*FCP/Motion Banding Solution*

NorBro

Major Contributor
So I discovered something today that I'm sure people out there somewhere may already know, but I think it's worth sharing since others may not...and it's really exciting!

For me - basically for the last 10 years or so - banding in motion graphics has been a major problem in H.264 deliverables.

When you're using lights in Motion on super clean canvases or even a simple vignette in FCP X, your final product will have rings, wavy lines/bands in it unless you're doing something about it (with a compromise) and/or exporting 10/12-bit high-quality formats. But that doesn't always work for everything in video (although it's rarely an issue with non-moving creations).

But generally if you have a short project in Motion, you can export it as ProRes and the problem is mostly solved. Years ago, however, doing this wasn't realistic for quantity and length until computers greatly improved (the solution below was simply not possible for most people's hardware) and video sharing platforms allowed you to upload more data (if your internet was fast enough).

When I make stuff in Motion - which looks gorgeous inside that software - and send 12-bit PR 444XQ to FCP for further editing, I have never been able to produce H.264 without banding/rings in it from the lights used in Motion (it even occurs with soft gradients in 2D projects).

I don't know how other applications handle everything, but during my decade-long search for a simple solution which included adding noise, generators, masks, behaviors and dithering (via Compressor), I found similar questions asked about After Effects. Most of the time only adding a little noise immediately reduces it, but unfortunately you lose your solid backgrounds as you attempt to strike a balance between the present and newly introduced artifacts. And one of the most bizarre occurrences in FCP is if you pause in your timeline, any banding disappears. But the second you move your playhead or play your video, you can see it again (and it's probably related to the better quality/performance setting).

___

So today I'm working on a very simple presentation and this problem is turning into a nightmare. This thing is like an hour long - similar to a PowerPoint but with more savvy editing - and I just want it to look good. And I'm searching solutions and trying a few new different approaches to no avail.

And by the way...I know most people don't use Motion 5 so if you're still reading then you might appreciate this.

So totally randomly I'm back in Compressor looking around (which I barely use anymore), and I end up trying HEVC as the codec for a MOV with its 'Profile' parameter set to 10-bit. (The default is 8-bit.)

And instantly it's so much better with HEVC. (And of course the files are smaller which is part of the goal.)

In the past, besides PR I have tried Apple Intermediate Codec, Photo-JPEG, Animation and Uncompressed 10-Bit 4:2:2 and nothing has worked.

___

I'm most likely late to the party even though it has only been a couple of years since HEVC's introduction into this software, but even now after searching for HEVC/FCP/Motion/banding/etc. together, I'm not seeing anyone recommending it.

Also, I created a custom Compressor setting/droplet and added it as an export destination in FCP so I can skip sending the projects to Compressor.

This is like the biggest breakthrough I've ever had. lol

___

Before and after...these are frames from a video. (The issue - as mentioned - is practically non-existent for static designs.)

bsgTOUR.png


FJLgQJu.png
 
Last edited:
Sometimes it gets to a point where it is just better to hire out and get a pro. I do AE work but there are instances where we need a really polished product and it would take too much time for me with my decent/intermediate experience. Galera https://galera.agency/ nailed it.
 
Last edited:
NorBro;n375300 ...and I end up trying HEVC as the codec for a MOV with its 'Profile' parameter set to 10-bit. (The default is 8-bit.) And instantly it's so much better with HEVC. (And of course the files are smaller which is part of the goal.)[/QUOTE said:
Totally 100% agree. 10-bit HEVC (H.265) is well ahead of H.264. For nearly everything now, I'm delivering H.265 HEVC in an MP4 wrapper. I also supply an H.264 file but try to impress on clients to use the H.265 if they can. Overall, a better image and one that can be rendered to a smaller file. 8-bit concatenation on 10-bit graphics has been a millstone around the production neck for years. Plus, I'm finding HEVC incredibly fast in Resolve with a decent GPU. The highest render rate I've seen with 'Updates During Renders' turned off has been up to 7.1 real-time rendering. Happy with that!

Chris Young
 
This is funny to read back on, I was really excited, lol.

The update at some point then to include HEVC really wiped out 10+ years of frustration trying to make certain titles with mixed gradients/lighting in Motion & FCP.
 
Chris, do you see a visual quality difference when comparing 10-bit 4:2:2 H.265 to 10-bit 4:2:2 H.264 shot with the same camera, on the same shot, etc.?

Just to interject - my understanding was that h265 codecs theoretically achieve equivalent image quality as h264 with about half the space. (Based on some sort of signal/noise analysis and with the caveat that the difference is more pronounced at smaller bitrates.)

I haven't seen a quality difference between these codecs on my cameras but I think we should also be asking about the bitrate. The a7siii, for example, offers a 10-bit 4:2:2 long-gop h264 bitrate of 100Mbps. It offers a 10-bit 4:2:2 long-gop h265 bitrate of either 50Mbps or 100Mbps. The idea is that the 50Mbps should match the quality of the 100Mbps H264. So, the question should be if the 100Mbps H265 exceeds the quality of the 100Mbps H264. But we also know that even in theory, there shouldn't be much difference at all at this (relatively) higher bitrate.

And I admit I'm perplexed by NorBro's findings in post, although I'm excited to try it out.

Here's a chart of some sort of quantitative measure of image quality for h264 vs h265 graphed against bitrate:

The-relationship-between-video-quality-SSIM-and-bitrate-of-both-compression-standards.png
https://www.researchgate.net/figure...of-both-compression-standards_fig15_273187466
 
Just to interject - my understanding was that h265 codecs theoretically achieve equivalent image quality as h264 with about half the space. ]

That is my understanding too. Just a more compressed file. But I don't care about bit rates or file sizes. So that's why I'd like to find out if there is any visual difference whatsoever between the two. I won't even ask Chris to prove it. :) If he says there is a visual difference i will take his word for it.
 
Just to make sure we're all on the same page, I was talking about banding when making graphics in Motion/FCP, and nothing to do with cameras.

A camera's relationship with H.265 can be different than a software's.

The findings are pretty straightforward...it exists with H.264 and doesn't with H.265, anyone can test it themselves if they have the software.

I mean, that's it; it's there with one and it's not there with the other, no other evidence is necessary.

___

Now, critically-thinking, you could argue if the way H.264 is implemented into FCP and Motion by Apple is the problem, but that's not the case study here.

This was to find a solution to make the final master/export look better, how it looks in the software when creating it.
 
Just to make sure we're all on the same page, I was talking about banding when making graphics in Motion/FCP, and nothing to do with cameras.

Yeah, we are drifting off topic but I'd still like to hear what Chris says if he has done comparisons.

BTW, was your H.264 that you didn't like 10-bit 4:2:2? Just curious.
 
Yeah, we are drifting off topic but I'd still like to hear what Chris says if he has done comparisons.

BTW, was your H.264 that you didn't like 10-bit 4:2:2? Just curious.

It was 8-bit 4:2:2 and likely always the problem.

When I wrote the above 2 years ago I wasn't absolutely sure, but looks like it. (Apple never supported 10-bit H.264, at least back then.)

I should open up Compressor and revisit all of this again at some point.
 
It was 8-bit 4:2:2 and likely always the problem.
When I wrote the above 2 years ago I wasn't absolutely sure, but looks like it. (Apple never supported 10-bit H.264, at least back then.)

So is it fair to say the root problem was 10-bit vs. 8-bit, and not H.265 vs. H.264?
Just trying to get a handle on whether or not there is any benefit to H.265 (when all other things are equal) besides just smaller file sizes.
 
I think so, but the way I was conversating about it wasn't meant to imply an apples to apples H.265 vs. H.264 comparison, but a, "This is the H.264 option we had to always use in Compressor vs. check out this new H.265 option in Compressor and wow, what a difference!".

I also don't remember why I said above that some of the other formats didn't work when they are 10-bit; there was likely more to the workflow that I'm just not remembering or something changed.
 
Maybe even more off topic.... but now I'm curious... this thread appeared in my "new posts" section yesterday BEFORE Tamesis added his / her plug, even though it was over 2 years old. ???
 
Maybe he/she or someone wrote something & then quickly deleted it and there was a temporary window where the thread was bumped even without the comment until the forum refreshed itself (or when he/she wrote what's above).

That's the only thing I can think of as it would never be bumped without some type of activity.
 
Maybe even more off topic.... but now I'm curious... this thread appeared in my "new posts" section yesterday BEFORE Tamesis added his / her plug, even though it was over 2 years old. ???

Oh look at that. Makes sense - norbro never sounds that positive or optimistic these days
 
you need some sleep, too many weddings

making no sense with that reply to his inquiry or that wild assumption from idk what

I thought this was a new thread because it showed up in new posts and I didn't look at the timestamp and yours was the only post. And I thought hey, NorBro sounds more chipper than usual. Then it turned out it was 2 years old
 
Back
Top