Darby Edelen
Forum Replies Created
-
Hey Dan, I was wondering if you might have any insight into why the wiggle() function seems to be biased toward the lower values in the wiggle range.
I was doing something very similar to what the original poster requested the other day and found that even across a large sampling of the timeline well over 50% of the values were returning below the average.
If I increased the number of octaves then it became even more dramatic.
Does that seem like acceptable behavior or a bug to you? Personally I’d prefer the values to be evenly dispersed around the center but I may not be considering certain details required in implementation.
I wrote a little code to check the average value returned by a wiggle around 0° on rotation:
subf = 1;
fd = thisComp.frameDuration / subf;
f = thisComp.duration / fd;
r = thisComp.layer("Null 1").transform.rotation;
sum = 0;
for(x = 0; x < f; x++){
rf = r.valueAtTime(x * fd);
sum += rf;
}
sum / f;The subf variable (number of times to sample between frames) doesn’t seem to be very necessary so I’ve left it at 1.
I applied this to a text layer and I’ve only been able to find slight positive bias (+0.35) on 2 out of 13 layers when duplicating the layer to generate new seeds. Most of the layers have negative bias around -2. Maybe I’m just unlucky? 🙂
As an additional test I pointed this at a Null with a noise() expression applied:
f = 50;
noise(f * time);In this case the average never goes above ±0.03 when varying the frequency and rarely goes above ±0.01 while sampling once per frame. I started increasing the subframe variable in this case and the averages moved toward 0, even the -0.03 went down to -0.0001 at 20 subframe samples. Not surprisingly it looks like noise() centers around 0. I can’t imagine why wiggle() wouldn’t!
Darby Edelen
-
[Jack Gibbons] “I’m using File>Create ProEXR Layer Comps. I’m under the impression this does the same thing as the Extrator but for all the channels at the same time?”
So you reload the EXR sequence and then use Create ProEXR Layer Comps on the reloaded sequence but don’t get all of the compositions you were expecting? Hrmmm, maybe try using Replace Footage on the sequence instead?
Darby Edelen
-
The codecs I mentioned in my last post are all QuickTime. You can use Adobe Media Encoder or another solution to transcode to them, in order of my preference:
1. PhotoJPEG (100% quality is practically visually lossless, 95% is still quite good and provides a lot of file size savings)
2. ProRes (practically visually lossless, quite good file size, not compatible with every system especially windows)
3. Animation (entirely lossless, files are big)
Keep in mind this should help your experience in AE but it likely won’t solve all your troubles.
Darby Edelen
-
I don’t have any R3D footage handy here at home to give you details but I’d start by trying some different debayer settings. I might also check the footage in other applications to see whether that has any impact (REDCINE-X maybe). If the problem only appears in AE then I’d take a closer look at the RAW options and project color settings.
Sorry I can’t provide more specific help, I’ve never encountered this exact issue.
Darby Edelen
-
Format is a more general term, but yes that’s basically what I was asking.
.mp4 is either an H.264 or MPEG-4 video which may not decode well in AE. I’d recommend transcoding to a format that is more friendly like Animation (big files, maximum quality), ProRes (bigger files than MP4, smaller than animation, high quality) or PhotoJPEG (bigger files than MP4, smaller than animation, fairly high quality).
If the file you’re working with is quite large then I’d recommend transcoding just a small part of it to see whether that improves AE’s performance.
Darby Edelen
-
What codec is the video using? Some codecs (like AVC/H.264) take a while to read/decode.
Darby Edelen
-
Just a quick addendum to my last post. The expression there will re-evaluate the control layer on every frame, so if the black/white sections change over time (fractal noise with evolution animated for example) the Nulls will choose a new random position whenever their position is no longer over black.
Darby Edelen
-
Have you tried removing and re-applying the EXtractoR effect on the footage in your composition after reloading it?
Darby Edelen
-
The preview will look correct no matter which channels you have checked in the Bake Texture tag. I’d consider that a bug but it’s likely because its showing a composite of all the channels (even unchecked ones).
The Surface Color channel does not bake reflections in my experience (and according to the documentation), you need to check the Reflection channel for that.
Darby Edelen
-
I can think of a few options that may work. Some will work better than others depending how your project is set up.
The first thing that comes to mind is the Surface deformer. However, this is not for the feint of heart. It takes some getting used to. In your case it could allow you to deform the geometry onto the UV coordinates of the source geometry so it requires a meaningful UV map. You’d also need to split the source geometry to create a proxy that consists of only the polygons the deformed geometry should follow (which you can hide from the render). Some research into the Surface deformer would be advised 🙂
An alternate would be to select an edge that runs along your source geometry that roughly matches the path you’d like the glass object to deform to. Use the Mesh > Commands > Edge to Spline command to create a spline that follows the edge. Then you can use the Spline Wrap deformer to deform your glass object along that spline.
There are some details that will change depending on the look you want and how your project is set up but those two options should be good starting points.
Also for both of these methods you’ll need to make sure that the glass geometry is sufficiently subdivided so that it can deform around corners and along curves.
Darby Edelen