Looking for solutions to border contamination when rendering shape layers

Forum|Forum|11 months ago
April 26, 2025
3 replies
773 views

As mentioned in the title, I discovered an issue where the borders between the object and the background become contaminated after rendering a video.
To investigate, I conducted several tests:

First image: Apple ProRes 4444 8bit
Second image: QuickTime Animation 32bit
Third image: PNG sequence

For the test, I placed a solid background at 1920x1080 resolution
and positioned a 960x540 red-colored shape layer exactly at the center.
This was done to achieve a perfect 1px:1px match.
(The rendered videos were captured as BMP screenshots and then magnified for observation.)

While working on an outsourced project, this issue was found around a logo,
and the client raised a complaint.
I attempted to resolve the issue, but while the PNG sequence had no problems,
the final delivery must be a video file with audio, as it is an advertisement.
This puts me in a difficult situation.

Correct answer RobShultz

What resolution is your graphic/composition?

Warren Heaton

Community Expert

@Bunny.c

Are you able to share a link to the project file that you're working with?

I am not able to reproduce this. If I create a 960-by-540 Solid with red Fill inside of a 1920-by-1080 Comp with White as the background color and then render it to H264 MP4, QuickTime ProRes 422 HQ, QuickTime ProRes 4444 with alpha, and PNG with alpha, it results with clean edges in each render. The project is set to 16-bpc and I used the default Render Settings template ("Best Settings") and default Output Module templates ("H264 - Match Render Settings - 15 Mbps", "High Quality", "High Quality with Alpha", and "PSD Sequence" modified to be a PNG sequence).

My system info:

Application: After Effects v25.2.2.2

OS: macOS v15.3.1, RAM: 128.00 GB GB, CPUs (logical): 16

- Warren

R

Rick Gerard

Community Expert

If you are looking at rendered video or images from renders, you must ensure that the scale is 100% so that the pixels align with the pixel grid on your screen. It does not matter what the format is. Your video playback device's size (scale) and position on the screen matter. You can also run into issues with compression codecs that compress color using 4:2:2 or 4:2:0 color compression. Any color compression requires that the different colored pixels be on an even-numbered pixel grid, or the colors will be blended. Your red and white examples could be perfect examples of that.

If your client's logo has curves or you cannot perfectly align it on an even number of pixels, you may have to fiddle with the background colors to mitigate the anti-aliasing and color compression problems. If the edge colored problems are visible when the video is at 100% size and (4K on a 4K monitor), and aligning the graphic on an even numbered pixel grid does not solve the color compression problem for the client, a subtle color shift in the background is going to be the only solution that will mitigate the situation and potentially eliminate the problem. You may need to gently educate your client on the realities of color compression because nothing is delivered to the public with 4:4:4 color, and the audience will not scale up their marketing videos to 800% and check the edges of the logo when deciding whether or not to purchase the product.

For others viewing this thread, shape layer edges and colored solid layers not aligned with the pixel grid can cause problems. The greater the color/contrast difference between the pixels, the more obvious the aliasing and color sampling problem becomes.

To fix shape layer problems, you must ensure that rectangular shapes with an even number of pixels have the values for Contents/Shape/Rectangle/Position set to whole numbers, and odd-numbered pixels are set to half-pixel (.5) values. Diagonal or angled lines and curves will be anti-aliased because pixels are arranged on a grid. Horizontal and vertical edges and lines can always perfectly align with the pixel grid. It just takes some work.

When working in Illustrator, the View menu has options to Snap to Pixel and Show Pixel Grid, but these options do not exist in After Effects. If you create artwork for video using Illustrator, you should always work with those options turned on and position your artwork in its Hero position.

B

Bunny.cAuthor

Inspiring

Thank you very much for your kind and detailed response.
However, as I mentioned in my original post, for testing purposes, I set up a 960x540 rectangle within a 1920x1080 composition, as shown in the content I attached this time.

Even within the After Effects preview, the borders appear perfectly sharp, and when rendered as a PNG sequence, the borders remain precise.
However, when rendering to a video format, the borders collapse...

R

Rick Gerard

Community Expert

As I said, it's 4:2:0 or 4:2:2 compression. Try rendering ProRez 444 in a 16-bit comp. You will eliminate color sub-sampling and intra-frame compression.

If you are going for distribution to the general public, you will have 4:2:2 or 4:2:0 compression. The only way to remove the artifacts from the color compression is to fiddle with the position or the colors. You can mitigate a little bit with position, but if the image is moving, you will also have intra-frame compression, which will cause additional pixel-level artifacts.

Here is a comparison between ProRez 444 and ProRez 422 compression and the original red on white composition.

Position is your first tool for removing color artifacts. The second is to modify the color differences between the foreground and background. Compression and delivery codecs are not friendly to pixel peeping. You have to understand the limitations and work around them. Your only option is to make the video look as good as you can when YouTube, social media, or cable companies and broadcasters get through re-compressing your work. They all do it.