Kalleheikki Kannisto
1
COW Coin
Forum Replies Created
-
The Pathfinder palette Unite button (first one) does exactly this. One-click operation in this case (before and after):
Kalleheikki Kannisto
Senior Graphic Designer -
Here’s a painstaking method of making video match to audio with After Effects and Twixtor:
https://library.creativecow.net/articles/freitag_lori/twixtor-lip-synch/video-tutorialIt should be possible to adapt the same idea to Premiere using Time Remapping and optical flow frame blending. Haven’t tried it, but I would expect it to work.
Kalleheikki Kannisto
Senior Graphic Designer -
You are correct: it isn’t possible to fix this in Premiere or with any regular de-interlace method, since the footage has been scaled in addition to having been output with interlacing, resulting in footage with blurred interlacing.
Here is a highly magic and complex fix in After Effects (CC2018). Explanation follows.
The upper and lower fields were track matted through a field matte (horizontal lines). The footage had to be scaled down by 88.888 % to make the footage match field lines (footage had been enlarged by 112,5% when output). Only a single-pixel slice was taken from each field to minimize the effect of the opposite field due to blurring. This resulted in an almost clean but sliced version of each field (single-pixel-high field with 3 empty pixel rows between them). Each was then copied three times and moved one, two and three pixels up respectively to fill in the empty rows. The upper and lower fields were precomposed separately. The upper field was then moved up two pixels to be in the correct location in relation to the lower field. The offset between the upper and lower field timing was removed with a Time remap expression and finally the upper and lower fields were blended together at 50% opacity to keep the details from both passes. A small vertical blur of one pixel followed by unsharp mask sharpening was done to regain the look of the original footage.
12971_weddinginterlacesamplefix.mp4.zip
Kalleheikki Kannisto
Senior Graphic Designer -
Hmm. Never noticed that. 14400 pixels sounds like the maximum extents of the drawing area. No harm in deleting that artboard, if it interferes with something in your workflow (such as exporting with artboards).
Kalleheikki Kannisto
Senior Graphic Designer -
General outline:
0) Create a white background and bring you photo in as the second layer.
1) Curve the face lighter.
2) Curve some of the yellow out. A major factor that makes the retro look is the lack of yellow: yellow color tends to fade out over time in the sun first, followed by magenta.
3) Make a layer mask and paint out everything but the mouth, nose tip and eyes (and, if the final hairdo is in the shot already, also keep the hair).
4) Adjust layer mask and color until you got a decent resemblance of the reference image. Note that the nose tip is faded out more than the rest of the remaining features, so you would either curves it lighter individually or make it more transparent with the layer mask.Kalleheikki Kannisto
Senior Graphic Designer -
I think you’re on the right track with using motion blur. Although it is hard to tell from the tiny reference shot, I suspect that it is a relatively long exposure which essentially gives you “motion blurred” reflections as the waves move some during the exposure. The question is how to get the needed organic variation into it so that some parts are blurred more than others.
You might want to try generating a selection mask with fractal noise that has been blurred and then squished vertically. That way you can use the selection (or layer mask) to vary the amount of motion blur in different parts.
For an even more organic look, or for final touches that break the fully directional motion blur, you could try the smear tool.
A light touch with the wave effect could give you the needed variation without actually smearing the reflection further.
Kalleheikki Kannisto
Senior Graphic Designer -
Audio-to-keyframes approach might be a better option there. You get the result from the right channel, the left channel and the combination separately, which gives you three slightly different outputs and then you can do some math such as right minus left channel to get the fourth output.
Kalleheikki Kannisto
Senior Graphic Designer -
The first place to investigate would be the graph display of the position animation.
Kalleheikki Kannisto
Senior Graphic Designer -
The long and the short of it is, you can’t currently do single-axis scaling with the Aux channels. The best workaround is to use Group nulls, parenting them if you need to emulate child-parent transformations as if the sub-object was a child of the main object. This becomes tricky if you’ve already used up all groups, as you need to use Unified Render. I find Unified render unreliable when trying to control the same group number animation.
Other options are very limited:
1) You may be able to use the Animation Engine, but with complex setups it is even more difficult than using nulls and unified render.
2) You may be able to set up (if the object is of suitable shape) an extrusion model and adjust the original path instead of the scale. You will introduce anchor point and refresh problems this way, but it is possible to overcome both.
Parented nulls are the most viable option.
Kalleheikki Kannisto
Senior Graphic Designer -
[Scott Petterson] “It’s how the masks work in AE: if you keep Illustrator out of the equation, and just go directly to AE and make a mask, the mask will be inverted in any area where the mask overlaps itself. “
That’s correct, which is why you need to to edit them out in Illustrator to have no such overlaps before copy-pasting into AE. Pathfinder operations come in handy there.
Kalleheikki Kannisto
Senior Graphic Designer
