital
Aptos, CA -
Yes, I would recommend also looking into creating several instances of the individual to be rotated in 3D space, rotated at even intervals to one another (2 instances at perpendicular to each other for example). Then look at a box blur for hiding some of the hard edges, and a compound blur perhaps to blur the higher (and usually wider) parts of the person a bit more.
Also, I recommend pre-comping the footage as 3D layers fading from one to another and rotating instances of this pre-comp in your final composition (fade up the duplicate instances after the rotation has started). This gives you an added benefit, if you duplicate the pre-composition and turn the original layers to guide layers (they won’t be rendered) you can create whirly swooshes that follow the layers rotating (turn on collapse transformations for the pre-comps in the final comp).
If this all sounds a bit confusing I’d be willing to send you a sample .aep =)
Darby Edelen
DVD Menu Artist
Left Coast Digital
Aptos, CA -
L = thisComp.layer("planet");This first line just assigns your layer to the variable ‘L’ so it can be easily accessed in future declarations & functions.
L.toWorld(L.anchorPoint);This is where the magic happens, this transforms the ‘L’ layer’s anchor point position from ‘layer space’ (i.e. where it resides on the layer) to ‘world space’ (i.e. where it resides in the 3D space of the composition). This point in world space is the value that’s returned by the expression.
I definitely recommend you look at other layer transformations when you’re feeling more comfortable with expressions, they help a ton with integrating 2D effects into ‘3D’ space.
Darby Edelen
DVD Menu Artist
Left Coast Digital
Aptos, CA -
[Paul Trunkfield] “Also, is using the wiggle setting the best for simulating jerky movement or shall i try and rock their chairs whilst filming.”
Definitely rock the chairs. However, I honestly don’t think you can end up with anything convincing using the technique you’ve outlined (and I hate to be a pessimist). The actors will simply not look as though they’re riding on a rollercoaster. When you ride a rollercoaster your body is thrown to the side as you make a turn, the sense of gravity decreases when you are going into a quick drop and increases as you’re coming out of it, the ‘wind’ direction changes (as it’s really you moving through air, not air moving past you). Having your actors try to mimic these movements would be ill advised, when you’re on a roller coaster quickly turning to the right your body’s momentum pulls you left but you tighten your muscles and lean right, the actors would have to lean left and right at the same time!
I think your best chance using this technique (see, i’m not all criticism) would be to limit yourself to very brief shot-counter shot sets (instead of the head on approach you suggested). This way you can more readily match the actors’ actions (and wind direction) to what should be happening in the shot (i.e. turning left, going up, going down) without having to worry too much about timing it all properly for the background footage.
Darby Edelen
DVD Menu Artist
Left Coast Digital
Aptos, CA -
Unless the camera was focused on the plane where the odometer resides, and you were either zoomed in or close enough to the odometer to make it take up more than 1/1000th of the picture, and you were using a pretty nice camera, this sounds pretty nearly impossible.
I can tell you, however, that a series of small transformations will have less of a disastrous effect than one large one in Photoshop. Instead of scaling something up by 100% try scaling it up by 10% 7 or 8 times (which will not be exactly 200% larger, but close) and you should save some detail. However, depending on the clarity of your source picture (as previously mentioned) this may not help you much.
Darby Edelen
DVD Menu Artist
Left Coast Digital
Aptos, CA